JP4134476B2 - Vehicle control apparatus and recording medium - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a vehicle control device that controls a vehicle, and more particularly to a vehicle control device that learns a steady control amount deviation in feedback control.
[0002]
[Prior art]
2. Description of the Related Art Conventionally, in a vehicle control device that calculates a control amount related to control of a vehicle, such as an air-fuel ratio that is a ratio of fuel and air supplied to the engine of the vehicle and an engine speed, the air-fuel ratio and the speed In order to match (converge) the control amount with the target value, correction of the control amount according to the environment / state is widely performed in general. For example, correction is performed based on input signals from various sensors such as a water temperature sensor and an atmospheric pressure sensor, and the control amount is calculated. However, it is theoretically impossible to make corrections that take into account all the factors that change from moment to moment. Therefore, feedback control is combined and correction based on the feedback amount is also performed.
[0003]
In this feedback control, the “steady deviation” of the control amount from the target value is calculated and stored as a learning value in order to quickly match the control amount with the target value from the time when the ignition switch is turned on. Is common.
An example of conventional feedback control will be described with reference to FIG.
[0004]
FIG. 6 is a timing chart showing the control amount, the feedback amount, and the learning value in association with each other. The control amount calculated at time T0 is shown as the base control amount. This is a control amount calculated in a state where no correction is performed. This controlled variable is converged to the target value.
[0005]
First, looking at the period from time T0 to T1 (hereinafter, the period from time α to time β is described as period [α, β]), the control amount is below the target value, so the feedback amount is reduced. It is increased at a constant rate and corrected so that the control amount approaches the target value.
[0006]
At time T1, the feedback amount is decreased (skip), and the amount is calculated as a learning value. This is as indicated by the symbol k in FIG.
Similarly, in both the period [T1, T2] and the period [T2, T3], the control amount is lower than the target value, so the feedback amount is increased at a constant rate. At times T2 and T3, the feedback amount is decreased and the amount is added to the learning value.
[0007]
Therefore, in this example, the control amount corrected from the base control amount at an arbitrary time in the timing chart is shown as the sum of the feedback amount and the learning value. For example, at time t1, the control amount is corrected by the feedback amount indicated by symbol A and the learning value indicated by symbol B.
[0008]
Thereafter, in the period [T3, T4], the control amount exceeds the target value at the time t2, so the feedback amount is decreased from the time t2.
The feedback amount is skipped so as to approach the reference value “0”. For example, at time T5, since the feedback amount is less than “0”, the feedback amount is increased and the learning value is decreased accordingly.
[0009]
That is, the feedback amount is for correcting an instantaneous deviation from the target value, and the learning value corrects the deviation between the base control amount and the target value for this feedback amount. That is, the deviation between the base control amount and the target value in FIG. 6 corresponds to the “steady deviation” described above.
[0010]
Here, in order to facilitate understanding, the learning value is a direct correction amount. However, the learning value only needs to be able to correct the above-described steady deviation.
If the “steady deviation” from the target value is stored as a learned value as described above, the control amount can be quickly converged to the target value from the time when the ignition switch of the vehicle is turned on. .
[0011]
For example, as shown in FIG. 7, a case is considered where the ignition switch of the vehicle is turned off at time t1 in a state where the control amount has converged to the target value. Then, when the ignition switch is turned on again at time t2, that is, after the operation is stopped in the period “t1, t2”, if the deviation between the target value and the base control amount is stored as the learning value, As shown by the solid line, the control amount quickly converges to the target value, and if the learning value is not stored, the correction is made from the base control amount, so the convergence of the control amount is delayed as shown by the two-dot chain line. become.
[0012]
As described above, conventionally, a configuration in which a steady deviation in feedback control is stored as a learning value is employed, and a device for shortening the time required for convergence of the control amount has been devised. This is because in a state where the control amount does not converge to the target value, an unfavorable situation occurs in terms of control. For example, it is conceivable that the rotational speed fluctuates or the emission deteriorates.
