JP6985676B2 - Compression self-ignition engine controller - Google Patents

Description

The technique disclosed herein relates to a control device for a compression self-igniting engine.

Patent Document 1 describes a device that simulates the dynamic performance of a power train of an automobile by a model. This simulation device expresses a functional model that expresses the individual units that make up the target device by system equations, and by updating and discretizing the parameters by the mechanism model for each sampling clock, the non-linearity of the characteristics by the mechanism is exhibited. Linearize at each sampling time.

Japanese Patent No. 3987199

By the way, a compression self-ignition engine that burns a dilute air-fuel mixture in a combustion chamber by self-ignition is known for the purpose of improving fuel efficiency and exhaust gas performance. Since this combustion largely depends on the self-ignition phenomenon due to the chemical reaction between oxygen and fuel, it is highly sensitive to state quantities such as in-cylinder temperature and concentration. When the load of the engine or the temperature in the combustion chamber is relatively high, noise is likely to be generated due to steep combustion. On the contrary, when the load of the engine or the temperature in the combustion chamber is low, the fuel efficiency is deteriorated due to the decrease in the combustion efficiency. In the worst case, if the load of the engine or the temperature in the combustion chamber is too low, the combustion fluctuation becomes too large, leading to misfire. Compressed self-ignition combustion is characterized by a narrow range of state quantities that enable stable combustion. Therefore, in the compression self-ignition type engine, in order to realize stable combustion with good fuel economy, it is necessary to accurately estimate and control the state quantity in the combustion chamber.

In the conventional engine control, a method of determining the operation amount of various devices with respect to the target operating condition has been adopted by using the control map obtained by the steady test performed in advance and the correction map. In a compression self-ignition engine, a method of controlling the state quantity in the combustion chamber by an internal EGR (Exhaust Gas Recirculation) is promising. Even if the amount of operation is the same during steady operation and transient operation, the composition and ratio of internal EGR may differ. Conventional map control using steady-state maps and correction maps has limitations in dealing with various driving conditions on the road. Further, when applying the conventional map control to the control of the compression self-ignition type engine, it is necessary to carry out a preliminary experiment including the transient operation, but there is also a problem that the conditions of the transient operation become enormous.

The technique disclosed herein has been made in view of this point, and its purpose is to provide a new control system for a compression self-igniting engine.

Specifically, the technique disclosed herein relates to a control device for a compression self-igniting engine. This control device includes an engine configured to self-ignite the air-fuel mixture in the combustion chamber every cycle, and fresh air and burned gas attached to the engine and into the combustion chamber. By coordinating the introduction, a state quantity setting device configured to bring the inside of the combustion chamber into a desired state and to inject fuel attached to the engine and supplied into the combustion chamber. The injector, the sensor attached to the engine and configured to detect parameters related to the pressure in the combustion chamber and output the detection signal, and the state amount setting device, the injector, and the sensor. A controller configured to operate the engine by connecting each of them and outputting an operation signal to the state amount setting device and the injector, the operation signal of the controller, and the detection signal of the sensor. It is provided with an estimator configured to receive and estimate the amount of state in the combustion chamber and output a state estimation signal to the controller.

Then, the controller receives the detection signal of the sensor and the state estimation signal of the estimator, and the combustion in the combustion chamber is the combustion corresponding to the target output of the engine. By setting the operation amount of the state amount setting device and the operation amount of the injector, respectively, and using the linear parameter fluctuation model that models the state in the combustion chamber shown in the equation (1), the estimator is used. The state quantity in the combustion chamber is estimated _{, and the first coefficient A LPV} and the second coefficient B _LPV in the linear parameter fluctuation model change according to the torque of the engine and the rotation speed of the engine, respectively.

However, x _k is one or both of the amount of oxygen in the combustion chamber and the amount of fuel in the combustion chamber as the state amount at a predetermined timing in the current cycle, and in the combustion chamber. The temperature of. x _{k + 1} is one or both of the amount of oxygen in the combustion chamber and the amount of fuel in the combustion chamber, and the temperature in the combustion chamber as the state quantity at a predetermined timing in the next cycle. be. u _k is as the operation amount in the current cycle, a EGR rate and fuel injection amount.

Here, the "combustion chamber" is not limited to the meaning of the space formed when the piston in the cylinder reaches the compression top dead center. The term "combustion chamber" may be used in a broad sense. That is, "inside the combustion chamber" may be paraphrased as "inside the cylinder".

According to the above configuration, the controller makes the combustion in the combustion chamber compressed self-ignition combustion corresponding to the target output of the engine based on the detection signal of the sensor and the state estimation signal of the estimator. Status amount setting Set the operation amount of the device and the operation amount of the injector. The controller outputs an operation signal to the state quantity setting device and the injector. Upon receiving the operation signal, the state quantity setting device adjusts the introduction of fresh air and burned gas into the combustion chamber to bring the inside of the combustion chamber into a desired state. Upon receiving the operation signal, the injector injects fuel to be supplied into the combustion chamber. In the combustion chamber, the air-fuel mixture is burned by compression self-ignition, and the engine is operated to the desired output.

The estimator estimates the state quantity (more accurately, a plurality of state quantities) in the combustion chamber by using the linear parameter fluctuation model shown in the equation (1). The linear parameter fluctuation model may be expressed by constructing a discretized model that expresses the state in the combustion chamber by a plurality of discrete points, and by linearizing the discretized model with respect to a plurality of equilibrium points. The multiple discrete points are, for example, the exhaust valve restart valve timing, the intake valve closing timing, the predetermined crank angle timing before the compression top dead center, the ignition timing, the maximum pressure timing, the combustion end timing, and the exhaust valve opening timing. It may be a part or all of the above.

By estimating the state in the combustion chamber using the linear parameter fluctuation model of the equation (1), the computational load of the estimator becomes smaller than that using the nonlinear model. On the other hand, since the first coefficient A _LPV and the second coefficient B _LPV of the equation (1) change according to the torque of the engine and the rotation speed of the engine, the estimation accuracy is higher than that of using a single linear model. Therefore, the calculation load of the estimator can be suppressed and the estimation accuracy can be improved at the same time. The torque of the engine may be expressed by, for example, the indicated Mean Effective Pressure (IMEP).

