JP4947019B2 - Air-fuel ratio control device for internal combustion engine - Google Patents

Description

  The present invention relates to an exhaust air passage of an internal combustion engine, and an upstream air-fuel ratio sensor and a downstream that are respectively disposed upstream and downstream of a “catalyst (exhaust purification catalyst, three-way catalyst) provided in the exhaust passage” The present invention relates to an air-fuel ratio control device for an internal combustion engine that controls “the air-fuel ratio of an air-fuel mixture supplied to the engine (hereinafter sometimes simply referred to as“ the air-fuel ratio of the engine ”)” based on the output value of the side air-fuel ratio sensor. .

  Conventionally, an upstream air-fuel ratio sensor, a catalyst, and a downstream air-fuel ratio sensor are provided from upstream to downstream of the exhaust passage of the internal combustion engine, and the output value of the upstream air-fuel ratio sensor and the output value of the downstream air-fuel ratio sensor are An air-fuel ratio control device that controls the air-fuel ratio of an engine based on the above is widely used.

  More specifically, the conventional air-fuel ratio control apparatus calculates a sub-feedback amount for making the output value of the downstream air-fuel ratio sensor coincide with the downstream target value. Further, the conventional air-fuel ratio control device performs “learning” for updating the learning value based on “a value corresponding to the steady component included in the calculated sub feedback amount”, and “updates the sub feedback amount. Is corrected according to the “learned value”. Then, the conventional air-fuel ratio control device is configured so that the main feedback amount for making the air-fuel ratio of the engine coincide with the stoichiometric air-fuel ratio based on the output value of the upstream air-fuel ratio sensor, the sub-feedback amount, and the learning value. And the engine air-fuel ratio (for example, fuel injection amount) is feedback controlled based on the calculated main feedback amount.

  In the present specification, calculating a new main feedback amount (updating) and using the main feedback amount for controlling the air-fuel ratio of the engine is also referred to as “execution of main feedback control”. Similarly, newly calculating (updating) the sub-feedback amount and using the sub-feedback amount for controlling the air-fuel ratio of the engine is also referred to as “sub-feedback control is executed”.

The learned value is stored in a backup RAM (standby RAM) provided in the air-fuel ratio control device. The backup RAM is supplied with power from the battery regardless of the position of the ignition key switch of the vehicle on which the engine is mounted. The backup RAM can hold the “stored value (data)” as long as power is supplied from the battery. Therefore, for example, the convergence value of the sub-feedback amount from the time when the sub-feedback control is started as the downstream air-fuel ratio sensor is activated to the time when the sub-feedback amount reaches the convergence value (steady value). Can be compensated by the learning value. As a result, immediately after the start of the sub-feedback control, the air / fuel ratio of the engine can be controlled to be an air / fuel ratio in the vicinity of an appropriate value. Note that the learning value may be stored in a nonvolatile memory such as an EEPROM (see Patent Document 1).
JP 2007-162626 A

  By the way, an upstream air-fuel ratio sensor such as a limit current type is, for example, an output value indicated by a solid line CT in FIG. 3 with respect to an upstream air-fuel ratio (the air-fuel ratio of exhaust gas reaching the upstream air-fuel ratio sensor) abyfs. (Normal value) Vabyfs ”is output. More specifically, the output value Vabyfs of the upstream air-fuel ratio sensor increases as the upstream air-fuel ratio abyfs becomes larger than the stoichiometric air-fuel ratio (lean air-fuel ratio). Furthermore, when the upstream air-fuel ratio abyfs is the stoichiometric air-fuel ratio, the output value Vabyfs becomes a predetermined value Vstoich. The control device stores the “relationship between the output value Vabyfs and the upstream air-fuel ratio abyfs” indicated by the solid line CT in the form of a table (look-up table) Mapabyfs. For example, when the upstream air-fuel ratio sensor is outputting the output value V1 (V1 <Vstoich), the control device determines that “the upstream air-fuel ratio abyfs is A1” based on the output value V1 and the table Mapabyfs. Recognize.

  However, the output value of the upstream air-fuel ratio sensor may deviate from the normal value due to various causes. For example, when a large thermal stress is generated inside the air-fuel ratio detection element of the upstream air-fuel ratio sensor due to adhesion of moisture in the exhaust gas to the upstream air-fuel ratio sensor, and as a result, a crack occurs in the element ( There is a case where an element crack occurs. In this case, the output value of the upstream air-fuel ratio sensor changes as indicated by the broken line CL in FIG. That is, the output value Vabyfs (broken line CL) of the upstream air-fuel ratio sensor is further away from the value Vstoich than the normal value (solid line CT). Such an abnormality of the output value Vabyfs is also referred to as “sensor output value expansion abnormality”.

  In this case, the output value Vabyfs of the upstream air-fuel ratio sensor is such that the true upstream air-fuel ratio abyfs is “an air-fuel ratio that is closer to the theoretical air-fuel ratio than the upstream air-fuel ratio A1 (a lean air-fuel ratio, a large air-fuel ratio). When it is “A2”, the value is V1. Therefore, the control device recognizes that the current upstream air-fuel ratio abyfs is “richer (smaller) air-fuel ratio A1” than “air-fuel ratio A2 that is the true upstream air-fuel ratio abyfs”.

  As a result, when air-fuel ratio control with the main feedback amount is executed, the air-fuel ratio of the engine is overcorrected to the lean side. On the other hand, as is well known, the air-fuel ratio of the exhaust gas flowing out downstream of the catalyst is a value corresponding to the “average value of the air-fuel ratio of the engine”. Therefore, the above-described “overcorrection of the air-fuel ratio to the lean side” appears as a change in “a sub feedback amount based on the output value of the downstream air-fuel ratio sensor”. That is, the sub feedback amount changes to “a value for correcting the air-fuel ratio of the engine to a richer side”. For this reason, the learning value of the sub feedback amount also changes to “a value that corrects the air-fuel ratio of the engine to a richer side”.

  In addition, when the output value Vabyfs of the upstream air-fuel ratio sensor is a normal value, when the upstream air-fuel ratio sensor is outputting the output value V2 (V2> Vstoich), the control device determines that the output value V2 and the table Mapabyfs Based on this, the upstream air-fuel ratio abyfs is recognized as A3. On the other hand, when the sensor output value expansion abnormality occurs, the output value Vabyfs indicates that the true upstream air-fuel ratio abyfs is “the air-fuel ratio closer to the stoichiometric air-fuel ratio than the upstream air-fuel ratio A3 (rich air-fuel ratio, smaller The value V2 when the air-fuel ratio is A4 ". Therefore, the control device recognizes that the current upstream air-fuel ratio abyfs is “a leaner (larger) air-fuel ratio A3” than “the air-fuel ratio A4 which is the true upstream air-fuel ratio abyfs”.

  As a result, when air-fuel ratio control with the main feedback amount is executed, the air-fuel ratio of the engine is overcorrected to the rich side. Therefore, this “overcorrection of the air-fuel ratio to the rich side” appears as a change in “a sub-feedback amount based on the output value of the downstream air-fuel ratio sensor”. That is, the sub feedback amount changes to “a value that corrects the air-fuel ratio of the engine to a leaner side”. For this reason, the learning value of the sub feedback amount also changes to “a value for correcting the air-fuel ratio of the engine to be leaner”.

  In addition, when such a sensor output value expansion abnormality occurs, the upstream air-fuel ratio acquired from the actual output value Vabyfs of the upstream air-fuel ratio sensor and the table Mapabyfs and the true upstream air-fuel ratio A situation occurs in which the difference is not symmetrical between the lean side and the rich side with respect to the stoichiometric air-fuel ratio (asymmetry of the air-fuel ratio detection error). As a result, the learning value of the sub feedback amount converges to a value different from that when the output value Vabyfs is a normal value. This point will be described more specifically below.

  Assume that the air-fuel ratio A2 is an air-fuel ratio that is smaller than the stoichiometric air-fuel ratio by a positive value dAF, and the air-fuel ratio A4 is an air-fuel ratio that is larger than the stoichiometric air-fuel ratio by a positive value dAF. At this time, the magnitude of the difference between the value A1 and the value A2 is usually larger than the magnitude of the difference between the value A3 and the value A4. As a result, the air-fuel ratio of the engine is overcorrected to the lean side of the stoichiometric air-fuel ratio by the main feedback control. Therefore, the learned value of the sub feedback amount converges to “a value for correcting the engine air-fuel ratio to be richer”.

  Thus, when the sensor output value expansion abnormality occurs, the learning value changes toward a value (convergence value) different from the value immediately before the sensor output value expansion abnormality occurs. However, the learning value is not changed rapidly in order to prevent erroneous learning. Therefore, according to the conventional control device, there is a problem that it takes a relatively long time until the learning value converges immediately after the occurrence of the sensor output value expansion abnormality, and as a result, the emission deteriorates.

  On the other hand, the output value of the upstream air-fuel ratio sensor is indicated by a one-dot chain line CS in FIG. 3 due to the clogging of the diffusion resistance layer covering the air-fuel ratio detection element or the deterioration of the air-fuel ratio detection element itself over time. May vary. That is, the output value Vabyfs (one-dot chain line CS) of the upstream air-fuel ratio sensor is closer to the value Vstoich than the normal value (solid line CT). Such an abnormality of the output value Vabyfs is also referred to as “sensor output value reduction abnormality”.

  As a result, the upstream air-fuel ratio acquired from the output value Vabyfs of the upstream air-fuel ratio sensor and the table Mapabyfs does not represent the true upstream air-fuel ratio, so when the air-fuel ratio control by the main feedback amount is executed The air-fuel ratio of the engine is overcorrected to the rich side or the lean side. This overcorrection of the air-fuel ratio appears in the sub-feedback amount based on the output value of the downstream air-fuel ratio sensor, and accordingly, the learning value of the sub-feedback amount also changes.

  Further, even when such a sensor output value reduction abnormality occurs, the upstream air-fuel ratio acquired by the actual output value Vabyfs of the upstream air-fuel ratio sensor and the table Mapabyfs, the true upstream air-fuel ratio, There arises a situation in which the difference between the two is not symmetrical between the lean side and the rich side with respect to the theoretical air-fuel ratio. That is, in this case as well, the above-described “asymmetricity of the air-fuel ratio detection error” occurs. As a result, the air-fuel ratio of the engine is generally overcorrected to the rich side with respect to the stoichiometric air-fuel ratio by the main feedback control. Therefore, the learned value of the sub feedback amount changes toward the “value for correcting the leaner side of the engine air / fuel ratio”.

  As described above, even when the sensor output value reduction abnormality occurs, the learning value changes toward a value (convergence value) different from the value immediately before the sensor output value reduction abnormality occurs. Therefore, according to the conventional control device, there is a problem that it takes a relatively long time until the learning value converges immediately after the occurrence of the sensor output value reduction abnormality, and as a result, the emission deteriorates.

  The present invention has been made to address the above problems. That is, one of the objects of the present invention is that when the output value of the upstream air-fuel ratio sensor deviates from the normal value, the learning value of the sub feedback amount is quickly brought close to the convergence value, thereby reducing the emission. An object of the present invention is to provide an air-fuel ratio control device that can be avoided.

  Specifically, an air-fuel ratio control apparatus for an internal combustion engine according to the present invention includes a catalyst (three-way catalyst) disposed in an exhaust passage of the internal combustion engine, an upstream air-fuel ratio sensor, a downstream air-fuel ratio sensor, Sub feedback amount calculation means, learning means (sub feedback amount learning means, sub feedback amount learning value update means), air-fuel ratio control means, first abnormality determination means, and learning value setting means.