[0013]
[Problems to be solved by the invention]
As described above, in the conventional configuration that absorbs the “steady deviation” from the target value by calculating the learning value, the feedback amount takes a value within a predetermined range from a reference value such as “0”, for example. It has become. This is because, for example, a predetermined storage area on the RAM is allocated to the feedback amount, but on the premise that correction using the learning value is performed, a storage area capable of expressing a value in a range corresponding thereto is allocated. Because. That is, the maximum value that can be expressed in the storage area is the upper limit value, and the minimum value is the lower limit value.
[0014]
On the other hand, the above-mentioned “steady deviation” occurs due to a change over time of an engine that is a control target, individual differences, or the like. Further, it is brought about by changes in the environment that cannot be corrected from various input signals. For this reason, when the operation suspension period of the vehicle becomes longer, the amount of “steady deviation” changes. In other words, the expression “steady” is used in comparison with an instantaneously changing amount such as a feedback amount, but this “steady deviation” also changes in a longer period. For example, when the operation is stopped for about half a year from summer to winter, the difference in outside air temperature increases, and the amount of steady deviation changes greatly.
[0015]
Therefore, the learning value stored for correcting this “steady deviation” may not be valid, and problems may occur under the assumption that there is an upper limit and a lower limit in the feedback amount as described above. It was. This will be described with reference to FIG. In FIG. 4, transitions of the control amount, the feedback amount, and the learning value are indicated by a solid line and a two-dot chain line. Here, the transition indicated by the solid line is referred to.
[0016]
As shown in FIG. 4, consider a case where the ignition switch of the vehicle is turned off at time t1 in a state where the control amount has converged to the target value. At this time, the amount of deviation between the base control amount and the target value (the amount indicated by the symbol C in the figure), that is, the amount of “steady deviation” is learned, and this is stored as a learned value. In this example, the learning value is C. Thereafter, the ignition switch is turned on again at time t2. However, if the period “t1, t2” becomes longer, for example, half a year, the amount of deviation between the base control amount and the target value changes greatly.In FIG. 4, only ΔC changes, and the amount of deviation becomes C ′. ing.
[0017]
At this time, since the learning value is stored as C, at time t2, correction by the learning value is performed, and the control amount greatly exceeds the target value. Therefore, in order to converge the control amount to the target value, the feedback amount is calculated so as to decrease at a constant rate. At this time, as described above, at time t3, the feedback amount is skipped in a direction approaching “0”, and the amount is reflected in the learning value.
[0018]
However, since the feedback amount is further decreased from time t3, the feedback amount reaches the lower limit value in the period [t4, t5]. That is, a feedback amount saturates. Thus, the control amount is not corrected to the target value side during the period [t4, t5]. As a result, the convergence of the control amount is delayed.
[0019]
As a technique for solving this, it is conceivable to reset the upper limit value and the lower limit value of the feedback amount. That is, the feedback amount takes a value in a larger range.
As described above, for example, a predetermined storage area on the RAM is assigned to the feedback amount, and the range that the feedback amount can take is determined by this storage area.
[0020]
Therefore, in order to increase the range of values that the feedback amount can take, it is conceivable to increase the storage area allocated to the feedback amount. However, considering that the storage area is finite, it is not desirable to increase the storage area. It is also conceivable to change the LSB while leaving the storage area as it is. However, in this case, the change in the feedback amount becomes rough, which is not desirable from the viewpoint of performing smooth control.
[0021]
The present invention has been made in order to solve the above-described problem, and the delay in convergence of the control amount caused by the vehicle operation suspension period being long and the learned value being invalid is expressed as the upper limit value of the feedback amount and The purpose is to prevent the lower limit without being manipulated.