As a result, the controller can operate the state quantity setting device and the injector based on the highly accurate state estimation of the combustion chamber. The compression self-ignition engine realizes stable combustion with good fuel economy.

The controls disclosed herein also include an engine configured to self-ignite the air-fuel mixture in the combustion chamber and fresh air and fresh air into the combustion chamber attached to the engine and into the combustion chamber on a cycle-by-cycle basis. By adjusting the introduction of the burnt gas, a state amount setting device configured to bring the inside of the combustion chamber into a desired state and a fuel attached to the engine and supplied into the combustion chamber are injected. The injector configured to detect the parameters related to the pressure in the combustion chamber, the sensor attached to the engine, and the sensor configured to output the detection signal, the state amount setting device, the injector, and the like. And the controller configured to operate the engine by connecting each of the sensors and outputting an operation signal to the state amount setting device and the injector, and the operation signal of the controller and the sensor. It includes an estimator configured to receive the detection signal, estimate the amount of state in the combustion chamber, and output the state estimation signal to the controller.
Then, the controller receives the detection signal of the sensor and the state estimation signal of the estimator, and the combustion in the combustion chamber is the combustion corresponding to the target output of the engine. The operation amount of the state amount setting device and the operation amount of the injector are set respectively, and the estimator uses the linear parameter fluctuations modeled on the states in the combustion chamber shown in the equations (1) and (2). By using the model, the state quantity in the combustion chamber is estimated, and the first coefficient A _LPV , the second coefficient B _LPV , and the third coefficient C _LPV in the linear parameter fluctuation model are the torques of the engine, respectively. It changes according to.

However, y _k is the indicated mean effective pressure and the maximum pressure increase rate (dP / dθ) as the output of the engine in the current cycle.

By using a linear parameter fluctuation model including equations (1) and (2), it is possible to suppress the computational load of the estimator and improve the estimation accuracy, and the compression self-ignition engine realizes stable combustion with good fuel economy. ..

The estimator is configured by a Kalman filter represented by the equation (3) using the linear parameter fluctuation model, and the first coefficient A _LPV , the second coefficient B _LPV , and the third coefficient C _LPV are each cycle-by-cycle. In addition, it may be set according to the target torque of the engine.

However, g _LPV is Kalman gain.

The Kalman filter (Equation (3)) can estimate the state quantity (state variable) in the combustion chamber of the engine in real time from the input / output values under the condition where noise is present. _{By setting the first coefficient A LPV} , the second coefficient B _LPV , and the third coefficient C _LPV of the equations (1), (2), and (3) for each cycle according to the target torque of the engine. It is possible to improve the estimation accuracy of the state quantity in the combustion chamber while reducing the calculation load of the estimator. The target torque of the engine may be, for example, a target IMEP.

On the other hand, the estimator may be configured by the Kalman filter represented by the equation (4) using the linear parameter variation model.

However, f is a discretized model that models the state in the combustion chamber, and the linear parameter fluctuation model is obtained from the discretized model. The g _LPV is a Kalman gain calculated from the first coefficient A _LPV , the second coefficient B _LPV , and the third coefficient C _{LPV , and the g LPV} corresponds to the target torque of the engine for each cycle. Is set. y _k is the actual output of the engine obtained based on the detection signal of the sensor.

By using the discretized model for the pre-estimation of the Kalman filter, it is possible to improve the estimation accuracy while keeping the computational load of the estimator relatively small. _{By calculating the Kalman gain g LPV} of the post-estimation of the Kalman filter from _{the first coefficient A LPV} , the second coefficient B _LPV , and the third coefficient C _LPV of the linear parameter fluctuation model, the estimation accuracy is maintained high. The computational load can be significantly reduced compared to the case where the Kalman gain is obtained from the differentiation of the discretized model. With this configuration, it is possible to reduce the computational load of the estimator and increase the estimation accuracy at a high level.

The control device for the compression self-igniting engine comprises a supercharging system attached to the engine and configured to supercharge the gas to be introduced into the combustion chamber, the first coefficient A _LPV , the first coefficient. The two-coefficient B _LPV and the third-coefficient C _LPV may be changed according to the torque and the supercharging pressure of the engine, respectively.

By doing so, while avoiding an increase in the calculation load of the estimator, the amount of state in the combustion chamber in the engine with a supercharger, particularly the temperature in the combustion chamber and the temperature in the combustion chamber. The accuracy of estimating the amount of oxygen can be improved.

The first coefficient A _LPV , the second coefficient B _LPV , and the third coefficient C _LPV may be further changed according to the rotation speed of the engine, respectively.

By doing so, it is possible to improve the estimation accuracy of the state quantity in the combustion chamber over a wide range of the operating area of the engine while avoiding an increase in the calculation load of the estimator.

The first coefficient A _LPV , the second coefficient B _LPV , and the third coefficient C _LPV may be represented by a linear function expression or a quadratic function expression, respectively.

_{By making the first coefficient A LPV} , the second coefficient B _LPV , and the third coefficient C _LPV of the linear parameter fluctuation model a continuous function, the computational load of the estimator is reduced and the estimation accuracy is increased. It is compatible with what you do.

The controller has a feedback controller, which measures the deviation between the target output of the engine and the output of the engine based on the detection signal of the sensor, and the state estimation signal of the estimator. Upon receiving the feedback, the feedback operation amounts u _{FB and k} may be output, and the control rule of the feedback controller may be the equation (5).

However, x _{i and k} are cumulative terms of the deviation. K _LPV is the feedback gain.

In addition to the deviation between the target output of the engine and the actual output of the engine, the control performance of the controller can be improved by determining the feedback operation amount based on the state amount in the combustion chamber estimated by the estimator. can.

The feedback gain K _LPV is set based on the linear parameter fluctuation models of the equations (1) and (2), and the feedback gain K _LPV is set for each cycle according to the target torque of the engine. May be done.