  The upstream air-fuel ratio sensor is disposed in a portion of the exhaust passage that is upstream of the catalyst. The upstream air-fuel ratio sensor outputs an output value corresponding to the air-fuel ratio of the gas flowing through the part where it is disposed.

  The downstream air-fuel ratio sensor is disposed in a portion of the exhaust passage and downstream of the catalyst. The downstream air-fuel ratio sensor outputs an output value corresponding to the air-fuel ratio of the gas flowing through the part where it is disposed.

  The sub feedback amount calculating means calculates a sub feedback amount for making the output value of the downstream air-fuel ratio sensor coincide with the downstream target value by, for example, proportional integral control or proportional integral derivative control. The downstream target value is generally set to a value corresponding to the theoretical air-fuel ratio. That is, the downstream target value is set to the output value of the downstream air-fuel ratio sensor when the stoichiometric air-fuel ratio gas reaches the downstream air-fuel ratio sensor. As will be described later, the sub-feedback amount is used for feedback control of the air-fuel ratio of the air-fuel mixture supplied to the engine (engine air-fuel ratio).

  The learning means performs learning to update a learning value based on a value corresponding to a stationary component (for example, an integral term in proportional-integral control or an integral term in proportional-integral-derivative control) included in the sub-feedback amount. Yes. The learning means corrects the sub feedback amount according to the “updated learning value”. That is, each time the learning unit updates the learning value, the learning unit corrects the sub feedback amount in accordance with the updated learning value. Thereby, double correction of the sub feedback amount and the learning value is avoided.

  The air-fuel ratio control means calculates the “air-fuel ratio of the air-fuel mixture supplied to the engine” based on the output value of the upstream air-fuel ratio sensor, the corrected sub feedback amount, and the updated learned value. “Air-fuel ratio feedback control” for matching the stoichiometric air-fuel ratio is executed.

  The first abnormality determination means determines whether or not a “first abnormality state” has occurred. Examples of the first abnormal state include the sensor output value expansion abnormality and the sensor output value reduction abnormality described above. That is, in the first abnormal state, “the actual output value that is the actual output value of the upstream air-fuel ratio sensor” is “the output value of the upstream air-fuel ratio sensor when the upstream air-fuel ratio sensor is normal, that is, normal This is a state different from “value”.

  The learning value setting means uses the output value of the upstream air-fuel ratio sensor at the time of “predetermined gas arrival”, which is “when the predetermined gas has arrived”, and the convergence value of the learning value. The relationship with a certain expected convergence value is stored in advance. The output value of the upstream air-fuel ratio sensor when the predetermined gas arrives represents the degree of the first abnormal state (sensor output value expansion abnormality or sensor output value reduction abnormality) occurring in the upstream air-fuel ratio sensor. Therefore, the relationship between the output value of the upstream air-fuel ratio sensor when the predetermined gas reaches and the convergence value of the learning value when such a first abnormal state occurs is uniquely determined, and is obtained in advance through experiments. Can be kept.

  The learning value setting means further determines that the actual convergence value of the learning value is based on the “the actual output value when the predetermined gas arrives” and the “relation” when it is determined that the first abnormal state has occurred. And the learning value is set (replaced) with the actual predicted convergence value acquired.

  According to this, “the actual output value that is the actual output value of the upstream air-fuel ratio sensor is different from the output value (normal value) of the upstream air-fuel ratio sensor when the upstream air-fuel ratio sensor is normal. When one abnormal state occurs, the learning value is immediately changed to an expected convergence value (convergence expected value), so that the learning value becomes a value close to an appropriate value immediately after the first abnormal state occurs. As a result, the air-fuel ratio control apparatus of the present invention can shorten the time from the occurrence of the first abnormal state to the time when the learning value converges. As a result, “the learning value deviates from the appropriate value. Can be avoided.

In this case, the learning value setting means
When it is determined that the second abnormal state has occurred, even if it is determined that the first abnormal state has occurred, the “acquired actual convergence expected value of the learning value (first It is preferable that the “setting of the learning value to the predicted convergence value when an abnormal state occurs” is not executed. This second abnormal state is an “abnormal state other than the first abnormal state” and “an abnormal state in which the convergence value of the learning values differs before and after the occurrence of the second abnormal state”.

For example, an abnormal state in which a fuel injection valve provided for a specific cylinder among a plurality of cylinders injects a larger amount of fuel than “indicated fuel amount” (fuel injection valve rich deviation abnormality) When this occurs, HC, CO, and H ₂ discharged from the specific cylinder increase rapidly. As a result, the upstream air-fuel ratio sensor is affected by H _{2 having} a small molecular diameter, and generates an output value corresponding to the air-fuel ratio richer than the actual average air-fuel ratio. As a result, the center of the engine air-fuel ratio shifts to the lean side from the stoichiometric air-fuel ratio by main feedback control based on the output value of the upstream air-fuel ratio sensor. Therefore, the sub-feedback amount changes to “a value that shifts the engine air-fuel ratio to the richer side” compared to immediately before the occurrence of the rich deviation abnormality of the fuel injection valve. Therefore, the learning value of the sub feedback amount also converges to “a value that shifts the air-fuel ratio of the engine to a richer side”.

  Further, for example, when misfire occurs (when the air-fuel mixture does not burn in the combustion chamber), the learned value of the sub-feedback amount is “lean engine air-fuel ratio more lean than the learned value immediately before misfire occurs. May converge to "value to shift to".

  These “abnormality deviation of fuel injection valve (air-fuel ratio imbalance between cylinders)” and “misfire” are representative examples of the second abnormal state described above. When the first abnormal state occurs when such a second abnormal state occurs, the learned value of the sub feedback amount is calculated by the learning value setting means as “when the first abnormality occurs (first Convergence to a value different from the “convergence expected value when an abnormal condition occurs alone”. Therefore, in such a case, if the learning value is set to “the predicted convergence value expected when the first abnormality occurs alone”, the learning value may be moved away from the actual convergence value. is there.

  On the other hand, according to the above configuration, the learning value is not set to the acquired actual convergence expected value when it is determined that the second abnormal state has occurred. The possibility that the value deviates from the actual convergence value can be reduced. As a result, it is possible to avoid the emission from getting worse.

On the other hand, the learning value setting means includes:
The relationship (that is, the relationship between the output value of the upstream air-fuel ratio sensor when the predetermined gas arrives and the predicted convergence value of the learning value) is expressed as “the output value of the upstream air-fuel ratio sensor when the predetermined gas arrives”. And “the difference between the normal value when the predetermined gas is reached” and “the normal value that is the convergence value of the learning value that is expected when the convergence expected value and the upstream air-fuel ratio sensor are normal”. It can also be stored as a “learning value difference relationship” indicating a “relation” between the “difference from the predicted convergence value”.

And the learning value setting means includes
When it is determined that the first abnormal state has occurred, based on the difference between the actual output value when the predetermined gas arrives and the normal value when the predetermined gas arrives, and the learning value difference relationship, It is preferable that the amount of change of the learning value is predicted, and “a value obtained by changing the learning value at the current time by the amount of the predicted change” is acquired as “the actual predicted convergence value”.

  According to this, the first abnormality is based on the difference between the convergence value of the learning value when the upstream air-fuel ratio sensor is normal and the convergence value of the learning value when the first abnormal state occurs alone. The learning value immediately before the occurrence of the state is corrected. Therefore, even when the first abnormal state occurs when the second abnormal state as described above occurs, only the amount of change in the learned value resulting from the occurrence of the first abnormal state is reflected in the learned value. Therefore, the learning value becomes closer to the appropriate value (convergence value). As a result, the emission can be made better.

<First Embodiment>
Hereinafter, an air-fuel ratio control apparatus for an internal combustion engine according to a first embodiment of the present invention (hereinafter simply referred to as a “first control apparatus”) will be described with reference to the drawings. The first control device is also a fuel injection amount control device that controls the amount of fuel injected and supplied to the engine (fuel injection amount).

(Constitution)
FIG. 1 shows a schematic configuration of an internal combustion engine 10 to which the first control device is applied. The engine 10 is a four-cycle / spark ignition type / multi-cylinder (four cylinders in this example) / gasoline fuel engine. The engine 10 includes a main body 20, an intake system 30, and an exhaust system 40.

  The main body portion 20 includes a cylinder block portion and a cylinder head portion. The main body portion 20 includes a plurality (four) of combustion chambers (first cylinder # 1 to fourth cylinder # 4) 21 including a piston top surface, a cylinder wall surface, and a lower surface of the cylinder head portion.

  In the cylinder head portion, an intake port 22 for supplying a mixture of air and fuel to each combustion chamber (each cylinder) 21 and an exhaust port for discharging exhaust gas (burned gas) from each combustion chamber 21 23 is formed. The intake port 22 is opened and closed by an unillustrated intake valve, and the exhaust port 23 is opened and closed by an unillustrated exhaust valve.

  A plurality (four) of spark plugs 24 are fixed to the cylinder head portion. Each spark plug 24 is disposed such that its spark generating part is exposed at the center of each combustion chamber 21 and in the vicinity of the lower surface of the cylinder head part. Each spark plug 24 generates an ignition spark from the spark generating portion in response to the ignition signal. Further, a plurality (four) of fuel injection valves (injectors) 25 are fixed to the cylinder head portion. One fuel injection valve 25 is provided for each intake port 22. The fuel injection valve 25 responds to the injection instruction signal and injects the fuel of the instruction injection amount included in the injection instruction signal into the corresponding intake port 22.

  Further, an intake valve control device 26 is provided in the cylinder head portion. The intake valve control device 26 has a known configuration that adjusts and controls the relative rotation angle (phase angle) between an intake camshaft (not shown) and an intake cam (not shown) by hydraulic pressure. The intake valve control device 26 operates based on an instruction signal (drive signal) and can change the valve opening timing (intake valve opening timing) of the intake valve.

  The intake system 30 includes an intake manifold 31, an intake pipe 32, an air filter 33, a throttle valve 34, and a throttle valve actuator 34a.

  The intake manifold 31 includes a plurality of branch portions connected to each intake port 22 and a surge tank portion in which the branch portions are gathered. The intake pipe 32 is connected to the surge tank portion. The intake manifold 31, the intake pipe 32, and the plurality of intake ports 22 constitute an intake passage. The air filter 33 is provided at the end of the intake pipe 32. The throttle valve 34 is rotatably attached to the intake pipe 32 at a position between the air filter 33 and the intake manifold 31. The throttle valve 34 changes the opening cross-sectional area of the intake passage formed by the intake pipe 32 by rotating. The throttle valve actuator 34a is formed of a DC motor, and rotates the throttle valve 34 in response to an instruction signal (drive signal).

  The exhaust system 40 includes an exhaust manifold 41, an exhaust pipe (exhaust pipe) 42, an upstream side catalyst 43, and a downstream side catalyst 44.

  The exhaust manifold 41 includes a plurality of branch portions 41a connected to each exhaust port 23 and a collective portion 41b in which the branch portions 41a are gathered. The exhaust pipe 42 is connected to a collective portion 41 b of the exhaust manifold 41. The exhaust manifold 41, the exhaust pipe 42, and the plurality of exhaust ports 23 constitute a passage through which exhaust gas passes. In the present specification, the collecting portion 41b of the exhaust manifold 41 and the exhaust pipe 42 are referred to as “exhaust passage” for convenience.