[0022]
[Means for Solving the Problems and Effects of the Invention]
The vehicle control device according to claim 1, which is made to achieve the above-described object, performs feedback control based on the calculated control amount and the target value in order to converge the control amount used for vehicle control to the target value. A feedback amount to be calculated is calculated, a steady deviation of the control amount from the target value is learned as a learning value, and at least the feedback amount and the control amount are corrected using the learning value. The reason why “at least” is used here is that correction based on various input signals from a water temperature sensor, an atmospheric pressure sensor or the like can be considered as described above.
[0023]
Here, in particular, in the present invention, the vehicle operation end point is stored in the storage means. It is conceivable that the driving end time is, for example, the time when the ignition switch of the vehicle is turned off. Therefore, the storage means here is realized by a non-volatile memory device such as an EEPROM or a flash ROM, or a memory device that is constantly supplied with power, such as a standby RAM.
[0024]
The elapsed time calculation means calculates the elapsed time from the driving end time stored in the storage means at the start of driving of the vehicle. It is conceivable that the driving start time is, for example, the time when the ignition switch of the vehicle is turned on.
Then, based on the elapsed time calculated by the elapsed time calculating means, the weight setting means sets the feedback amount weight to be larger than the normal time and sets the learning value weight to be smaller than the normal time.
[0025]
Normally, it is assumed that each of the feedback amount and the learning value is reflected with a predetermined weight in the correction of the control amount. For example, in the example described above as the prior art, the feedback amount and the learning value are used as they are for the correction of the control amount, and thus the feedback amount and the learning value are each reflected in the correction of the control amount with a weight of “1”. Therefore, in this example, the weight setting means sets the weight of the feedback amount to, for example, “1.5” larger than “1”, and sets the weight of the learning value to, for example, “0.7” smaller than “1”. It is a condition to do.
[0026]
When the feedback amount and the learning value weight are set, the feedback amount and the learning value weight may be individually set as described above. For example, as shown in claim 2, a weight operation constant based on the elapsed time is calculated. The weights of both the feedback amount and the learning value may be set based on the calculated weight operation constant.
[0027]
In this case, a weight operation constant is calculated, and weights for both the feedback amount and the learning value are set. For example, if the weight operation constant is calculated as “0.4”, the weight operation constant is added to the weight of the feedback amount and subtracted from the weight of the learning value. At this time, the weight of the feedback amount is calculated as “1.4”, and the weight of the learning value is calculated as “0.6”. This is advantageous in that the weight setting process is simplified.
[0028]
The technical idea of the present invention is as follows.
That is, the control amount is corrected based on at least the feedback amount and the learning value. However, if the elapsed time (operation stoppage period) becomes as long as half a year, there is a high possibility that the learning value becomes invalid. On the other hand, the feedback amount is based on the calculated control amount and the target value. If this feedback amount is used, the control amount can always be corrected to the target value side.
[0029]
Therefore, in the present invention, the weight of the learning value is made smaller than that at the normal time to prevent the control amount from being largely corrected by the inappropriate learning value, and the weight of the feedback amount is made larger than that at the normal time, Make up for the amount of correction.
As described above, the feedback amount has an upper limit value / lower limit value. In view of this, the correction of leaning from the learning value to the feedback amount was realized by changing the weighting of the both without changing the feedback amount and the learning value directly, that is, by changing the degree of reflection of both to the correction amount.
[0030]
By changing the weights of the two, the total period of time during which the controlled variable transitions to the vicinity of the target value can be shortened as compared with the conventional configuration in which correction is performed using an invalid learning value.
When the control amount deviates significantly from the target value, the feedback amount becomes the upper limit value or the lower limit value (see the period [t4, t5] in FIG. 4), and as a result, the convergence of the control amount may be delayed. It was. In the present invention, as described above, the possibility that the feedback amount becomes the upper limit value or the lower limit value is reduced by shortening the period during which the control amount deviates greatly from the target value as much as possible. For example, the period during which the control amount is not corrected, such as the period [t4, t5] shown in FIG. 4, can be reduced, and the convergence of the control amount to the target value can be prevented from being delayed.