According to the study by the inventors of the present application, the optimum value of the feedback gain for correcting the deviation between the target output of the engine and the actual output of the engine from the linear parameter fluctuation model of the equations (1) and (2), and the state quantity. The optimum value of the feedback gain for correcting the difference between the two can be obtained. The feedback gain changes according to the target torque of the engine. This improves the control performance of the controller.

The controller further comprises a feedforward controller, which is configured to receive the target output and the state estimation signal of the estimator and output a feedforward manipulated variable. , May be.

By combining the feedback controller and the feedforward controller, the control performance of the controller is improved.

As described above, according to the control device of the compression self-ignition engine, the state quantity in the combustion chamber can be estimated with a relatively small calculation load and with relatively high accuracy. The compression self-ignition engine realizes stable combustion with good fuel consumption by operating the state quantity setting device and the injector based on the highly accurate estimation of the state of the combustion chamber by the controller.

FIG. 1 is a diagram illustrating the configuration of a compression self-ignition engine. FIG. 2 is a block diagram illustrating a configuration of a control device for a compression self-ignition engine. FIG. 3 is a block diagram illustrating the configuration of the controller. FIG. 4 is a flowchart illustrating the procedure for determining the coefficient of the linear model. FIG. 5 is a diagram illustrating steady-state set points related to the construction of a model. FIG. 6 is a flowchart illustrating a procedure for estimating a state quantity in a combustion chamber. FIG. 7 is a diagram illustrating the characteristics of each element of the first coefficient. FIG. 8 is a diagram illustrating the characteristics of each element of the second coefficient. FIG. 9 is a diagram illustrating the characteristics of each element of the third coefficient. FIG. 10 is a diagram illustrating the characteristics of each element of the feedback gain. FIG. 11 is a diagram illustrating the discrete points of the discretized model and the valve timings of the intake valve and the exhaust valve.

Hereinafter, embodiments of the combustion control device of the compression self-ignition engine will be described in detail with reference to the drawings. The following description is an example of a control device for a compression self-igniting engine. FIG. 1 is a diagram illustrating the configuration of a compression self-ignition engine. FIG. 2 is a block diagram illustrating a configuration of a control device for a compression self-ignition engine.

The engine 1 is mounted on a four-wheeled vehicle. By driving the engine 1, the automobile runs. The fuel of the engine 1 is gasoline in this configuration example. The fuel may be gasoline containing bioethanol or the like. The fuel of the engine 1 may be any fuel as long as it is at least a liquid fuel containing gasoline.

(Engine configuration)
The engine 1 includes a cylinder block 12 and a cylinder head 13 mounted on the cylinder block 12. A plurality of cylinders 11 are formed inside the cylinder block 12. FIG. 1 shows only one cylinder 11. The engine 1 is a multi-cylinder engine.

A piston 3 is slidably inserted in each cylinder 11. The piston 3 is connected to the crankshaft 15 via a connecting rod 14. The piston 3 forms a combustion chamber 17 together with the cylinder 11 and the cylinder head 13. The "combustion chamber" is not limited to the meaning of the space formed when the piston 3 reaches the compression top dead center. The term "combustion chamber" may be used in a broad sense. That is, the "combustion chamber" may mean the space formed by the piston 3, the cylinder 11 and the cylinder head 13 regardless of the position of the piston 3. Although detailed illustration is omitted, a cavity is formed on the upper surface of the piston 3. Further, the lower surface of the cylinder head 13, that is, the ceiling surface of the combustion chamber 17, is composed of two inclined surfaces. The combustion chamber 17 has a so-called pent roof shape.

The geometric compression ratio of the engine 1 is set high for the purpose of improving the theoretical thermal efficiency and stabilizing the CI (Compression Ignition) combustion described later. Specifically, the geometric compression ratio of the engine 1 is 17 or more. The geometric compression ratio may be 18, for example. The geometric compression ratio may be appropriately set in the range of 17 or more and 20 or less.

The cylinder head 13 is formed with two intake ports 18 for each cylinder 11 (note that only one intake port 18 is shown in FIG. 1). The intake port 18 communicates with the combustion chamber 17. An intake valve 21 is provided at the intake port 18. The intake valve 21 opens and closes between the combustion chamber 17 and the intake port 18. The intake valve mechanism opens and closes the intake valve 21 at a predetermined timing. In this configuration example, the intake valve mechanism has an intake S-VT (Sequential-Valve Timing) 231 as a variable valve system, as shown in FIG. The intake S-VT231 is configured to continuously change the rotational phase of the intake camshaft within a predetermined angular range. The valve opening timing and valve closing timing of the intake valve 21 change continuously. The intake S-VT231 may be, for example, an electric type or a hydraulic type.

The intake valve mechanism also has an intake CVVL (Continuously Variable Valve Lift) 232 as a variable valve system in this configuration example, as shown in FIG. The intake CVVL232 is configured to continuously change the lift amount of the intake valve 21 within a predetermined range. As the configuration of the intake CVVL232, a known configuration may be appropriately adopted. The configuration of the intake CVVL232 is arbitrary.

The cylinder head 13 is also formed with two exhaust ports 19 for each cylinder 11 (note that only one exhaust port 19 is shown in FIG. 1). The exhaust port 19 communicates with the combustion chamber 17. An exhaust valve 22 is provided at the exhaust port 19. The exhaust valve 22 opens and closes between the combustion chamber 17 and the exhaust port 19. The exhaust valve mechanism opens and closes the exhaust valve 22 at a predetermined timing. In this configuration example, the exhaust valve mechanism has an exhaust S-VT241 as a variable valve system, as shown in FIG. The exhaust S-VT241 is configured to continuously change the rotational phase of the exhaust camshaft within a predetermined angular range. The valve opening timing and valve closing timing of the exhaust valve 22 change continuously. The exhaust S-VT241 may be an electric type or a hydraulic type.