The upstream catalyst 43 is a three-way catalyst that supports “noble metal as a catalyst material” and “ceria (CeO 2)” on a support made of ceramic and has an oxygen storage / release function (oxygen storage function). The upstream catalyst 43 is disposed (intervened) in the exhaust pipe 42. When the upstream catalyst 43 reaches a predetermined activation temperature, it exhibits a “catalytic function for simultaneously purifying unburnt substances (HC, CO, H _2, etc.) and nitrogen oxides (NOx)” and “oxygen storage function”. .

  The downstream catalyst 44 is a three-way catalyst similar to the upstream catalyst 43. The downstream catalyst 44 is disposed (intervened) in the exhaust pipe 42 downstream of the upstream catalyst 43.

  Further, the first control device includes a hot-wire air flow meter 51, a throttle position sensor 52, an engine speed sensor 53, a water temperature sensor 54, an upstream air-fuel ratio sensor 55, a downstream air-fuel ratio sensor 56, and an accelerator opening sensor 57. I have.

The hot-wire air flow meter 51 detects the mass flow rate of the intake air flowing through the intake pipe 32 and outputs a signal representing the mass flow rate (intake air amount per unit time of the engine 10) Ga.
The throttle position sensor 52 detects the opening degree of the throttle valve 34 and outputs a signal representing the throttle valve opening degree TA.

  The engine rotational speed sensor 53 outputs a signal having a narrow pulse every time the intake camshaft rotates 5 ° and a wide pulse every time the intake camshaft rotates 360 °. A signal output from the engine rotation speed sensor 53 is converted into a signal representing the engine rotation speed NE by the electric control device 60. Further, the electric control device 60 acquires the crank angle (absolute crank angle) of the engine 10 based on signals from the engine rotation speed sensor 53 and a crank angle sensor (not shown).

  The water temperature sensor 54 detects the temperature of the cooling water of the internal combustion engine 10 and outputs a signal representing the cooling water temperature THW.

  The upstream air-fuel ratio sensor 55 is disposed in either the exhaust manifold 41 or the exhaust pipe 42 (that is, the exhaust passage) at a position between the collecting portion 41 b of the exhaust manifold 41 and the upstream catalyst 43. The upstream air-fuel ratio sensor 55 is disclosed in, for example, “limit current type wide area air-fuel ratio including a diffusion resistance layer” disclosed in JP-A-11-72473, JP-A-2000-65782, JP-A-2004-69547, and the like. Sensor ".

  As shown in FIG. 2, the upstream air-fuel ratio sensor 55 detects that the upstream air-fuel ratio becomes higher when the voltage V applied between the exhaust-side electrode and the atmosphere-side electrode of the upstream air-fuel ratio sensor 55 exceeds a certain voltage Vp. An output value Vabyfs corresponding to abyfs (a value corresponding to the limit current value Ip) is output. In other words, as shown by the solid line CT in FIG. 3, the upstream air-fuel ratio sensor 55 is “the exhaust gas air-fuel ratio (that is, the air-fuel ratio of the exhaust gas flowing through the portion where the upstream air-fuel ratio sensor 55 is disposed (that is, An “output value Vabyfs” corresponding to the “upstream air-fuel ratio abyfs)” is generated. The upstream air-fuel ratio abyfs is the air-fuel ratio of the gas flowing into the upstream catalyst 43, and can be said to be the air-fuel ratio of the air-fuel mixture supplied to the engine 10.

  The output value Vabyfs of the upstream air-fuel ratio sensor 55 increases as the air-fuel ratio of the upstream air-fuel ratio abyfs increases (becomes the lean air-fuel ratio). The output value Vabyfs matches the value Vstoich when the upstream air-fuel ratio abyfs is the stoichiometric air-fuel ratio.

  The electric control device 60 described later stores the data indicated by the solid line CT in FIG. 3 in the form of “look-up table (map) Mapabyfs (Vabyfs)”, and applies the actual output value Vabyfs to the table Mapabyfs. Accordingly, the “actual upstream air-fuel ratio abyfs” is acquired.

  Referring again to FIG. 1, the downstream air-fuel ratio sensor 56 is disposed in the exhaust pipe 42 (that is, the exhaust passage) at a position between the upstream catalyst 43 and the downstream catalyst 44. The downstream air-fuel ratio sensor 56 is a well-known concentration cell type oxygen concentration sensor (O2 sensor). The downstream air-fuel ratio sensor 56 outputs an output value Voxs corresponding to “the air-fuel ratio of the exhaust gas flowing through the portion of the exhaust passage where the downstream air-fuel ratio sensor 56 is disposed (that is, the downstream air-fuel ratio afdown)”. It is designed to output. The downstream air-fuel ratio afdown is the air-fuel ratio of the gas flowing out from the upstream catalyst 43. As is well known, the air-fuel ratio of the gas flowing out from the upstream side catalyst 43 accurately represents the average value of the air-fuel ratio of the air-fuel mixture supplied to the engine 10.

  As shown in FIG. 4, the output value Voxs of the downstream air-fuel ratio sensor 56 becomes the maximum output value max (for example, about 0.9 V) when the downstream air-fuel ratio afdown is richer than the stoichiometric air-fuel ratio. When the air-fuel ratio afdown is leaner than the stoichiometric air-fuel ratio, the minimum output value min (for example, about 0.1 V) is obtained. Furthermore, the output value Voxs becomes “a voltage Vst approximately between the maximum output value max and the minimum output value min (intermediate voltage Vst, for example, about 0.5 V)” when the downstream air-fuel ratio afdown is the stoichiometric air-fuel ratio. . The output value Voxs suddenly changes from the maximum output value max to the minimum output value min when the downstream air-fuel ratio afdown changes from an air-fuel ratio richer than the stoichiometric air-fuel ratio to a lean air-fuel ratio. When the air-fuel ratio afdown changes from an air-fuel ratio leaner than the stoichiometric air-fuel ratio to a rich air-fuel ratio, it suddenly changes from the minimum output value min to the maximum output value max.

  Referring to FIG. 1 again, the accelerator opening sensor 57 detects the operation amount of the accelerator pedal AP operated by the driver, and outputs a signal indicating the operation amount Accp of the accelerator pedal AP.

  The electric control device 60 includes a “CPU, ROM, RAM, a backup RAM that stores data while the power is on and holds the stored data even when the power is shut off, an interface including an AD converter, etc. Is a “well-known microcomputer”.

  The interface of the electric control device 60 is connected to the sensors 51 to 57 so as to supply signals from the sensors 51 to 57 to the CPU. Further, the interface sends an instruction signal (drive signal) or the like to the ignition plug 24 of each cylinder, the fuel injection valve 25 of each cylinder, the intake valve control device 26, the throttle valve actuator 34a, etc. in accordance with an instruction from the CPU. It is like that. The electric control device 60 sends an instruction signal to the throttle valve actuator 34a so that the throttle valve opening TA increases as the acquired accelerator pedal operation amount Accp increases.

(Outline of operation of first control device)
Next, an outline of the operation of the first control device will be described. As will be described in detail later, the first control device proportionally integrates the sub feedback amount Vafsfb for making the output value Voxs of the downstream air-fuel ratio sensor 56 coincide with the “downstream target value Voxsref corresponding to the theoretical air-fuel ratio”. Calculation is performed according to control (PI control) or proportional-integral-derivative control (PID control). Further, the first control device is also referred to as a sub feedback amount learning value Vafsfbg (hereinafter referred to as “sub FB learning value Vafsfbg”) based on “a value corresponding to a steady component (integral term) included in the sub feedback amount Vafsfb”. .) Is updated at a predetermined timing. At this time, if the sub FB learning value Vafsfbg increases by the change amount ΔG, the first control device decreases the sub feedback amount Vafsfb by the change amount ΔG.

  In addition, the first control device calculates the control air-fuel ratio abyfsc based on the output value Vabyfs of the upstream air-fuel ratio sensor 55, the sub feedback amount Vafsfb, the learned value Vafsfbg, and the table Mapabyfs. Then, the first control device calculates a main feedback amount for making the control air-fuel ratio abyfsc coincide with the “upstream air-fuel ratio abyfs which is the theoretical air-fuel ratio”, and based on the calculated main feedback amount, The fuel ratio (fuel injection amount) is controlled.

  Now, it is assumed that the “sensor output value expansion abnormality” described above has occurred. At this time, if the true upstream air-fuel ratio abyfs is an air-fuel ratio leaner than the stoichiometric air-fuel ratio, the output value Vabyfs of the upstream air-fuel ratio sensor is indicated by a solid line CT as shown by the broken line CL in FIG. It becomes larger than “output value (normal value) Vabyfs when upstream air-fuel ratio sensor 55 is normal”. Further, if the true upstream air-fuel ratio abyfs is richer than the stoichiometric air-fuel ratio, the output value Vabyfs of the upstream air-fuel ratio sensor is shown by the solid line CT as shown by the broken line CL in FIG. It becomes smaller than the normal value Vabyfs.

  Further, as described above, when the “sensor output value expansion abnormality of the upstream air-fuel ratio sensor 55” occurs, the upstream air-fuel ratio acquired by the actual output value Vabyfs of the upstream air-fuel ratio sensor and the table Mapabyfs Thus, a situation occurs in which the difference from the true upstream air-fuel ratio is not symmetrical between the lean side and the rich side (the asymmetry of the air-fuel ratio detection error) with respect to the theoretical air-fuel ratio.

  That is, the air-fuel ratio A2 shown in FIG. 3 is an air-fuel ratio that is smaller than the stoichiometric air-fuel ratio by a positive value dAF, and the air-fuel ratio A4 shown in FIG. Assume a large air / fuel ratio. When the output value expansion abnormality occurs in the upstream air-fuel ratio sensor 55, the output value Vabyfs becomes V1 when the true air-fuel ratio is A2. Therefore, the detected upstream air-fuel ratio is A1. Further, when the output value expansion abnormality has occurred in the upstream air-fuel ratio sensor 55, the output value Vabyfs becomes V2 when the true air-fuel ratio is A4. Therefore, the detected upstream air-fuel ratio is A3.

  At this time, as apparent from FIG. 3, the magnitude of the difference between the detected upstream air-fuel ratio A1 and the true air-fuel ratio A2 (| A2-A1 |) is determined from the detected upstream air-fuel ratio A3 and the true air-fuel ratio A4. And the magnitude of the difference (| A3-A4 |). As a result, the air-fuel ratio of the engine is overcorrected to the lean side of the stoichiometric air-fuel ratio by the main feedback control. Therefore, the sub feedback amount Vafsfb changes to “a value for correcting the air-fuel ratio of the engine to be richer”, and as a result, the sub FB learning value Vafsfbg is also “to correct the air-fuel ratio of the engine to be richer. To the value of (converges).

  FIG. 5 shows that “sub-FB learning” is performed in a situation where fuel cut control is executed in the period of time t1 to t2, and “the sensor output value expansion abnormality of the upstream air-fuel ratio sensor 55” occurs during the fuel cut control. 6 is a time chart showing a “change in value Vafsfbg”. In the situation shown in FIG. 5, the convergence value of the sub FB learning value Vafsfbg changes from the value Vafsfbg0 to the value VafsfbgL before and after the fuel cut control. In general, the sub feedback control is stopped during the fuel cut control.