[0031]
The reason why the weight of the feedback amount is set to be large as well as the weight of the learning value is set small is as follows.
If the weight of the learning value is set to be small, correction due to an invalid learning value is suppressed, so that the control amount can be prevented from greatly deviating from the target value due to correction due to an invalid learning value. However, since the correction by the learning value is suppressed, there is a possibility that the control amount deviates from the target value. For example, FIG. 4 shows the case where the amount of deviation between the base control amount and the target value becomes smaller in the period [t1, t2]. Conversely, the amount of deviation becomes larger in the period [t1, t2]. That is, the base control amount may be shifted downward. In this case, even if the control amount is corrected by the learning value C, the control amount is less than the target value. However, if the weight of the learning value C is reduced, the control amount further deviates from the target value. Become. Therefore, the weight of the feedback amount is set to be larger than that at the normal time, and the correction based on the feedback amount is performed because the correction based on the learning value is not performed.
[0032]
By the way, it is conceivable to set the weight based on the elapsed time, for example, with the configuration shown in claim 3. That is, it further includes a correspondence storage unit that stores a correspondence between the elapsed time and the weight setting information for setting the weight. In this case, the weight setting unit sets the feedback amount and the weight of the learning value with reference to the correspondence stored in the correspondence storage unit.
[0033]
Here, the weight setting information for setting the weight may be the weight itself. Moreover, it is good also as a numerical value for changing weight in normal time like the above-mentioned weight operation constant. If the feedback amount and the weight of the learning value are individually set, the correspondence relationship may be stored for each of the feedback amount and the learning value. Further, as described above, if the weights of both the feedback amount and the learning value are set from the weight operation constant, only one correspondence relationship for associating the elapsed time with the weight operation constant may be stored. This correspondence is as illustrated in FIG.
[0034]
As described above, when the elapsed time is a long period such as half a year, there is a high possibility that the learning value becomes invalid. Therefore, as shown in FIG. 3, when the elapsed time exceeds the predetermined time S, it is conceivable to change and set the feedback amount and the learning value weight. That is, as shown in claim 4, it is conceivable that the weight setting means sets the feedback amount and the weight of the learning value when the elapsed time exceeds a predetermined time.
[0035]
The weight operation constant shown in FIG. 3 increases the weight of the feedback amount and decreases the weight of the learning value. When the predetermined time S is exceeded, the weight operation constant increases at a constant rate, and the maximum value is “1”. It has become. However, the present invention is not limited to this, and the map from the elapsed time to the weight operation constant shown in FIG. 3 may be a monotonically increasing function in a broad sense. That is, it is only necessary that the weighting operation constant f (a) ≦ f (b) with respect to the elapsed time a <b (provided that f (a) = f (b) with respect to any a <b. except.). For example, it is conceivable to increase the weighting operation constant in a step shape according to the elapsed time.
[0036]
By the way, the present invention realizes correction in which the learning value is tilted from the learning value to the feedback amount according to the elapsed time (operation stop period). For this reason, the weight of the feedback amount is increased while the weight of the learning value is increased. It was small. However, if the weight of the feedback amount increases, the convergence of the subsequent control amount deteriorates. This is because the controlled variable vibrates around the target value. On the other hand, since the learning value is updated and approaches a valid value, it is not necessary to keep the weight of the learning value small.
[0037]
Therefore, as shown in claim 5, when the weight setting means sets the feedback amount and the learning value weight, the weight setting means returns the set weight to the normal weight after an appropriate period thereafter. Good. For example, the feedback amount and the weight of the learning value are changed and set for a predetermined time from the time when the ignition switch is turned on, and after the predetermined time has elapsed, the weight is returned to the normal time. At this time, the weight of the feedback amount may be gradually decreased, while the weight of the learning value may be gradually increased and finally returned to the normal weight. Further, the appropriate period is not limited to a certain period. For example, it may be determined according to the degree of weight change. When the weight is changed more greatly, the weight is returned to the normal weight after a longer period. With the configuration as described above, the convergence of the control amount does not deteriorate.