The exhaust valve mechanism also has an IDEVA 242 as a variable valve system in this configuration example, as shown in FIG. Although not shown, the IDEVA 242 is configured to change the lift amount of the exhaust valve 22 by opening the exhaust valve 22 by either the first cam portion or the second cam portion. .. Although not shown in the cam profile of the first cam portion, the lift amount of the exhaust valve 22 gradually increases from zero to the maximum lift amount as the crank angle progresses, and then the lift amount gradually increases. It has one cam mountain that decreases to zero. On the other hand, the cam profile of the second cam portion has two cam ridges, a large lift portion and a small lift portion, as illustrated in FIG. The large lift portion is a portion where the lift amount of the exhaust valve gradually increases from zero to the maximum lift amount as the crank angle progresses, and then the lift amount gradually decreases. The large lift portion is a first cam ridge that opens the exhaust valve 22 in the exhaust stroke. The small lift portion (exhaust resumption valve) is a portion where the lift amount of the exhaust valve 22 is gradually increased to zero after the large lift portion is followed by increasing the lift amount of the exhaust valve 22 again. The small lift portion is a second cam ridge that opens the exhaust valve 22 in the intake stroke.

The engine 1 adjusts the length of the overlap period related to the valve opening timing of the intake valve 21 and the valve closing timing of the exhaust valve 22 by the intake valve mechanism and the exhaust valve mechanism. As a result, depending on the operating state of the engine 1, the burned gas in the combustion chamber 17 may be scavenged, or the hot burned gas may be trapped in the combustion chamber 17 (that is, the internal EGR gas may be trapped in the combustion chamber 17). Introduced in). Further, when the engine 1 is in a predetermined operating state, the exhaust valve 22 is opened in the intake stroke by the exhaust valve mechanism. This introduces the internal EGR gas into the combustion chamber 17. The intake valve mechanism and the exhaust valve mechanism are examples of state quantity setting devices.

An injector 6 is attached to the cylinder head 13 for each cylinder 11. The injector 6 is configured to inject fuel directly into the combustion chamber 17. The injector 6 is arranged so that its central axis is along the central axis of the cylinder 11. The injector 6 faces the cavity. The central axis of the injector 6 may be deviated from the central axis of the cylinder 11.

Although detailed illustration is omitted, the injector 6 is composed of a multi-injection type fuel injection valve having a plurality of injection ports. The injector 6 injects fuel so that the fuel spray spreads radially from the center of the combustion chamber 17.

The injector 6 is not limited to the multi-injection type injector. As the injector 6, an external valve opening type injector may be adopted.

A fuel supply system 61 is connected to the injector 6. The fuel supply system 61 includes a fuel tank 63 configured to store fuel, and a fuel supply path 62 that connects the fuel tank 63 and the injector 6 to each other. A fuel pump 65 and a common rail 64 are interposed in the fuel supply path 62. The fuel pump 65 pumps fuel to the common rail 64. In this configuration example, the fuel pump 65 is a plunger type pump driven by a crankshaft 15. The common rail 64 is configured to store the fuel pumped from the fuel pump 65 at a high fuel pressure. When the injector 6 opens, the fuel stored in the common rail 64 is injected into the combustion chamber 17 from the injection port of the injector 6. The fuel supply system 61 is configured to be able to supply fuel having a high pressure of 30 MPa or more to the injector 6. The maximum fuel pressure of the fuel supply system 61 may be, for example, about 120 MPa. The pressure of the fuel supplied to the injector 6 may be changed according to the operating state of the engine 1. The configuration of the fuel supply system 61 is not limited to the above configuration.

A spark plug 25 is attached to the cylinder head 13 for each cylinder 11. The spark plug 25 forcibly ignites the air-fuel mixture in the combustion chamber 17. In this configuration example, the spark plug 25 is arranged on the intake side of the cylinder 11 with the central axis interposed therebetween. The spark plug 25 is adjacent to the injector 6. The spark plug 25 is located between the two intake ports 18. The spark plug 25 is attached to the cylinder head 13 so as to be inclined from above to below toward the center of the combustion chamber 17. Although detailed illustration is omitted, the electrode of the spark plug 25 faces the inside of the combustion chamber 17 and is located near the ceiling surface of the combustion chamber 17.

An intake passage 40 is connected to one side surface of the engine 1. The intake passage 40 communicates with the intake port 18 of each cylinder 11. The intake passage 40 is a passage through which the gas introduced into the combustion chamber 17 flows. An air cleaner 41 for filtering fresh air is disposed at the upstream end of the intake passage 40. A surge tank 42 is arranged near the downstream end of the intake passage 40. The intake passage 40 downstream of the surge tank 42 constitutes an independent passage that branches for each cylinder 11, although detailed illustration is omitted. The downstream end of the independent passage is connected to the intake port 18 of each cylinder 11.

A throttle valve 43 is arranged between the air cleaner 41 and the surge tank 42 in the intake passage 40. The throttle valve 43 is configured to adjust the amount of fresh air introduced into the combustion chamber 17 by adjusting the opening degree of the valve. The throttle valve 43 is an example of a state quantity setting device.

The intake passage 40 is also provided with a supercharger 44 downstream of the throttle valve 43. The supercharger 44 is configured to supercharge the intake air introduced into the combustion chamber 17. In this configuration example, the turbocharger 44 is a mechanical turbocharger driven by the engine 1. The mechanical turbocharger 44 may be, for example, a roots type. The configuration of the mechanical turbocharger 44 may be any configuration. The mechanical turbocharger 44 may be a Rishorum type or a centrifugal type.

An electromagnetic clutch 45 is interposed between the supercharger 44 and the engine 1. The electromagnetic clutch 45 transmits a driving force from the engine 1 to the supercharger 44 or cuts off the transmission of the driving force between the supercharger 44 and the engine 1. As will be described later, the turbocharger 44 is switched on and off by switching the shutoff and connection of the electromagnetic clutch 45 by the ECU 10. That is, the engine 1 can switch between supercharging the gas introduced into the combustion chamber 17 by the supercharger 44 and not supercharging the gas introduced into the combustion chamber 17 by the supercharger 44. It is configured so that it can be done.

Instead of the mechanical turbocharger 44, or in addition to the mechanical turbocharger 44, a turbocharger operated by exhaust energy may be attached to the engine 1.