  In this case, according to the conventional control device, the sub feedback control is resumed from the end of the fuel cut (time t2), and the sub feedback amount Vafsfb gradually changes to a value that changes the air-fuel ratio of the engine to the rich side. As a result, the sub FB learning value Vafsfbg gradually changes from “value immediately before fuel cut Vafsfbg0” to “convergence value VafsfbgL”, and reaches the convergence value VafsfbgL at time t3 after time T has elapsed from time t2. To do. In other words, the sub FB learning value Vafsfbg is greatly different from the appropriate value (convergence value) from time t2 to time t3. Further, since feedback control involves an unavoidable control delay, the sub feedback amount Vafsfb does not immediately compensate for the “deviation from the appropriate value of the sub FB learning value Vafsfbg”. As a result, the air-fuel ratio of the engine deviates from the stoichiometric air-fuel ratio, and emission may deteriorate.

  On the other hand, the first control apparatus determines whether or not “a sensor output value expansion abnormality of the upstream air-fuel ratio sensor 55” has occurred during a period from time t1 to time t2 (during fuel cut control). When the first control apparatus determines that the “sensor output value expansion abnormality of the upstream air-fuel ratio sensor 55” has occurred, the first FB learning value Vafsfbg is between the time t1 and the time t2 (at the time t2 at the latest). The value is immediately set to “the convergence value VafsfbgL expected when the sensor output value expansion abnormality of the upstream air-fuel ratio sensor 55 occurs”. As a result, it can be excluded that the emission is greatly deteriorated after the sensor output value expansion abnormality occurs.

  In actuality, the first control device does not perform any of both “the sensor output value expansion abnormality of the upstream air-fuel ratio sensor 55” and “the sensor output value reduction abnormality of the upstream air-fuel ratio sensor 55” during the fuel cut control. It is determined whether or not an error has occurred. In other words, during the fuel cut control, the first control device sets the “actual output value Vabyfs of the upstream air-fuel ratio sensor 55” to “the upstream side when the upstream air-fuel ratio sensor 55 is normal”. First abnormality determining means for determining “whether or not a first abnormal state has occurred” that is different from the “normal value that is the output value Vabyfs of the air-fuel ratio sensor” is provided. When the first control device determines that the first abnormal state has occurred, the sub-FB learning value Vafsfbg is determined as “the convergence value (convergence expected value) VafsfbgL of the sub FB learning value Vafsfbg expected when the first abnormal state occurs”. Immediately. The above is the outline of the operation of the first control device.

(Actual operation)
Next, the actual operation of the first control device will be described.

<Fuel injection amount control>
The CPU of the first control device performs the “routine for calculating the fuel injection amount Fi and instructing fuel injection” shown in FIG. 6 according to a predetermined crank angle (for example, the crank angle of the predetermined cylinder before the intake top dead center). Every time BTDC is 90 ° CA), the cylinder is repeatedly executed for the cylinder (hereinafter also referred to as “fuel injection cylinder”). Therefore, when the predetermined timing comes, the CPU starts processing from step 600, and in step 610, whether the value of the fuel cut flag XCF (hereinafter referred to as “F / C flag XFC”) is “0”. Determine whether or not.

  The value of the F / C flag XFC is set to “1” until a fuel cut return condition is satisfied after a fuel cut start condition described later is satisfied, and is set to “0” in other cases. That is, the value of the F / C flag XFC is set to “1” when the fuel cut control is to be executed. The value of the F / C flag XFC is set to “0” by the initial routine executed when the “ignition key switch (not shown) on which the engine 10 is mounted” is switched from the off position to the on position. It is set up.

  Now, it is assumed that the time is before time t1 shown in FIG. 5 and the value of the F / C flag XFC is “0”. In this case, the CPU sequentially performs the processing from step 620 to step 650 described below, proceeds to step 695, and once ends this routine.

  Step 620: The CPU executes “fuel injection based on“ the intake air amount Ga measured by the air flow meter 51, the engine rotational speed NE obtained based on the signal of the engine rotational speed sensor 53, and the lookup table MapMc ”. The “in-cylinder intake air amount Mc (k)”, which is “the amount of air sucked into the cylinder”, is acquired. The in-cylinder intake air amount Mc (k) is stored in the RAM while corresponding to each intake stroke. The in-cylinder intake air amount Mc (k) may be calculated by a well-known air model (a model constructed according to a physical law simulating the behavior of air in the intake passage).

  Step 630: The CPU obtains the basic fuel injection amount Fbase by dividing the in-cylinder intake air amount Mc (k) by the upstream target air-fuel ratio abyfr. The upstream target air-fuel ratio abyfr is set to the stoichiometric air-fuel ratio stoich.

Step 640: The CPU corrects the basic fuel injection amount Fbase with the main feedback amount DFi. More specifically, the CPU calculates the final fuel injection amount Fi by adding the main feedback amount DFi to the basic fuel injection amount Fbase. The main feedback amount DFi will be described later.
Step 650: The CPU injects the fuel of the final fuel injection amount Fi from the fuel injection valve 25 provided corresponding to the fuel injection cylinder.

<Calculation of main feedback amount>
The CPU repeatedly executes the “main feedback amount calculation routine” shown in the flowchart of FIG. 7 every elapse of a predetermined time. Therefore, when the predetermined timing comes, the CPU starts the process from step 700 and proceeds to step 705 to determine whether or not the “main feedback control condition (upstream air-fuel ratio feedback control condition)” is satisfied.

The main feedback control condition is satisfied when all of the following conditions are satisfied.
(A1) The upstream air-fuel ratio sensor 55 is activated.
(A2) The engine load (load factor) KL is less than or equal to the threshold KLth.
(A3) Fuel cut control is not being performed (the value of the F / C flag XFC is not “1”).

Here, the load factor KL is obtained by the following equation (1). Instead of the load factor KL, an accelerator pedal operation amount Accp may be used. In the equation (1), Mc is the in-cylinder intake air amount, ρ is the air density (unit is (g / l)), L is the exhaust amount of the engine 10 (unit is (l)), and “4” is the engine. The number of cylinders is 10.
KL = (Mc / (ρ · L / 4)) · 100% (1)

  The description will be continued assuming that the main feedback control condition is satisfied. In this case, the CPU makes a “Yes” determination at step 705 to sequentially perform the processing from step 710 to step 740 described below, and proceeds to step 795 to end the present routine tentatively.

Step 710: The CPU acquires the feedback control output value Vabyfc according to the following equation (2). In equation (2), Vabyfs is the output value of the upstream air-fuel ratio sensor 55, Vafsfb is the sub-feedback amount calculated based on the output value Voxs of the downstream air-fuel ratio sensor 56, and Vafsfbg is the learning value of the sub-feedback amount (sub FB learning value Vafsfbg). These values are all values obtained at the present time. A method of calculating the sub feedback amount Vafsfb and the sub FB learning value Vafsfbg will be described later.
Vabyfc = Vabyfs + (Vafsfb + Vafsfbg) (2)

Step 715: The CPU obtains the feedback control air-fuel ratio abyfsc by applying the feedback control output value Vabyfc to the table Mapabyfs indicated by the solid line CT in FIG. 3 as shown in the following equation (3).
abyfsc = Mapabyfs (Vabyfc) (3)

Step 720: In accordance with the following equation (4), the CPU “in-cylinder fuel supply amount Fc (k−N)” which is “the amount of fuel actually supplied to the combustion chamber 21 at a time point N cycles before the current time point”. " That is, the CPU divides “the in-cylinder intake air amount Mc (k−N) at a point N cycles before the current point (ie, N · 720 ° crank angle)” by “the feedback control air-fuel ratio abyfsc”. Thus, the in-cylinder fuel supply amount Fc (k−N) is obtained.
Fc (k−N) = Mc (k−N) / abyfsc (4)

  Thus, in order to obtain the in-cylinder fuel supply amount Fc (k−N), the in-cylinder intake air amount Mc (k−N) N strokes before the current stroke is divided by the feedback control air-fuel ratio abyfsc. This is because “a time corresponding to the N stroke” is required until “the exhaust gas generated by the combustion of the air-fuel mixture in the combustion chamber 21” reaches the upstream air-fuel ratio sensor 55.

Step 725: In accordance with the following equation (5), the CPU “target in-cylinder fuel supply amount Fcr (k) which is“ the amount of fuel that should have been supplied to the combustion chamber 21 at the time N cycles before the current time ”. -N) ". That is, the CPU divides the in-cylinder intake air amount Mc (k−N) N strokes before the current time by the upstream target air-fuel ratio abyfr (= stoich), thereby obtaining the target in-cylinder fuel supply amount Fcr (k−N )
Fcr = Mc (k−N) / abyfr (5)

Step 730: The CPU acquires the in-cylinder fuel supply amount deviation DFc according to the above equation (6). That is, the CPU obtains the in-cylinder fuel supply amount deviation DFc by subtracting the in-cylinder fuel supply amount Fc (k−N) from the target in-cylinder fuel supply amount Fcr (k−N). This in-cylinder fuel supply amount deviation DFc is an amount representing the excess or deficiency of the fuel supplied into the cylinder at the time point before the N stroke.
DFc = Fcr (k−N) −Fc (k−N) (6)

Step 735: The CPU obtains the main feedback amount DFi according to the above equation (7). In this equation (7), Gp is a preset proportional gain, and Gi is a preset integral gain. Further, the “value SDFc” in the equation (7) is “an integral value of the in-cylinder fuel supply amount deviation DFc”. That is, the CPU calculates the “main feedback amount DFi” by proportional-integral control for making the feedback control air-fuel ratio abyfsc coincide with the upstream target air-fuel ratio abyfr.
DFi = Gp · DFc + Gi · SDFc (7)

  Step 740: The CPU adds the in-cylinder fuel supply amount deviation DFc obtained in the above step 730 to the integral value SDFc of the in-cylinder fuel supply amount deviation DFc at that time, so that a new in-cylinder fuel supply amount deviation DFc is obtained. An integral value SDFc is obtained.

  As described above, the main feedback amount DFi is obtained by proportional integral control, and this main feedback amount DFi is reflected in the final fuel injection amount Fi by the processing of step 640 of FIG. 6 described above.

  By the way, “sub-feedback amount Vafsfb and sub-FB learning value Vafsfbg” on the right side of the above equation (2) is set to “value corresponding to the theoretical air-fuel ratio” as “the output value Voxs of the downstream air-fuel ratio sensor 56”. It can be considered as an “auxiliary correction amount” for matching with the downstream target value Voxsref. Therefore, since the feedback control air-fuel ratio abyfsc is a value substantially based on the output value Vabyfs of the upstream air-fuel ratio sensor 55, the main feedback amount DFi is “the engine represented by the output value Vabyfs of the upstream air-fuel ratio sensor 55”. It can be said that this is a correction amount for making “the air-fuel ratio of the air-fuel ratio” coincide with “the upstream target air-fuel ratio abyfr (theoretical air-fuel ratio)”.

  On the other hand, if the main feedback control condition is not satisfied at the time of determination in step 705 of FIG. 7, the CPU determines “No” in step 705 and proceeds to step 745 to set the value of the main feedback amount DFi to “0”. To "". Next, in step 750, the CPU stores “0” in the integral value SDFc of the in-cylinder fuel supply amount deviation. Thereafter, the CPU proceeds to step 795 to end the present routine tentatively. Thus, when the main feedback control condition is not satisfied, the main feedback amount DFi is set to “0”. Accordingly, the basic fuel injection amount Fbase is not corrected by the main feedback amount DFi.