[0038]
In addition, the function which implement | achieves the elapsed time calculation means and weight setting means of such a vehicle control apparatus with a computer system can be provided as a program started on the computer system side, for example. In the case of such a program, for example, the program is recorded on a computer-readable recording medium such as a floppy disk, a magneto-optical disk, a CD-ROM, or a hard disk, and is used by being loaded into a computer system and started up as necessary. it can. In addition, the ROM or backup RAM may be recorded as a computer-readable recording medium, and the ROM or backup RAM may be incorporated into a computer system.
[0039]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, an embodiment of the present invention will be described with reference to the drawings.
FIG. 1 is a block diagram showing a configuration of an engine control apparatus (hereinafter referred to as “ECU”) 1 of an embodiment mounted on a vehicle.
[0040]
The ECU 1 includes a power supply circuit 21, an input circuit 22, an A / D converter 23, and an output circuit 24 with the microcomputer 10 as the center.
The microcomputer 10 includes a CPU 11 that executes various processes for controlling an engine mounted on the vehicle, an input port 12 that inputs a signal from the input circuit 22, and an output port that outputs a signal to the output circuit 24. 13, a ROM 14 for storing a program executed by the CPU 11, a RAM 15 for temporarily storing a control calculation result by the CPU 11, and a serial communication I / O for exchanging data with the A / D converter 23 described above. An F (interface) 16 and a standby RAM 17 are provided. The CPU 11, input port 12, output port 13, ROM 14, RAM 15, serial communication I / F 16 and standby RAM 17 are connected to each other via a bus 18.
[0041]
The ECU 1 includes a rotation angle sensor 33 that detects the engine speed NE, a speed sensor 34 that detects the traveling speed (vehicle speed) of the vehicle, a throttle sensor 35 that detects the throttle opening, an atmospheric pressure sensor 36 that measures atmospheric pressure, Various sensors such as a water temperature sensor 37 that detects the temperature of engine coolant, an injector 38 that performs fuel injection, and an actuator such as an ISC valve 39 that adjusts the engine speed are connected.
[0042]
Signals from the rotation angle sensor 33 and the speed sensor 34 are input to the CPU 11 from the input port 12 of the microcomputer 10 via the input circuit 22 described above. On the other hand, signals from the throttle sensor 35, the atmospheric pressure sensor 36, and the water temperature sensor 37 are converted into digital signals by the A / D converter 23 described above, and input to the CPU 11 via the serial communication I / F 16 of the microcomputer 10. The Further, the output circuit 24 described above operates the injector 38 and the ISC valve 39 in accordance with a drive signal output from the CPU 11 of the microcomputer 10 via the output port 13.
[0043]
The power supply circuit 21 receives an ignition voltage supplied from the vehicle battery 32 via the ignition switch 31 and operates to the CPU 11, input port 12, output port 13, ROM 14, RAM 15, and serial communication I / F 16 of the microcomputer 10. Output voltage. The standby RAM 17 is always supplied with power from the battery 32, and the memory contents are not lost even when the ignition switch 31 is turned off.
[0044]
In such an ECU 1, when the ignition switch 31 is turned on (turned on), an operating voltage is output from the power supply circuit 21 to the microcomputer 10. Then, the CPU 11 executes engine control processing based on a program stored in the ROM 14, and input signals from various sensors such as the rotation angle sensor 33, the speed sensor 34, the throttle sensor 35, the atmospheric pressure sensor 36, and the water temperature sensor 37. Based on the above, the engine is controlled by operating actuators such as the injector 38 and the ISC valve 39.