An intercooler 46 is arranged downstream of the supercharger 44 in the intake passage 40. The intercooler 46 is configured to cool the compressed gas in the turbocharger 44. The intercooler 46 may be configured to be water-cooled, for example.

A bypass passage 47 is connected to the intake passage 40. The bypass passage 47 connects the upstream portion of the supercharger 44 and the downstream portion of the intercooler 46 in the intake passage 40 to each other so as to bypass the supercharger 44 and the intercooler 46. An air bypass valve 48 is provided in the bypass passage 47. The air bypass valve 48 adjusts the flow rate of the gas flowing through the bypass passage 47.

When the supercharger 44 is turned off (that is, when the electromagnetic clutch 45 is disengaged), the air bypass valve 48 is fully opened. As a result, the gas flowing through the intake passage 40 bypasses the supercharger 44 and is introduced into the combustion chamber 17 of the engine 1. The engine 1 operates in a non-supercharged state, that is, in a naturally aspirated state.

When the turbocharger 44 is turned on (that is, when the electromagnetic clutch 45 is connected), a part of the gas that has passed through the turbocharger 44 flows back to the upstream of the turbocharger through the bypass passage 47. .. Since the reverse flow rate can be adjusted by adjusting the opening degree of the air bypass valve 48, the boost pressure of the gas introduced into the combustion chamber 17 can be adjusted. In this configuration example, the supercharging system 49 is configured by the supercharger 44, the electromagnetic clutch 45, the bypass passage 47, and the air bypass valve 48.

An exhaust passage 50 is connected to the other side surface of the engine 1. The exhaust passage 50 communicates with the exhaust port 19 of each cylinder 11. The exhaust passage 50 is a passage through which the exhaust gas discharged from the combustion chamber 17 flows. Although not shown in detail, the upstream portion of the exhaust passage 50 constitutes an independent passage that branches for each cylinder 11. The upstream end of the independent passage is connected to the exhaust port 19 of each cylinder 11. The exhaust passage 50 is provided with an exhaust gas purification system having one or more catalytic converters 51. The catalyst converter 51 includes a three-way catalyst. The exhaust gas purification system is not limited to the one including only the three-way catalyst.

An EGR passage 52 constituting an external EGR system is connected between the intake passage 40 and the exhaust passage 50. The EGR passage 52 is a passage for returning a part of the burned gas to the intake passage 40. The upstream end of the EGR passage 52 is connected to the upstream of the catalytic converter 51 in the exhaust passage 50. The downstream end of the EGR passage 52 is connected to the intake passage between the air cleaner 41 and the turbocharger 44 in the intake passage 40.

A water-cooled EGR cooler 53 is provided in the EGR passage 52. The EGR cooler 53 is configured to cool the burnt gas. The EGR passage 52 is also provided with an EGR valve 54. The EGR valve 54 is configured to regulate the flow rate of the burnt gas flowing through the EGR passage 52. By adjusting the opening degree of the EGR valve 54, the recirculation amount of the cooled burnt gas, that is, the external EGR gas can be adjusted.

In this configuration example, the EGR system 55 includes an external EGR system including an EGR passage 52 and an EGR valve 54, and an internal EGR system including the above-mentioned intake valve mechanism and exhaust valve mechanism. It is composed of and. The EGR system 55 is an example of a state quantity setting device.

As shown in FIG. 2, the control device of the compression self-ignition type engine includes an ECU (Engine Control Unit) 10 for operating the engine 1. The ECU 10 is a controller based on a well-known microcomputer, and is composed of a central processing unit (CPU) 101 for executing a program, and, for example, a RAM (Random Access Memory) or a ROM (Read Only Memory). It is provided with a memory 102 for storing programs and data, and an input / output bus 103 for inputting / outputting electric signals. The ECU 10 is an example of a controller.

As shown in FIGS. 1 and 2, various sensors SW1 to SW16 are connected to the ECU 10. The sensors SW1 to SW16 output a detection signal to the ECU 10. The sensors include the following sensors.

That is, the air flow sensor SW1 located downstream of the air cleaner 41 in the intake passage 40 and detecting the flow rate of fresh air flowing through the intake passage 40, the first intake temperature sensor SW2 for detecting the temperature of the fresh air, and the intake passage. The first pressure sensor SW3 located upstream of the turbocharger 44 in 40 and detecting the pressure of the gas flowing into the supercharger 44, downstream of the supercharger 44 in the intake passage 40, and the connection of the bypass passage 47. The second intake air temperature sensor SW4, which is located upstream of the position and detects the temperature of the gas flowing out from the turbocharger 44, is attached to the surge tank 42 and detects the pressure of the gas downstream of the turbocharger 44. The second pressure sensor SW5, the finger pressure sensor SW6 attached to the cylinder head 13 corresponding to each cylinder 11, and the finger pressure sensor SW6 for detecting the pressure in each combustion chamber 17, the exhaust passage 50, and the exhaust discharged from the combustion chamber 17. exhaust gas temperature sensor SW7 for detecting the temperature of the gas, arranged upstream of the catalytic converter 51 in the exhaust passage 50 and the linear O ₂ sensor SW8 for detecting the oxygen concentration in the exhaust gas, downstream of the catalytic converter 51 in the exhaust passage 50 _{Lambda O 2} sensor SW9 that is arranged and detects the oxygen concentration in the exhaust gas, water temperature sensor SW10 that is attached to the engine 1 and detects the temperature of the cooling water, is attached to the engine 1, and the rotation angle of the crank shaft 15. Crank angle sensor SW11 attached to the accelerator pedal mechanism and accelerator opening sensor SW12 attached to the accelerator pedal mechanism to detect the accelerator opening corresponding to the operation amount of the accelerator pedal, and the rotation angle of the intake cam shaft attached to the engine 1. The intake cam angle sensor SW13 to be detected, the exhaust cam angle sensor SW14 attached to the engine 1 and the exhaust cam shaft rotation angle to detect the rotation angle, the EGR passage 52, and the differential pressure upstream and downstream of the EGR valve 54 are detected. The EGR differential pressure sensor SW15 and the fuel pressure sensor SW16 attached to the common rail 64 of the fuel supply system 61 and detecting the pressure of the fuel supplied to the injector 6.