<Calculation of sub feedback amount and sub FB learning value>
The CPU executes the routine shown in FIG. 8 every elapse of a predetermined time in order to calculate the “sub feedback amount Vafsfb” and the “sub FB learning value Vafsfbg”. Therefore, when the predetermined timing comes, the CPU starts the process from step 800 and proceeds to step 805 to determine whether or not the “sub feedback control condition (downstream air-fuel ratio feedback control condition)” is satisfied.

The sub-feedback control condition is satisfied when all of the following conditions are satisfied.
(B1) The main feedback control condition is satisfied.
(B2) The downstream air-fuel ratio sensor 56 is activated.
(B3) The upstream target air-fuel ratio abyfr is set to the stoichiometric air-fuel ratio stoich.

  The description will be continued assuming that the sub-feedback control condition is satisfied. In this case, the CPU makes a “Yes” determination at step 805 to sequentially perform the processing from step 810 to step 820 described below to calculate the sub feedback amount Vafsfb.

Step 810: The CPU obtains an “output deviation amount DVoxs” which is a difference between the “downstream target value Voxsref” and the “output value Voxs of the downstream air-fuel ratio sensor 56” according to the following equation (8). The downstream target value Voxsref is set to a value Vst (0.5 V) corresponding to the theoretical air-fuel ratio. That is, the CPU obtains “output deviation amount DVoxs” by subtracting “current output value Voxs of downstream air-fuel ratio sensor 56” from “downstream target value Voxsref”. The downstream target value Voxsref may be set to a value slightly corresponding to the rich air-fuel ratio than the value Vst. However, since this value is also a value corresponding to the air-fuel ratio within the range of “the so-called window of the upstream catalyst 43”, it is a value substantially corresponding to the stoichiometric air-fuel ratio.
DVoxs = Voxsref−Voxs (8)

Step 815: The CPU obtains a sub feedback amount Vafsfb according to the following equation (9). In this equation (9), Kp is a preset proportional gain (proportional constant), and Ki is a preset integral gain (integral constant). SDVoxs is an integral value of the output deviation amount DVoxs.
Vafsfb = Kp · DVoxs + Ki · SDVoxs (9)

  Step 820: The CPU obtains a new output deviation amount integrated value SDVoxs by adding “the output deviation amount DVoxs obtained in step 810” to “the integrated value SDVoxs of the output deviation amount at that time”.

  In this way, the CPU calculates the “sub feedback amount Vafsfb” by proportional-integral control for making the output value Voxs of the downstream air-fuel ratio sensor 56 coincide with the downstream target value Voxsref. This sub-feedback amount Vafsfb is used to calculate the feedback control output value Vabyfc, as shown in the above-described equation (2). The CPU may calculate the sub FB learning value Vafsfbg by PID (proportional integral differential control) by adding the differential term Kd · DDVox to the right side of the above equation (9). In this case, Kd is a preset differential gain (constant). DDVoxs is a differential value (time differential value) of the output deviation amount DVoxs.

  Next, the CPU sequentially calculates the “sub FB learning value Vafsfbg” by correcting the “sub feedback amount Vafsfbg” by sequentially performing the processes of steps 825 to 840 described below, and then proceeds to step 895 to execute this routine. Is temporarily terminated.

  Step 825: The CPU stores the current sub FB learning value Vafsfbg as the pre-update learning value Vafsfbgbf.

Step 830: The CPU updates the sub FB learning value Vafsfbg according to the following equation (10). The left side Vafsfbg (k + 1) of the equation (10) represents the updated sub FB learning value Vafsfbg, and the right side Vafsfbg (k) represents the pre-update sub FB learning value Vafsfbg. The value α is an arbitrary value larger than 0 and smaller than 1.
Vafsfbg (k + 1) = α · Vafsfbg + (1−α) · Ki · SDVoxs (10)

  As is clear from the equation (10), the sub FB learning value Vafsfbg is a value obtained by performing “filter processing for noise removal” on the “integral term Ki · SDVoxs of the sub feedback amount Vafsfb”. In other words, the sub FB learning value Vafsfbg is a value corresponding to the stationary component (integral term) of the sub feedback amount Vafsfb. The sub FB learning value Vafsfbg is stored in the backup RAM.

Step 835: The CPU calculates a change amount (update amount) ΔG of the sub FB learning value Vafsfbg according to the following equation (11).
ΔG = Vafsfbg−Vafsfbgbf (11)

Step 840: The CPU corrects the sub feedback amount Vafsfb with the change amount ΔG according to the following equation (12). In equation (12), Vafsfb (k + 1) is the corrected sub-feedback amount Vafsfb, and Vafsfbg (k) is the uncorrected sub-feedback amount Vafsfb.
Vafsfb (k + 1) = Vafsfb (k) −ΔG (12)

  The processing of step 835 and step 840 will be described. As shown in the above equation (2), the CPU adds the “sub feedback amount Vafsfb and the sub FB learning value Vafsfbg” to the “output value Vabyfs of the upstream air-fuel ratio sensor 55”, thereby providing the feedback control output value Vabyfc. Get. The sub FB learning value Vafsfbg is a value obtained by incorporating a part of the integral term Ki · SDVoxs (stationary component) of the sub feedback amount Vafsfb. Therefore, when the sub FB learning value Vafsfbg is updated, if the sub feedback amount Vafsfb is not corrected according to the updated amount, double correction is performed by “the updated sub FB learning value Vafsfbg and the updated sub feedback amount Vafsfb”. Is done. Therefore, when the sub FB learning value Vafsfbg is updated, the CPU needs to correct the sub feedback amount Vafsfb in accordance with “update amount ΔG of the sub FB learning value Vafsfbg”.

  Therefore, when the CPU updates the sub FB learning value Vafsfbg so as to increase by the change amount ΔG, as shown in the equations (11) and (12), the CPU decreases the sub feedback amount Vafsfb by the change amount ΔG. In equation (11), Vafsfbgbf is the sub FB learning value Vafsfbg immediately before the update, and Vafsfbg is the sub FB learning value Vafsfbg immediately after the update. Accordingly, the change amount ΔG is a positive value or a negative value.

  Through the above processing, the sub feedback amount Vafsfb and the sub FB learning value Vafsfbg are updated every time a predetermined time elapses.

On the other hand, if the sub-feedback control condition is not satisfied, the CPU makes a “No” determination at step 805 in FIG. 8, performs the processing of step 845 and step 850 described below in order, and proceeds to step 895 to proceed to this routine. Is temporarily terminated.
Step 845: The CPU sets the value of the sub feedback amount Vafsfb to “0”.
Step 850: The CPU sets the integrated value SDVoxs of the output deviation amount to “0”.

  Thereby, as is apparent from the above equation (2), the feedback control output value Vabyfc is the sum of the output value Vabyfs of the upstream air-fuel ratio sensor 55 and the sub FB learning value Vafsfbg. That is, in this case, “update of the sub feedback amount Vafsfb” and “reflection of the sub feedback amount Vafsfb to the final fuel injection amount Fi” are stopped. However, at least the “sub FB learning value Vafsfbg” corresponding to “the integral term of the sub feedback amount Vafsfb” is reflected in the final fuel injection amount Fi.

<Fuel cut control>
In order to execute the fuel cut control, the CPU executes the fuel cut control routine shown in FIG. 9 every elapse of a predetermined time. Therefore, when the predetermined timing comes, the CPU starts the process from step 900 and proceeds to step 910 to determine whether or not the value of the F / C flag XFC is “1”.

Assume that the value of the F / C flag XFC is “0” (that is, fuel cut control is not executed). In this case, the CPU makes a “No” determination at step 910 to proceed to step 920 to determine whether or not a fuel cut start condition is satisfied. The fuel cut start condition is satisfied only when both condition 1 and condition 2 described below are satisfied.
(Condition 1) The opening degree TA of the throttle valve 34 is “zero (or a predetermined opening degree TAth or less)”.
(Condition 2) The engine speed NE is “fuel cut start rotational speed NEfcth or higher”.

  At this time, if the fuel cut start condition is not satisfied, the CPU makes a “No” determination at step 920 to directly proceed to step 995 to end the present routine tentatively. Therefore, the value of the F / C flag XFC is maintained at “0”.

  On the other hand, if the fuel cut start condition is satisfied, the CPU makes a “Yes” determination at step 920 to proceed to step 930 to set the value of the F / C flag XFC to “1”. As a result, the CPU makes a “No” determination at step 610 in FIG. 6 to directly proceed to step 695. As a result, since step 650 is not executed, fuel injection is stopped (fuel cut control is executed).

  Referring to FIG. 9 again, the CPU proceeds to step 940 following step 930 to set the value of the first abnormal state determination execution flag XIL (hereinafter referred to as “determination execution flag XIL”) to “1”. . When the value of this determination execution flag XIL is set to “1”, it is determined whether or not “the first abnormal state in which the output value Vabyfs of the upstream air-fuel ratio sensor 55 deviates from the normal value” has occurred. Is executed (see FIG. 10 described later). Thereafter, the CPU proceeds to step 995 to end the present routine tentatively.

  After the value of the F / C flag XFC is set to “1” in this way, when the CPU starts the routine processing shown in FIG. 9 again, the CPU makes a “Yes” determination at step 910 following step 900. Then, the process proceeds to Step 950. In step 950, the CPU determines whether or not a fuel cut return condition (fuel cut control end condition) is satisfied.

The fuel cut return condition is satisfied when at least one of condition 3 and condition 4 described below is satisfied.
(Condition 3) The throttle valve opening TA is larger than “zero (or a predetermined opening TAth)”.
(Condition 4) The engine rotational speed NE is smaller than the “fuel cut return rotational speed NEfcre”. The fuel cut return rotational speed NEfcre is a rotational speed that is smaller than the fuel cut start rotational speed NEfcth by a predetermined rotational speed ΔN.

  Assume that the fuel cut return condition is not satisfied. In this case, the CPU makes a “No” determination at step 950 to directly proceed to step 995 to end the present routine tentatively. Therefore, the value of the F / C flag XFC is maintained at “1”. As a result, the fuel cut control is continuously executed.

  On the other hand, when the fuel cut return condition is satisfied, when the CPU proceeds to step 950, the CPU makes a “Yes” determination at step 950 to proceed to step 960 to set the value of the F / C flag XFC to “0”. Set. Thereafter, the CPU proceeds to step 995 to end the present routine tentatively. As a result, the CPU makes a “Yes” determination at step 610 in FIG. 6 to execute the processing from step 620 to step 650. Accordingly, fuel injection is resumed (fuel cut control is stopped).

<Determination of first abnormal state and setting of sub FB learning value>
Further, the CPU repeatedly executes the “sub FB learning value setting routine (first abnormal state determination routine)” shown in FIG. 10 every elapse of a predetermined time. Accordingly, when the predetermined timing comes, the CPU starts processing from step 1000 and proceeds to step 1010 to determine whether or not the value of the determination execution flag XIL is “1”. At this time, if the value of the determination execution flag XIL is not “1”, the CPU makes a “No” determination at step 1010 to proceed to step 1060 to set the value of the determination execution flag XIL to “0”. Thereafter, the CPU proceeds to step 1095 to end the present routine tentatively.

  On the other hand, when the fuel cut start condition is satisfied and the value of the determination execution flag XIL is set to “1” by executing the processing of step 940 in FIG. In Step 1010, the determination is “Yes”, and the processing of the next Step 1020 and Step 1030 is executed. The CPU determines that “Yes” is determined in Step 1010, and after a predetermined time (a time longer than the time required for the gas by the fuel cut control to reach the upstream air-fuel ratio sensor 55 from the combustion chamber 21). It may be configured to proceed to the following step 1020 or later. In other words, in this example, it is assumed that the predetermined time is substantially “0”. This point is also applied to other embodiments described later.