[0045]
In this engine control process, for example, the engine speed, air-fuel ratio, and the like are calculated as control amounts. At this time, in order to make the calculated control amount coincide with the set target value, various corrections are performed when calculating the control amount. For example, correction based on input signals from various sensors such as the atmospheric pressure sensor 36 and the water temperature sensor 37 is performed. Further, in order to correct the control amount in accordance with the environment that changes every moment, the feedback amount is calculated based on the calculated control amount and the target value. Furthermore, “steady deviation” from the target value of the control amount is learned as a learning value. In general, the correction based on the feedback amount and the learning value is performed together with the correction based on the sensor signal.
[0046]
The CPU 11 stores a learning value representing “steady deviation” from the target value of the control amount in the standby RAM 17 so that it is not lost even when the ignition switch 31 is off. Thus, when the ignition switch 31 is turned on again, the learning value stored in the standby RAM 17 is read, and the control amount is corrected using this learning value.
[0047]
Next, a control amount calculation process executed by the CPU 11 of the microcomputer 10 will be described. The CPU 11 performs various controls on the engine. Here, the calculation of a general control amount will be described as an example.
FIG. 2 is a flowchart showing a control amount calculation process executed by the CPU 11. This control amount calculation process is executed when the ignition switch 31 is turned on.
[0048]
First, in the first step S100, the time during which the operation is stopped is calculated as the elapsed time. As will be described later, the time at which the process is executed (current time) is repeatedly stored in the standby RAM 17. Therefore, when the ignition switch 31 is turned off, the time immediately before that is stored. Therefore, here, the time stored in the standby RAM 17 is read, and the difference from the current time is calculated as the elapsed time. In this sense, the standby RAM 17 corresponds to “storage means”.
[0049]
In subsequent S110, a weighting operation constant W is determined. The weight operation constant W is determined based on the map shown in FIG. This map shows the correspondence between the elapsed time and the weight operation constant, and is stored in the ROM 14 in advance. Therefore, the ROM 14 corresponds to “correspondence storage means”. As can be seen from FIG. 3, when the elapsed time exceeds the predetermined time S, the weight operation constant is set larger than “0”, and in this case, the weight operation constant increases at a constant rate as the elapsed time becomes longer. . Note that the maximum value of the weight operation constant is “1”. For example, the predetermined time S may be half a year.
[0050]
In the next S120, each correction coefficient is calculated. Here, a correction coefficient is calculated based on signals from various sensors. In FIG. 2, the temperature correction coefficient TmpCmp based on the signal from the water temperature sensor 37 and the atmospheric pressure correction coefficient AtmCmp based on the signal from the atmospheric pressure sensor 36 are shown. Here, the feedback amount FBCmp is calculated based on the control amount and the target value. In the present embodiment, when the control amount exceeds the target value, the feedback amount FBCmp is decreased at a constant rate, and conversely, when the control amount is below the target value, the feedback amount FBCmp is increased at a constant rate. Note that the feedback amount may be calculated based on the deviation between the control amount and the target value.
[0051]
In subsequent S130, the total correction amount is calculated. The total correction amount TotalCmp is obtained from the temperature correction coefficient TmpCmp, the atmospheric pressure correction coefficient AtmCmp, the feedback amount FBCmp, the learning value Adpt, and the like. At this time, the feedback amount FBCmp and the learning value Adpt are weighted. That is, using the weighting operation constant W calculated in S110, the weight of the feedback amount is set to (1 + W) and larger than the normal weight “1”, and the learning value weight is set to (1−W). It is set smaller than the weight “1”. Then, the feedback amount FBCmp is multiplied by the weight (1 + W) to calculate the correction amount based on the feedback amount, and the learning value Adpt is multiplied by the weight (1-W) to calculate the correction amount based on the learning value.
[0052]
In the next S140, the control amount is determined and output using the calculated total correction amount TotalCmp. In subsequent S150, the learning value Adpt is updated. It is conceivable to update this based on the deviation of the feedback amount from the reference value.