Based on these detection signals, the ECU 10 determines the operating state of the engine 1 and calculates the operation amount of each device. The ECU 10 outputs a control signal related to the calculated operation amount to the injector 6, the spark plug 25, the intake S-VT231, the intake CVVL232, the exhaust S-VT241, the exhaust IDEVA242, the fuel supply system 61, the throttle valve 43, and the EGR valve 54. The output is output to the electromagnetic clutch 45 of the supercharger 44 and the air bypass valve 48. For example, the ECU 10 adjusts the opening degree of the air bypass valve 48 based on the front-rear differential pressure of the supercharger 44 obtained from the detection signals of the first pressure sensor SW3 and the second pressure sensor SW5 to adjust the boost pressure. adjust. Further, the ECU 10 adjusts the opening degree of the EGR valve 54 based on the front-rear differential pressure of the EGR valve 54 obtained from the detection signal of the EGR differential pressure sensor SW15, thereby introducing an external EGR gas amount into the combustion chamber 17. To adjust.

The engine 1 burns the air-fuel mixture in the combustion chamber 17 by compression self-ignition for the main purpose of improving fuel efficiency and exhaust gas performance in a part of the operating region. Further, the engine 1 performs combustion by spark ignition (that is, Spark Ignition: SI) in other regions.

(Engine control)
FIG. 3 shows the configuration of the control device of the engine 1. The control device is composed of the ECU 10. The control device includes a controller 100 and an estimator 104.

The controller 100 operates the state quantity setting device (that is,) for each cycle so that the IMEP as the output of the engine 1 and the maximum pressure increase rate (dP / dθ) become the target IMEP and the target maximum dP / dθ. EGR rate) and the operating amount of the injector 6 (that is, the fuel injection amount) are set. The ECU 10 sets the target IMEP and the target maximum dP / dθ based on the detection signals of various sensors SW1 to SW16.

The controller has 100, a feedforward controller 105, a subtractor 106, a feedback controller 107, and an adder 108. The feedforward controller 105 and the feedback controller 107 require a state quantity (state variable) in the combustion chamber 17. The estimator 104 estimates a state variable that cannot be measured in real time, and outputs a state estimate value to the controller 100.

The estimator 104 is configured by a Kalman filter. The estimator 104 includes a model that models the state of the engine 1 in the combustion chamber 17. The model includes a discretized model in which the state in the combustion chamber 17 is represented by a plurality of discrete points, and a linear parameter variation model in which the discretized model is represented by linearizing the discrete points with respect to a plurality of equilibrium points. .. As shown in FIG. 11, the plurality of discrete points are the restart valve timing (EVO2) of the exhaust valve 22, the valve closing timing (IVC) of the intake valve 21, 10 degrees before the compression top dead center (-10degATDC), and the ignition timing. , Maximum pressure timing (Pmax), combustion end timing, and exhaust valve 22 opening timing (EVO). In constructing the discretization model, polytropic changes are assumed in the compression stroke and expansion stroke, and the temperature and pressure at each discrete point are obtained. For the prediction of ignition timing, consider the Livengood-Wu integral of the ignition delay time obtained in the experiment by RCM. From the ignition timing to the end of combustion, the amount of state at the end of combustion is calculated in consideration of energy conservation.

The linear parameter variation model obtained by linearizing the discretized model is represented by the state-space equations of equations (1) and (2).

Here, x _k is as the state quantity of the valve closing timing of the intake valve 21 in the current cycle (IVC), the amount of oxygen in the combustion chamber 17 (n_O2 _k), fuel quantity (n_ftotal _k in the combustion chamber 17 ) And the temperature in the combustion chamber (Tivc _k ). x _{k + 1 is the amount of oxygen in the combustion chamber 17 (n_O2 k + 1} ), the amount of fuel in the combustion chamber 17 (n_ftotal _{k + 1} ), and the combustion chamber as the state quantities of the valve closing timing of the intake valve 21 in the next cycle. The temperature inside (Tivc _{k + 1} ). u _k is as quantity operation in the current cycle, EGR rate (EGR _k) and the fuel injection amount (n_finj _k).

Further, y _{k is the indicated mean effective pressure (IMEP k} ) and the maximum pressure rise rate (dP / dθ _k ) as the output of the engine 1 in the current cycle.

The subscript k indicates the value at the kth cycle.

The linear parameter fluctuation model of the equations (1) and (2) is expressed by the following equations (1)'(2)' in a matrix format.

Here, the first coefficient A _LPV , the second coefficient B _LPV , and the third coefficient C _LPV of the linear parameter fluctuation model change according to the torque of the engine 1, respectively. The torque of engine 1 is represented by the target IMEP.

The first coefficient A _LPV is a coefficient matrix, and the characteristics of each element (A ₁₁ , A ₁₂ , A ₁₃ , A ₂₁ , A ₂₂ , A ₂₃ , A ₃₁ , A ₃₂ , A ₃₃ ) are shown in FIG. .. That is, A ₁₁ has a characteristic that it becomes smaller as the target IMEP becomes higher. A ₁₂ has a characteristic that it increases as the target IMEP increases. A ₁₃ has a characteristic that it becomes smaller as the target IMEP becomes higher. A ₂₁ has a characteristic that it becomes smaller as the target IMEP becomes higher. A ₂₂ has a characteristic that it becomes smaller as the target IMEP becomes higher. A ₂₃ has a characteristic that it increases as the target IMEP increases. A ₃₁ has a characteristic that it increases as the target IMEP increases. A ₃₂ has a characteristic that it becomes smaller as the target IMEP becomes higher. A ₃₃ has a characteristic that it increases once as the target IMEP increases and then decreases.