Step 1020: The CPU acquires the current output value Vabyfs of the upstream air-fuel ratio sensor 55.
Step 1030: The CPU changes the sensor output by subtracting “normal value Vabyfsfc during fuel cut (hereinafter referred to as“ normal value Vabyfsfc during FC ”)” from “output value Vabyfs acquired at step 1020”. Acquired as a quantity ΔV. As shown in FIG. 2, the normal value Vabyfsfc during FC is obtained when the upstream air-fuel ratio sensor 55 is normal and the fuel cut control is executed (that is, the upstream air-fuel ratio sensor 55 has a predetermined value). This is the output value Vabyfs output by the upstream air-fuel ratio sensor 55 (when the atmosphere, which is a gas having oxygen concentration) has reached.

  Next, the CPU proceeds to step 1040 to determine whether or not the magnitude (absolute value) of the sensor output change amount ΔV is larger than the abnormality determination threshold ε (ε> 0). At this time, if the magnitude of the sensor output change amount ΔV is equal to or smaller than the abnormality determination threshold ε, it can be determined that the first abnormal state has not occurred. Therefore, when the magnitude of the sensor output change amount ΔV is equal to or smaller than the abnormality determination threshold ε, the CPU makes a “No” determination at step 1040 to proceed to step 1060 to set the value of the determination execution flag XIL to “0”. To do. Thereafter, the CPU proceeds to step 1095 to end the present routine tentatively.

  On the other hand, when the “sensor output value expansion abnormality” in the first abnormal state has occurred, as shown in FIG. 2, “the output value Vabyfs acquired in step 1020” is the normal value Vabyfsfc during FC. Much larger than. That is, the sensor output change amount ΔV is larger than the abnormality determination threshold value ε. In this case, the CPU makes a “Yes” determination at step 1040 to proceed to step 1050. Similarly, when the “sensor output value reduction abnormality” in the first abnormal state has occurred, as shown in FIG. 2, “the output value Vabyfs acquired in step 1020” is obtained from the normal value Vabyfsfc during FC. Is considerably smaller. That is, the absolute value of the sensor output change amount ΔV is larger than the abnormality determination threshold value ε. Also in this case, the CPU makes a “Yes” determination at step 1040 to proceed to step 1050.

  In step 1050, the CPU determines a predicted convergence value Vafsfbgy of the sub FB learning value Vafsfbg and sets the predicted convergence value Vafsfbgy as the sub FB learning value Vafsfbg. As a result, the sub FB learning value Vafsfbg is matched with the convergence expected value Vafsfbgy.

  More specifically, the CPU stores the relationship between the sensor output change amount ΔV and the expected convergence value Vafsfbgy in the ROM in the form of a lookup table. The predicted convergence value Vafsfbgy is a sub-range when the output value Vabyfs during fuel cut control of the upstream air-fuel ratio sensor 55 is in a first abnormal state in which the sensor output change amount ΔV deviates from the normal value Vabyfsfc during FC. It is a convergence value that the FB learning value Vafsfbg will converge. The data in this table is data obtained in advance by experiments.

  According to this table, as shown in block 1050 of FIG. 10, when the sensor output change amount ΔV is “0”, the convergence expected value Vafsfbgy becomes the reference value Vafsfbg0. Further, according to this table, the predicted convergence value Vafsfbgy is determined so as to increase as the sensor output change amount ΔV increases. That is, for example, when the sensor output change amount ΔV is ΔVL (ΔVL> 0) due to a sensor output value expansion abnormality, the convergence expected value Vafsfbgy is set to “VafsfbgL larger than the reference value Vafsfbg0”. Further, for example, when the sensor output change amount ΔV is ΔVR (ΔVR <0) due to a sensor output value reduction abnormality, the convergence expected value Vafsfbgy is set to “VafsfbgyR smaller than the reference value Vafsfbg0”. Thereafter, the CPU proceeds to step 1060 to set the value of the determination execution flag XIL to “0”, and proceeds to step 1095 to end the present routine tentatively.

  The time chart of FIG. 5 described above shows a change in the sub FB learning value Vafsfbg when the “sensor output value expansion abnormality” occurs such that the sensor output change amount ΔV becomes ΔVL. As shown in FIG. 5, according to the first control device, when it is determined that “sensor output value expansion abnormality” has occurred (immediately after time t1), the sub FB learning value Vafsfbg is the convergence expected value Vafsfbgy. Matched to some value VafsfbgL. Therefore, the sub FB learning value Vafsfbg becomes a value close to the appropriate value (convergence value) from the time (time t2) when the sub feedback control is resumed after the end of the fuel cut control. As a result, it is possible to avoid deterioration of emissions.

As described above, the first control device
Sub-feedback amount calculation means for calculating a sub-feedback amount for matching the output value of the downstream air-fuel ratio sensor with the downstream target value (see step 810 to step 820 in FIG. 8);
Learning means for performing learning for updating a learning value based on a value corresponding to a stationary component included in the sub feedback amount and correcting the sub feedback amount according to the updated learning value (steps 825 to 820 in FIG. 8) See step 840).
Air-fuel ratio feedback control for matching the air-fuel ratio of the air-fuel mixture supplied to the engine to the stoichiometric air-fuel ratio based on the output value of the upstream air-fuel ratio sensor, the corrected sub-feedback amount, and the updated learned value. Air-fuel ratio control means to be executed (see each step in FIG. 7 and step 640 in FIG. 6);
An air-fuel ratio control apparatus for an internal combustion engine comprising:
A first abnormal state in which an actual output value that is an actual output value of the upstream air-fuel ratio sensor is different from a normal value that is an output value of the upstream air-fuel ratio sensor when the upstream air-fuel ratio sensor is normal. First abnormality determination means (see step 1020 to step 1040 in FIG. 10) for determining whether or not it has occurred;
The output value of the upstream air-fuel ratio sensor when the predetermined gas reaches when the predetermined gas (for example, exhaust under fuel cut control, that is, the atmosphere) has reached the upstream air-fuel ratio sensor, and the learning The relationship with the predicted convergence value that is the convergence value of the value (relationship shown in the block of step 1050 in FIG. 10) is stored in advance, and when it is determined that the first abnormal state has occurred, the predetermined gas reaches The actual convergence predicted value of the learning value is acquired based on the actual output value at time (the output value Vabyfs acquired in step 1020 in FIG. 10) and the relationship, and the actual value obtained by acquiring the learning value at the same time Learning value setting means for setting the predicted convergence value (see step 1040 and step 1050 in FIG. 10);
Is an air-fuel ratio control device.

  Therefore, when the upstream air-fuel ratio sensor 55 enters the first abnormal state, the sub FB learning value Vafsfbg is immediately changed to the expected convergence value (convergence expected value Vafsfbgy). Therefore, the sub FB learning value Vafsfbg becomes a value close to an appropriate value immediately after the first abnormal state occurs. Thereby, the 1st control apparatus can shorten time from the time of the 1st abnormal state occurrence to the time when sub FB learning value Vafsfbg converges. As a result, it is possible to avoid “deterioration of emissions due to the deviation of the sub FB learning value Vafsfbg from the appropriate value”.

Second Embodiment
Hereinafter, an air-fuel ratio control apparatus for an internal combustion engine according to a second embodiment of the present invention (hereinafter simply referred to as “second control apparatus”) will be described. The second control device replaces the table (the table shown in the block of step 1050 in FIG. 10) used when the first control device determines the convergence expected value Vafsfbgy with the table shown in FIG. Only the first control device is different.

  The table shown in FIG. 11 is a table in which the expected convergence value Vafsfbgy is determined for each region to which the actual engine speed NE belongs. The expected convergence value Vafsfbgy changes according to the engine speed NE. More specifically, the expected convergence value Vafsfbgy changes as shown by the line L1 in FIG. 11 when the engine speed NE is smaller than 1500 rpm. The expected convergence value Vafsfbgy changes as indicated by the line L2 in FIG. 11 when the engine speed NE is 1500 rpm or higher and smaller than 3000 rpm. Further, the expected convergence value Vafsfbgy changes as indicated by a line L3 in FIG. 11 when the engine speed NE is 3000 rpm or more.

  Therefore, when the CPU of the second control device proceeds to step 1050, it obtains the predicted convergence value Vafsfbgy based on the engine speed NE at that time, the sensor output change amount ΔV, and the table shown in FIG. Then, the predicted convergence value Vafsfbgy is stored in the sub FB learning value Vafsfbg. As a result, the sub FB learning value Vafsfbg immediately after the occurrence of the first abnormal condition is set to a value close to an appropriate value even when the predicted convergence value Vafsfbgy when the first abnormal condition occurs differs depending on the engine speed NE. can do.

  The second control device can also be configured as follows. That is, when it is determined that the first abnormal condition has occurred (that is, when the process proceeds to step 1050), the second control device determines that “engine speed is based on the sensor output change ΔV and the table shown in FIG. The predicted convergence value Vafsfbgy is determined for each NE area, and the predicted convergence value Vafsfbgy is stored in the backup RAM. After that, every time the engine speed NE changes to a different area, the convergence expected value Vafsfbgy corresponding to the changed area is read from the backup RAM, and the read convergence expected value Vafsfbgy is read as the sub FB learning value. It may be configured to set in Vafsfbg.

  Furthermore, the table shown in FIG. 11 can replace the predicted convergence value Vafsfbgy with a table determined for each region to which the load KL of the engine 10 belongs. In this case, when it is determined that the first abnormal condition has occurred (that is, when the process proceeds to step 1050), the second control device determines that the load of the engine 10 is based on the sensor output change ΔV and the replaced table. The predicted convergence value Vafsfbgy is determined for each KL area ", and the predicted convergence value Vafsfbgy is stored in the backup RAM. After that, every time the engine load KL changes to a different region, the predicted convergence value Vafsfbgy corresponding to the changed region is read from the backup RAM, and the read predicted convergence value Vafsfbgy is read as the sub FB learning value. Configured to set to Vafsfbg.

  In addition, the table shown in FIG. 11 can replace the predicted convergence value Vafsfbgy with a table determined for each region to which the operating state determined by the engine speed NE and the load KL of the engine 10 belongs. In this case, when it is determined that the first abnormal condition has occurred (that is, when the process proceeds to step 1050), the second control device determines that the load of the engine 10 is based on the sensor output change ΔV and the replaced table. The predicted convergence value Vafsfbgy is determined for each KL area ", and the predicted convergence value Vafsfbgy is stored in the backup RAM. After that, every time the operating state determined by the engine rotational speed NE and the load KL of the engine 10 changes, the convergence expected value Vafsfbgy corresponding to the changed region is read from the backup RAM, and the read convergence prediction The value Vafsfbgy is configured to be set to the sub FB learning value Vafsfbg.

<Third Embodiment>
Hereinafter, an air-fuel ratio control apparatus for an internal combustion engine according to a third embodiment of the present invention (hereinafter simply referred to as “third control apparatus”) will be described.

(Outline of operation of third control device)
The sub FB learning value Vafsfbg is not only the case where the “first abnormal state” such as “sensor output value expansion abnormality and sensor output value reduction abnormality” described above has occurred, but also other abnormal states different from the first abnormal state ( Even when the second abnormal state occurs, the value converges to a value different from that before the second abnormal state occurs.