In the next S160, the weight operation constant W is updated. Here, the weight operation constant W is updated by subtracting the predetermined value DEC from the weight operation constant W. At this time, when the weight operation constant W becomes smaller than “0”, the weight operation constant W is set to “0”. In the next S170, the current time is stored in the standby RAM 17, and then the process proceeds to S120 and the processes from S120 are repeated.
[0053]
In the present embodiment, the weight operation constant W corresponding to the elapsed time is determined by the control amount calculation process described above (S110 in FIG. 2). Specifically, as the elapsed time exceeds the predetermined time S, a larger maximum weight operation constant W of “1” is determined (see FIG. 3). Then, the correction amount based on the feedback amount is calculated as FBCmp × (1 + W), while the correction amount based on the learning value is calculated as Adpt × (1−W) (S130). Thereafter, the control amount is determined and output using these (S140).
[0054]
Therefore, since the correction by the learning value Adpt that is not appropriate due to the increase in the elapsed time is suppressed and the correction by the feedback amount is increased, the conventional configuration in which correction is performed using the learning value that is not appropriate when viewed in total. In comparison, there is a higher possibility that the period during which the control amount changes to the target value can be shortened.
[0055]
If the period during which the control amount deviates greatly from the target value becomes longer, the feedback amount becomes the upper limit value or the lower limit value (see periods [t4, t5] in FIG. 4), and as a result, the convergence of the control amount may be delayed. there were. In the present embodiment, as described above, the possibility that the feedback amount becomes the upper limit value or the lower limit value is reduced by shortening the period during which the control amount greatly deviates from the target value as much as possible. For example, the period during which the control amount is not corrected, such as the period [t4, t5] shown in FIG. 4, can be reduced, and the convergence of the control amount to the target value can be prevented from being delayed.
[0056]
The transition of the control amount during the period [t2, t4] shown in FIG. 4 is shown in an enlarged manner in FIG. Next, the effect of this embodiment will be described more specifically based on FIG.
Since the feedback amount is “0” at time t2 (see FIG. 4), the control amount is corrected by the learning value Adpt × (1−W), and the control amount is extremely different from the target value by the correction by the inappropriate learning value Adpt. Can be prevented. Further, for example, at time t3, in addition to the correction by the learning value Adpt × (1−W), the correction is further performed by the feedback amount FBCmp × (1 + W), so that the control amount is largely corrected to the target value side. Therefore, the period during which the control amount deviates greatly from the target value is shortened, and the feedback amount does not become the upper limit value or the lower limit value as shown in FIG. Therefore, it is possible to prevent the convergence of the control amount from being delayed.
[0057]
Note that FIG. 4 shows the case where the amount of deviation between the base control amount and the target value is reduced in the period [t1, t2], but conversely, the amount of deviation is increased in the period [t1, t2]. That is, the base control amount may be shifted downward. In this case, even if the control amount is corrected by the learning value Adpt, the control amount is less than the target value. However, if the weight of the learning value Adpt is reduced to (1-W), the control amount is further reduced. It will deviate from the target value.
[0058]
However, since the weight of the feedback amount FBCmp is set to be larger (1 + W) than usual, the slope (absolute value) of the control amount toward the target value becomes larger than the conventional value, and as a result, the time period deviates greatly from the target value. Is likely to be shortened. Therefore, even in such a case, there is a high possibility that the convergence of the control amount can be prevented from being delayed.
[0059]
If the feedback amount weight (1 + W) is maintained as it is, the control amount vibrates around the target value, and the convergence of the subsequent control amount deteriorates.
Therefore, in the control amount calculation process described above, when the process from S120 is repeated, the weight operation constant W once determined is decreased by a predetermined value DEC and finally returned to “0” (in FIG. 2). S160). As a result, the convergence of the control amount is not deteriorated.
[0060]
The CPU 11 of the microcomputer 10 that executes the control amount calculation process described above corresponds to “elapsed time calculation means” and “weight setting means”. The process of S100 in FIG. 2 corresponds to the process as the elapsed time calculation unit, and the processes of S110, S130, and S160 correspond to the process as the weight setting unit.