The second coefficient B _LPV is a coefficient matrix, and the characteristics of each element (B ₁₁ , B ₁₂ , B ₂₁ , B _22, B ₃₁ , B ₃₂ ) are shown in FIG. That is, B ₁₁ has a characteristic that it increases as the target IMEP increases. B ₁₂ remains zero regardless of the size of the target IMEP. B ₂₁ has a characteristic that it becomes smaller as the target IMEP becomes higher. B ₂₂ remains zero regardless of the size of the target IMEP. B ₃₁ has a characteristic that it increases as the target IMEP increases. B ₃₂ remains zero regardless of the size of the target IMEP.

The third coefficient C _LPV is a coefficient matrix, and the characteristics of each element (C ₁₁ , C ₁₂ , C ₁₃ , C ₂₁ , C ₂₂ , C ₂₃ ) are shown in FIG. That is, C ₁₁ has a characteristic that it increases as the target IMEP increases. C ₁₂ has a characteristic that it increases as the target IMEP increases. C ₁₃ has a characteristic that it becomes smaller as the target IMEP becomes higher. C ₂₁ has a characteristic that it becomes smaller as the target IMEP becomes higher. C ₂₂ has a characteristic that it becomes smaller as the target IMEP becomes higher. C ₂₃ has a characteristic that it becomes smaller as the target IMEP becomes higher.

The first coefficient A _LPV , the second coefficient B _LPV , and the third coefficient C _LPV are expressed by a linear function expression or a quadratic function expression with respect to the target IMEP, respectively.

The estimator 104 is configured by a Kalman filter using the linear parameter variation model described above. The Kalman filter is represented by the equation (4).

f is the discretized model described above. The Kalman gain g _LPV is calculated from the first coefficient A _LPV , the second coefficient B _LPV , and the third coefficient C _LPV. Since the calculation of Kalman gain itself is known, the description thereof will be omitted. As described above, since the first coefficient A _LPV , the second coefficient B _LPV , and the third coefficient C _LPV each change according to the target IMEP, the Kalman gain g _LPV is also the target of the engine 1 for each cycle. It is set according to IMEP. u _k is an operation amount in previously described current cycle, as described later, which is the output of the controller 100. y _k is the actual output of the engine 1 as described above, and is obtained based on the detection signal of the acupressure sensor SW6 (see also the combustion waveform in FIG. 11). Incidentally, the ^ (hat) subscript, represents an estimate of the estimator 104, the subscript ^- (negative superscript) represents a pre-estimated value.

The Kalman filter performs pre-estimation by the discretization model f. The estimation accuracy can be improved while keeping the calculation load of the estimator 104 relatively small.

The Kalman gain g _LPV is determined by using a linear parameter variation model without using a discretized model (that is, a nonlinear model). When the Kalman gain is obtained from the derivative of the nonlinear model, the estimation accuracy is high, but the computational load of the estimator 104 is considerably large. On the other hand, when the Kalman gain g _LPV is obtained from the linear parameter fluctuation model, the calculation load does not increase, but the estimation accuracy can be improved as compared with the case where the Kalman gain is obtained from a single linear model.

The estimator 104 outputs an estimated state quantity (x ^ _k : n_O2 _k , n_ftotal _k , Tivc _{k) in the combustion chamber 17.}

The feedforward controller 105 receives a target IMEP, a target maximum dP / dθ, and a state quantity estimated value x ^ _k of the estimator 104. State estimated value x _{^ k,} as described above, the amount of oxygen (n_O2 _k), fuel quantity in the combustion chamber 17 in the valve closing timing, the combustion chamber 17 of the intake valve 21 (n_ftotal _k), and, It is the temperature (Tivc _k ) in the combustion chamber 17. The feedforward controller 105 sets and outputs the feedforward control amounts u _{FF and k} based on the target IMEP, the target maximum dP / dθ, and the state quantity estimated value x ^ _k.

The subtractor 106 calculates the deviation (that is, the engine output deviation) between the target IMEP and the target maximum dP / dθ and the IMEP and the maximum dP / dθ as the actual engine output. The actual IMEP and the actual maximum dP / dθ are calculated based on the pressure history in the combustion chamber 17 detected by the acupressure sensor SW6, as described above.

The feedback controller 107 is an LQG (Linear Quadratic Gaussian) servo device that determines the amount of feedback operation. The feedback controller 107 has a cumulative unit 109 that accumulates engine output deviations from the subtractor 106, and a calculation unit 110 that calculates a feedback operation amount.

The calculation unit 110 sets the feedback control quantities u _{FB and k} based on the cumulative term of the engine output deviation and the state quantity estimated value x ^ _k of the estimator 104. Specifically, the control rule of the feedback controller is expressed by the equation (5).

However, x _{i and k} are cumulative terms of engine output deviation. K is the feedback gain. As described above, x ^ _k is a state quantity estimated value of the estimator 104.

When the equation (5) is expressed in a matrix format, it becomes the following equation (5)'.

_{The feedback controller 107 has a feedback operation amount u FB, k} based on the deviation between the target output of the engine 1 and the actual output of the engine 1 and the estimated state quantity in the combustion chamber 17 estimated by the estimator 104. By setting, the control performance of the controller 100 can be improved.

Further, the optimum value of the feedback gain for correcting the deviation of the engine output and the feedback gain for correcting the difference in the state quantity in the combustion chamber 17 can be obtained from the above-mentioned linear parameter fluctuation model (Equations (1) and (2). )reference). Therefore, the feedback gain K _LPV changes according to the target IMEP of the engine 1.

FIG. 10 illustrates the characteristics of each element _{of the feedback gain K LPV (K 11} , K ₁₂ , K ₁₃ , K ₁₄ , K ₁₅ , K ₂₁ , K ₂₂ , K ₂₃ , K ₂₄ , K ₂₅ ). K ₁₁ has a characteristic that it becomes smaller as the target IMEP becomes higher. K ₁₂ has a characteristic that it increases as the target IMEP increases. K ₁₃ has a characteristic that it becomes smaller as the target IMEP becomes higher. K ₁₄ has a characteristic that it becomes smaller as the target IMEP becomes higher. K ₁₅ has a characteristic that it becomes smaller as the target IMEP becomes higher. K ₂₁ has a characteristic that it becomes smaller as the target IMEP becomes higher. K ₂₂ has a characteristic that it becomes smaller as the target IMEP becomes higher. K ₂₃ has a characteristic that it increases as the target IMEP increases. K ₂₄ has a characteristic that it becomes smaller as the target IMEP becomes higher. K ₂₅ has a characteristic that it becomes smaller as the target IMEP becomes higher. The feedback gain K is also expressed by a linear function expression or a quadratic function expression with respect to the target IMEP.