For example, an abnormal state (fuel injection valve rich deviation abnormality) occurs in which the fuel injection valve 25 provided for a specific cylinder among a plurality of cylinders injects “a larger amount of fuel than the instructed fuel amount”. In this case, HC, CO and H ₂ discharged from the specific cylinder increase rapidly. On the other hand, since the molecular diameter of hydrogen molecules is smaller than that of other unburned substances, the hydrogen molecules diffuse more quickly in the diffusion resistance layer of the upstream air-fuel ratio sensor 55. As a result, the upstream air-fuel ratio sensor 55 is affected by H _{2 having} a small molecular diameter, and generates an output value Vabyfs corresponding to the air-fuel ratio richer than the actual average air-fuel ratio.

  As a result, the center of the engine air-fuel ratio shifts to the lean side from the stoichiometric air-fuel ratio by main feedback control based on the output value of the upstream air-fuel ratio sensor 55. Therefore, the sub-feedback amount Vafsfb changes to “a value that shifts the air-fuel ratio of the engine 10 to the richer side” than immediately before “fuel injection valve rich deviation abnormality” occurs. Therefore, the sub FB learning value Vafsfbg also converges to “a value that shifts the air-fuel ratio of the engine 10 to the richer side”.

  Further, for example, when misfire occurs (when the air-fuel mixture does not burn in the combustion chamber), the sub FB learning value Vafsfbg is greater than the sub FB learning value Vafsfbg immediately before the misfire occurs. May converge to a value that shifts to the lean side.

  These “abnormality deviation of fuel injection valve (air-fuel ratio imbalance between cylinders)” and “misfire” are representative examples of the second abnormal state described above. When such a second abnormal state occurs, if the first abnormal state occurs repeatedly, the sub FB learning value Vafsfbg is determined (predicted) in step 1050 of FIG. It converges to a value different from the “convergence expected value Vafsfbgy at the time of occurrence (when the first abnormal state alone occurs)”. Accordingly, when the first abnormal condition occurs repeatedly when the second abnormal condition occurs, the sub FB learning value Vafsfbg is expressed as “the convergence expected value Vafsfbgy expected when the first abnormal condition occurs alone. On the other hand, there is a possibility that the sub FB learning value Vafsfbg may be moved away from the “actual convergence value”.

  FIG. 12 is a time chart showing an example of the sub FB learning value Vafsfbg in such a situation. In the example shown in FIG. 12, misfire as a second abnormal state occurs at time t1. Therefore, the sub FB learning value Vafsfbg decreases from the value Vafsfbg0 toward the value VafsfbgA after time t1, and converges to the value VafsfbgA after time t2.

  Thereafter, a sensor output value expansion abnormality as a first abnormal state occurs at time t3. In this case, the first control device sets the sub feedback amount Vafsfb to the convergence expected value Vafsfbgy at time t3.

  However, actually, the sub FB learning value Vafsfbg “converges to a value VafsfbgB between the value VafsfbgA and the value Vafsfbgy. Therefore, the first control device sets the sub FB learning value Vafsfbg to“ actual convergence value ”. ”Away from it.

  Therefore, the third control device does not set the sub FB learning value Vafsfbg to the “convergence expected value Vafsfbgy” when the first abnormal state occurs repeatedly when the second abnormal state occurs. (See the solid line after time t3 in FIG. 3). Therefore, the possibility that the sub feedback amount Vafsfb deviates from the actual convergence value can be reduced. As a result, it is possible to avoid the emission from getting worse.

(Actual operation)
Next, the actual operation of the third control device will be described.
The CPU of the third control device executes the routines shown in FIGS. 6 to 9 in the same manner as the CPU of the first control device. Further, the CPU of the third control device repeatedly executes the “sub FB learning value setting routine (first abnormal state determination routine)” shown in FIG. 13 instead of FIG. 10 every elapse of a predetermined time. The routines shown in FIGS. 6 to 9 have been described. Therefore, the following description will focus on the routine shown in FIG. Of the steps shown in FIG. 13, steps that perform the same processing as the steps shown in FIG. 10 are denoted by the same reference numerals as those given for such steps in FIG. 10.

  When the predetermined timing comes, the CPU starts processing from step 1300 in FIG. 13 and proceeds to step 1010 to determine whether or not the value of the determination execution flag XIL is “1”. As described above, the value of the determination execution flag XIL is set to “1” by the process of step 940 in FIG. 9 when the fuel cut start condition is satisfied.

  Assume that the value of the determination execution flag XIL is not “1”. According to this assumption, the CPU makes a “No” determination at step 1010 to proceed to step 1060 to set the value of the determination execution flag XIL to “0”. Thereafter, the CPU proceeds to step 1395 to end the present routine tentatively.

  On the other hand, when the value of determination execution flag XIL is set to “1” by executing the processing of step 940 in FIG. 9 in accordance with the establishment of the fuel cut start condition, the CPU executes step 1010 in FIG. Is determined as “Yes”, and the processing of the next step 1020 and step 1030 is executed. As a result, “sensor output change amount ΔV” that is a value obtained by subtracting the normal value Vabyfsfc during FC from the actual output value Vabyfs is acquired.

  Next, the CPU proceeds to step 1040 to determine whether or not the magnitude (absolute value) of the sensor output change amount ΔV is larger than the abnormality determination threshold ε (ε> 0). At this time, if the magnitude of the sensor output change amount ΔV is equal to or smaller than the abnormality determination threshold ε, it can be determined that the first abnormal state has not occurred. Therefore, when the magnitude of the sensor output change amount ΔV is equal to or smaller than the abnormality determination threshold ε, the CPU makes a “No” determination at step 1040 to proceed to step 1060 to set the value of the determination execution flag XIL to “0”. To do. Thereafter, the CPU proceeds to step 1395 to end the present routine tentatively.

  On the other hand, when the first abnormal state occurs, the magnitude of the sensor output change ΔV becomes larger than the abnormality determination threshold ε. In this case, the CPU makes a “Yes” determination at step 1040 to proceed to step 1310 to determine whether or not the above-described second abnormal state has occurred.

  Whether or not the second abnormal state has occurred is determined by a “second abnormal state occurrence determination routine” (not shown). For example, the CPU determines that the second abnormal state has occurred when the sub FB learning value Vafsfbg is equal to or greater than the second abnormal state determination threshold. Alternatively, the CPU can determine whether or not “misfire as the second abnormal state” has occurred based on the changing speed of the engine rotational speed NE.

  At this point, if the second abnormal state has not occurred, the CPU makes a “No” determination at step 1310 to execute the processing of step 1050 and step 1060 described above, and then proceeds to step 1395 to execute this routine. Is temporarily terminated. As a result, the sub-FB learning value Vafsfbg is set to the convergence expected value Vafsfbgy immediately after the first abnormal state occurs alone, as in the first control device. Accordingly, the third control device can also avoid the emission deterioration immediately after the first abnormal state occurs alone.

  On the other hand, if the second abnormal state has occurred at the time when the CPU executes the process of step 1310, the CPU makes a “Yes” determination at step 1310 to proceed directly to step 1060, and then proceeds to step 1395. Proceed to end this routine. As a result, even if it is determined that the first abnormal state has occurred (when it is determined “Yes” in step 1040), if the second abnormal state has already occurred, the sub FB learning value Vafsfbg is not set to the expected convergence value Vafsfbgy.

  As a result, the third control device can reduce the possibility that the sub FB learning value Vafsfbg deviates from the actual convergence value. As a result, it is possible to avoid the emission from getting worse.

<Fourth embodiment>
Hereinafter, an air-fuel ratio control apparatus for an internal combustion engine according to a fourth embodiment of the present invention (hereinafter simply referred to as “fourth control apparatus”) will be described.

(Outline of operation of the fourth control device)
In the fourth control device, when the first abnormal state occurs repeatedly when the second abnormal state occurs, the convergence value of the sub FB learning value Vafsfbg changes along with the occurrence of the first abnormal state. This is a device that changes the sub FB learning value Vafsfbg by the amount to be used.

  In other words, when the first abnormal state occurs, the fourth control device has a relationship between “the sensor output change amount ΔV” indicating the degree of the abnormality and the change amount of the sub FB learning value Vafsfbg (hereinafter referred to as “learning value difference”). "Relationship") is stored in advance.

More specifically, the learning value difference relationship is a relationship between the following difference Δ1 and the following difference Δ2.
The difference Δ1 is “the output value Vabyfs of the upstream air-fuel ratio sensor 55” when a predetermined gas (for example, the atmosphere that is the gas under fuel cut control) reaches the upstream air-fuel ratio sensor 55 (when the predetermined gas arrives). And “a normal value when a predetermined gas arrives (normal value Vabyfsfc during FC)”. When the difference Δ1 is a difference obtained during the fuel cut control, it coincides with the difference ΔV.
The difference Δ2 is expressed as “the predicted convergence value Vafsfbgy of the sub FB learning value Vafsfbg when the first abnormal state in which the upstream air-fuel ratio sensor 55 causes the difference Δ1 alone occurs” and “the upstream air-fuel ratio sensor 55 Is the difference from the convergence value of the sub FB learning value Vafsfbg expected when normal is normal (convergence expected value at normal time) Vafsfbg0 ”.

  Then, the fourth control device obtains the difference Δ2 from the learned value difference relationship and the obtained difference Δ1 together with “obtains the difference Δ1” when the first abnormal state occurs, and obtains the sub FB learned value Vafsfbg at that time. A value obtained by adding “the obtained difference Δ2” is set as a new sub FB learning value Vafsfbg. The sub FB learning value Vafsfbg at the time of occurrence of the first abnormal state is a value near the convergence value of the sub FB learning value Vafsfbg in the state where the second abnormal state has occurred before the second abnormal state has occurred. It is highly possible that Therefore, according to the fourth control apparatus, even when the first abnormal state occurs repeatedly when the second abnormal state occurs, the sub FB learning value Vafsfbg can be immediately brought close to the appropriate value. Therefore, it is possible to avoid the emission from deteriorating.

  FIG. 14 is a time chart showing an example of the sub FB learning value Vafsfbg in such a situation. In the example shown in FIG. 14, misfire as a second abnormal state occurs at time t1. Therefore, the sub FB learning value Vafsfbg decreases from the value Vafsfbg0 toward the value VafsfbgA after time t1, and converges to the value VafsfbgA after time t2.

  Thereafter, a sensor output value expansion abnormality as a first abnormal state occurs at time t3. In this case, the third control device according to the third embodiment does not change the sub feedback amount Vafsfb at time t3. Therefore, the sub FB learning value Vafsfbg converges to the value VafsfbgB at time t4.

  On the other hand, the fourth control device changes (in this case, increases) “sub-FB learning value Vafsfbg” by “difference Δ2” at time t3 (see the solid line in FIG. 14). As a result, the sub FB learning value Vafsfbg quickly approaches the actual convergence value, so that it is possible to avoid the emission from deteriorating.

(Actual operation)
Next, the actual operation of the fourth control device will be described.
The CPU of the fourth control device executes the routines shown in FIGS. 6 to 9 in the same manner as the CPU of the first control device. Further, the CPU of the fourth control device repeatedly executes the “sub FB learning value setting routine (first abnormal state determination routine)” shown in FIG. 15 instead of FIG. 10 every elapse of a predetermined time. The routines shown in FIGS. 6 to 9 have been described. Therefore, the following description will focus on the routine shown in FIG. Of the steps shown in FIG. 15, steps that perform the same processing as the steps shown in FIG. 10 are denoted by the same reference numerals as those given for such steps in FIG. 10.