[0061]
As described above, the present invention is not limited to such embodiments, and can be implemented in various forms without departing from the spirit of the present invention.
(A) In the above embodiment, the time when the ignition switch 31 of the vehicle is turned off (specifically, the time immediately before the vehicle is turned off) is stored in the standby RAM 17 as the time of “end of driving”. Instead of the standby RAM 17, an EEPROM or a flash ROM may be used.
[0062]
(B) In the above embodiment, when the elapsed time (operation stop period) exceeds a predetermined time, the weighting operation constant is increased at a constant rate (see FIG. 3). You may set according to. For example, the weight setting constant may be increased uniformly when the elapsed time exceeds a predetermined time. For example, the weight setting constant may be increased exponentially.
[Brief description of the drawings]
FIG. 1 is a block diagram illustrating a configuration of an engine control apparatus according to an embodiment.
FIG. 2 is a flowchart illustrating control amount calculation processing according to the embodiment.
FIG. 3 is an explanatory diagram showing a correspondence between elapsed time and weighting operation constant in the embodiment.
FIG. 4 is a timing chart exemplified by associating a control amount, a feedback amount, and a learning value.
FIG. 5 is a partial explanatory diagram of the timing chart of FIG. 4;
FIG. 6 is a timing chart illustrating the relationship among a control amount, a feedback amount, and a learned value in a conventional configuration.
FIG. 7 is an explanatory diagram showing an effect when a learning value is stored in a conventional configuration.
[Explanation of symbols]
1 ... Engine control device
10 ... Microcomputer 11 ... CPU
12 ... Input port 13 ... Output port
14 ... ROM 15 ... RAM
16 ... Serial communication I / F 17 ... Standby RAM
21 ... Power supply circuit 22 ... Input circuit
23 ... A / D converter 24 ... Output circuit
31 ... Ignition switch 32 ... Battery
33 ... Rotation angle sensor 34 ... Speed sensor
35 ... Throttle sensor 36 ... Atmospheric pressure sensor
37 ... Water temperature sensor 38 ... Injector
39 ... ISC valve

Claims (6)

In order to converge a control amount used for vehicle control to a target value, a feedback amount for performing feedback control is calculated based on the calculated control amount and the target value, and the control amount from the target value is constantly In a vehicle control device that learns a gap as a learning value and corrects the control amount using at least the feedback amount and the learning value,
Storage means for storing the end of driving of the vehicle;
An elapsed time calculating means for calculating an elapsed time from the driving end time stored in the storage means at a driving start time of the vehicle;
Based on the elapsed time calculated by the elapsed time calculation means, assuming that each of the feedback amount and the learned value is reflected with a predetermined weight in the correction of the control amount in normal times, the feedback amount A vehicle control device comprising weight setting means for setting a weight larger than normal and setting a weight of a learning value smaller than normal. The vehicle control device according to claim 1,
The weight control unit calculates a weight operation constant based on the elapsed time, and sets both the feedback amount and the learning value based on the calculated weight operation constant. In the vehicle control device according to claim 1 or 2,
Furthermore, correspondence relationship storage means for storing a correspondence relationship between the elapsed time and the weight setting information for setting the weight,
The vehicle control apparatus characterized in that the weight setting means sets the weight of the feedback amount and the learning value with reference to the correspondence stored in the correspondence storage. In the vehicle control device according to any one of claims 1 to 3,
The weight control means sets the feedback amount and the weight of the learning value when the elapsed time exceeds a predetermined time. In the vehicle control device according to any one of claims 1 to 4,
When the weight setting means sets the feedback amount and the weight of the learning value, the weight control means returns the set weight to the normal weight after an appropriate period thereafter. A computer-readable recording medium in which a program for causing a computer system to function as the elapsed time calculating means and the weight setting means of the vehicle control device according to claim 1 is recorded.

Images