  When the predetermined timing is reached, the CPU starts processing from step 1500 in FIG. 15 and proceeds to step 1010 to determine whether or not the value of the determination execution flag XIL is “1”. As described above, the value of the determination execution flag XIL is set to “1” by the process of step 940 in FIG. 9 when the fuel cut start condition is satisfied.

  Assume that the value of the determination execution flag XIL is not “1”. According to this assumption, the CPU makes a “No” determination at step 1010 to proceed to step 1060 to set the value of the determination execution flag XIL to “0”. Thereafter, the CPU proceeds to step 1595 to end the present routine tentatively.

  On the other hand, if the value of determination execution flag XIL is set to “1” by executing the process of step 940 in FIG. 9 in accordance with the establishment of the fuel cut start condition, the CPU executes step 1010 in FIG. Is determined as “Yes”, and the processing of the next step 1020 and step 1030 is executed. As a result, “sensor output change amount ΔV = Δ1”, which is a value obtained by subtracting the normal value Vabyfsfc during FC from the actual output value Vabyfs, is acquired.

Next, the CPU proceeds to step 1040 to determine whether or not the magnitude (absolute value) of the sensor output change amount ΔV is larger than the abnormality determination threshold ε (ε> 0). At this time, if the magnitude of the sensor output change amount ΔV is equal to or smaller than the abnormality determination threshold ε, it can be determined that the first abnormal state has not occurred. Therefore, when the magnitude of the sensor output change amount ΔV is equal to or less than the abnormality determination threshold ε,
The CPU makes a “No” determination at step 1040 to proceed to step 1060 to set the value of the determination execution flag XIL to “0”. Thereafter, the CPU proceeds to step 1595 to end the present routine tentatively.

  On the other hand, when the first abnormal state occurs, the magnitude of the sensor output change ΔV becomes larger than the abnormality determination threshold ε. In this case, the CPU makes a “Yes” determination at step 1040 to proceed to step 1510. By applying “the sensor output change ΔV acquired at step 1030” to the “learning value difference relationship” described above, “ The difference Δ2 (that is, the correction amount ΔVafsfbg of the sub FB learning value Vafsfbg) ”is acquired. In this example, the difference Δ2 (correction amount ΔVafsfbg) increases as the sensor output change amount ΔV increases. The difference Δ2 (correction amount ΔVafsfbg) takes a positive value when the sensor output change amount ΔV is a positive value, and takes a positive value when the sensor output change amount ΔV is a negative value. In this case as well, as in the second embodiment, the learning value difference relationship may be further set for each engine speed NE region and / or for each engine load KL.

  Next, the CPU proceeds to step 1520 to update the sub FB learning value Vafsfbg by adding the difference Δ2 (correction amount ΔVafsfbg) to the sub FB learning value Vafsfbg at that time. That is, “the sum of the sub FB learning value Vafsfbg immediately before that point and the difference Δ2” is set in the sub FB learning value Vafsfbg. Thereafter, the CPU proceeds to step 1595 to end the present routine tentatively.

  As a result, when it is determined that the first abnormal state has occurred, the learning value Vafsfbg is corrected (increased or decreased) by the difference Δ2. The difference Δ2 is the amount of change in the learning value Vafsfbg when the first abnormal state occurs alone. Therefore, according to the fourth control device, the sub FB learning value Vafsfbg immediately approaches an appropriate value regardless of whether or not the second abnormal state has occurred when the first abnormal state has occurred. Therefore, it is possible to avoid the emission from deteriorating.

  As described above, the air-fuel ratio control apparatus according to each embodiment of the present invention promptly sets the sub FB learning value Vafsfbg to an appropriate value (convergence) when the first abnormal state occurs in the upstream air-fuel ratio sensor 55. Value). Therefore, these air-fuel ratio control devices can avoid “deterioration of emissions due to the deviation of the sub FB learning value Vafsfbg from the appropriate value”.

  Various modifications can be employed within the scope of the present invention. For example, the air-fuel ratio control apparatus according to each of the above embodiments sets the update rate of the sub FB learning value Vafsfbg to a value other than the first period until a predetermined time elapses after the first abnormal state occurs (first period). The update speed of the sub FB learning value Vafsfbg during the period of time may be increased. The update speed of the sub FB learning value Vafsfbg can be easily increased by changing the value α in step 830 in FIG. 8 to a small value, for example. According to this, the time from the occurrence of the first abnormal state to the convergence of the sub FB learning value Vafsfbg can be further shortened.

  Further, for example, as disclosed in JP-A-2005-273524, for example, the air-fuel ratio feedback control includes a main feedback amount based on a value obtained by high-pass filtering the output value of the upstream air-fuel ratio sensor, and a downstream air-fuel ratio sensor. Alternatively, a sub-feedback amount based on a value obtained by performing low-pass filter processing on the output value may be calculated independently, and air-fuel ratio feedback control may be executed based on the sub-feedback amount. Further, in the air-fuel ratio feedback control, as disclosed in, for example, Japanese Patent Laid-Open No. 06-010738, a sub feedback amount is calculated based on the output value Voxs of the downstream air-fuel ratio sensor 56, and the upstream is calculated based on the sub feedback amount. The main feedback amount (air / fuel ratio correction coefficient) created based on the output value of the side air / fuel ratio sensor may be changed.

  In addition, each of the air-fuel ratio control devices performs the determination of the first abnormal state during the fuel cut control. Instead, the air-fuel ratio control apparatus has a reference upstream air-fuel ratio sensor disposed in a position adjacent to the upstream air-fuel ratio sensor 55 in addition to the upstream air-fuel ratio sensor 55, and the reference upstream air-fuel ratio sensor 55 The first abnormal condition may be determined based on the difference between the output value of the side air-fuel ratio sensor and the output value of the upstream air-fuel ratio sensor 55.

1 is a schematic diagram of an internal combustion engine to which an air-fuel ratio control device (first control device) according to a first embodiment of the present invention is applied. 2 is a graph showing a relationship among an applied voltage, an output value (limit current value), and an air-fuel ratio for the upstream air-fuel ratio sensor shown in FIG. 2 is a graph showing the relationship between the upstream air-fuel ratio and the output value of the upstream air-fuel ratio sensor shown in FIG. 2 is a graph showing the relationship between the downstream air-fuel ratio and the output value of the downstream air-fuel ratio sensor shown in FIG. It is a time chart for demonstrating the action | operation of a 1st control apparatus. 2 is a flowchart showing a fuel injection control routine executed by a CPU shown in FIG. 2 is a flowchart showing a routine executed by the CPU shown in FIG. 1 to calculate a main feedback amount. 2 is a flowchart showing a routine executed by the CPU shown in FIG. 1 to calculate a sub feedback amount and a sub FB learning value. This is a routine for setting a flag for the CPU shown in FIG. 1 to execute fuel cut control. 3 is a flowchart showing a routine executed by the CPU shown in FIG. 1 to set a sub FB learning value to an appropriate value when a first abnormal state occurs. It is a table referred when CPU of the air fuel ratio control apparatus (2nd control apparatus) which concerns on 2nd Embodiment of this invention determines the convergence expected value of a sub FB learning value. It is a time chart for demonstrating the action | operation of the air fuel ratio control apparatus (3rd control apparatus) which concerns on 3rd Embodiment of this invention. It is the flowchart which showed the routine which CPU of a 3rd control apparatus performs in order to set a sub FB learning value to an appropriate value at the time of 1st abnormal condition generation | occurrence | production. It is a time chart for demonstrating the action | operation of the air fuel ratio control apparatus (4th control apparatus) which concerns on 4th Embodiment of this invention. It is the flowchart which showed the routine performed in order that the CPU of a 4th control apparatus may set a sub FB learning value to an appropriate value at the time of 1st abnormal condition generation | occurrence | production.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 10 ... Internal combustion engine, 20 ... Main part, 21 ... Combustion chamber, 25 ... Fuel injection valve, 30 ... Intake system, 34 ... Throttle valve, 40 ... Exhaust system, 41 ... Exhaust manifold, 42 ... Exhaust pipe, 43 ... Upstream side Catalyst, 44 ... downstream catalyst, 55 ... upstream air-fuel ratio sensor, 56 ... downstream air-fuel ratio sensor, 60 ... electric control device.

Claims (3)

A catalyst disposed in an exhaust passage of the internal combustion engine;
An upstream air-fuel ratio sensor that is disposed in a portion upstream of the catalyst in the exhaust passage and outputs an output value corresponding to the air-fuel ratio of the gas flowing through the disposed portion;
A downstream air-fuel ratio sensor that is disposed in a portion downstream of the catalyst in the exhaust passage and outputs an output value corresponding to the air-fuel ratio of the gas flowing through the disposed portion;
Sub-feedback amount calculation means for calculating a sub-feedback amount for matching the output value of the downstream air-fuel ratio sensor with the downstream target value;
Learning means for performing learning to update a learning value based on a value corresponding to a stationary component included in the sub feedback amount and correcting the sub feedback amount according to the updated learning value;
Air-fuel ratio feedback control for matching the air-fuel ratio of the air-fuel mixture supplied to the engine to the stoichiometric air-fuel ratio based on the output value of the upstream air-fuel ratio sensor, the corrected sub-feedback amount, and the updated learned value. Air-fuel ratio control means to be executed;
An air-fuel ratio control apparatus for an internal combustion engine comprising:
A first abnormal state in which an actual output value that is an actual output value of the upstream air-fuel ratio sensor is different from a normal value that is an output value of the upstream air-fuel ratio sensor when the upstream air-fuel ratio sensor is normal. First abnormality determination means for determining whether or not it has occurred;
The relationship between the output value of the upstream air-fuel ratio sensor when the predetermined gas reaches the upstream air-fuel ratio sensor, which is the time when the predetermined gas arrives, and the predicted convergence value that is the convergence value of the learning value are stored in advance. In addition, when it is determined that the first abnormal state has occurred, an actual convergence expected value of the learning value is acquired based on the actual output value and the relationship when the predetermined gas reaches, and the learning value is obtained. A learning value setting means for setting the actual convergence expected value acquired,
An air-fuel ratio control device. An air-fuel ratio control apparatus for an internal combustion engine according to claim 1,
The learning value setting means includes
Determining whether or not a second abnormal state other than the first abnormal state and a convergence value of the learning value is different before and after the occurrence of the second abnormal state has occurred, When it is determined that the first abnormal state has occurred, if it is determined that the second abnormal state has already occurred, the learning value is not set to the acquired actual convergence expected value. An air-fuel ratio control device configured as described above. The air-fuel ratio control apparatus for an internal combustion engine according to claim 1,
The learning value setting means includes
When the relationship between the output value of the upstream air-fuel ratio sensor when the predetermined gas arrives and the normal value when the predetermined gas arrives, the expected convergence value, and the upstream air-fuel ratio sensor are normal Is stored as a learning value difference relationship indicating a relationship between the predicted value and the normal convergence expected value, which is a convergence value of the learning value, and when it is determined that the first abnormal state has occurred, The amount of change in the learned value is predicted based on the difference between the actual output value at the time of gas arrival and the normal value at the time of arrival of the predetermined gas, and the learning value difference relationship, An air-fuel ratio control apparatus configured to acquire a value obtained by changing the learning value in as the actual expected convergence value.

Images