_{The adder 108 adds the manipulated variable u FF, k} of the feed forward controller 105 and the manipulated variable _{u FB, k of the} feedback controller 107 to obtain an EGR rate as an manipulated variable of the state quantity setting device and an EGR rate. , The fuel injection amount as the operation amount of the injector 6 is determined. The controller 100 outputs an operation signal to each device. The operation amount of the state amount setting device and the operation amount of the injector 6 are also output to the estimator 104 as described above.

In the engine 1, the air-fuel mixture in the combustion chamber 17 is burned by compression self-ignition. The acupressure sensor SW6 measures the pressure history in the combustion chamber 17. Based on the pressure history measured by the acupressure sensor SW6, the actual IMEP of the engine 1 and the actual maximum dP / dθ are calculated. The calculated actual IMEP and the actual maximum dP / dθ are output to the subtractor 106 and the estimator 104 of the controller 100.

As described above, the estimator 104 is in the combustion chamber 17 of the engine 1 based on the operation amount of the state quantity setting device, the operation amount of the injector 6, and the actual IMEP and the actual maximum dP / dθ. The state quantity of is estimated for each cycle.

(Set the coefficient of the linear parameter fluctuation model and set the feedback gain)
FIG. 4 is a flowchart showing the setting procedure _{of the first coefficient A LPV} , the second coefficient B _LPV , and the third coefficient C _LPV of the linear parameter fluctuation model described above, and the setting procedure of the feedback gain K _LPV. This flow is performed before the control device is mounted on the vehicle.

In step S41 after the start, the combustion waveform at the steady set point is measured by an actual machine or a simulation. FIG. 5 illustrates a steady set point. The horizontal axis of FIG. 5 is the gross IMEP, and the vertical axis is the temperature in the combustion chamber 17 at the valve closing timing of the intake valve 21. The "stability limit" line indicates the limit at which combustion becomes unstable due to compression self-ignition. Below this line, the temperature inside the combustion chamber 17 is too low, and combustion due to compression self-ignition becomes unstable. The "combustion noise limit" line indicates the limit at which the combustion noise due to compression self-ignition exceeds the permissible value. Above this line, the temperature inside the combustion chamber is too high, and the combustion due to compression self-ignition becomes steep and the combustion noise becomes loud. The steady set point is the operating state of the engine 1 that maximizes fuel economy while avoiding these restrictions.

Returning to the flow of FIG. 4, in step S42 following step S41, the constructed discretized model is linearized by Taylor expansion around each measured stationary set point (equilibrium point), and the above-mentioned linear parameter fluctuation model and feedback are performed. Create a gain (see equations (1), (2) and (5)).

In step S43, a function of the target IMEP is created for each element of the matrix of _{each coefficient A LPV} , B _LPV , and C _LPV of the created linear parameter fluctuation model, and the matrix of the feedback gain K _LPV. That is, the characteristics of each element of FIGS. 7, 8, 9, and 10 are represented by the functional expression of the target IMEP.

In step S44, the function expression created in step S43 is stored in the memory 102 of the ECU, and the flow ends.

(Estimator estimation procedure)
FIG. 6 is a flowchart showing an estimation procedure of the estimator 104 while the engine 1 is in operation. In step S61 after the start, the state quantity estimated value x ^ _k-1 at the time of closing the intake valve 21 in the previous cycle is read. Specifically, the estimated state quantity x ^ _k-1 is the amount of oxygen in the combustion chamber 17 (n_O2 _k-1 ) and the amount of fuel in the combustion chamber 17 (n_ftotal _{k−) at the valve closing timing of the intake valve 21. 1} ), the temperature inside the combustion chamber 17 (Tivc _k-1 ).

In step S62, the state estimated value _{x ^ k-1} of the previous cycle, the amount of operation of the previous cycle _{u k-1} (i.e., EGR rate (EGR _k-1) and the fuel injection amount (n_finj _{k-1)) Therefore, the state quantity estimated value x ^ k} ⁻ at the valve closing timing of the intake valve 21 in the current cycle is estimated by the discrete model f. That is, the pre-estimation of the equation (4) is performed.

In the following step S63, the estimated output f (x ^ _k ⁻ ) of the current cycle is calculated from the state quantity estimated value x ^ _k ^{− and the discretized model f.}

_{Further, in step S64, the actual output y k} (IMEP (IMEP _k ) and maximum pressure increase rate (dP / dθ _k )) of the engine 1 is obtained by processing the in-cylinder pressure of the current cycle measured by the acupressure sensor SW6. demand.

Further, in step S65, the target IMEP (IMEP _{ref k} ) of the current cycle is read, and in the following step S66, the target IMEP and the Kalman gain from _{the first coefficient A LPV} , the second coefficient B _LPV , and the third coefficient C _LPV. g Calculate _{LPV k.} The Kalman gain may be calculated from the actual IMEP and the first coefficient A _LPV , the second coefficient B _LPV , and the third coefficient C _LPV.

In step S67 after steps S63, S64, and S66, _{the difference between the actual output y k} of the engine and the estimated output f (x ^ _k ) is multiplied by the Kalman gain g _{LPV k} to obtain the pre-estimated value x ^ _k. ^- by adding to, to update the state quantity x ^ _k of the current cycle. That is, the ex post facto estimation of the equation (4) is performed.

As described above, the estimator 104 estimates the state in the combustion chamber 17 using the linear parameter fluctuation model of the equations (1) and (2), so that the computational load of the estimator 104 is a nonlinear model. Is smaller than using. On the other hand, the first coefficient A _LPV , the second coefficient B _LPV , and the third coefficient C _LPV of the equations (1) and (2) change according to the target IMEP of the engine 1 (that is, the torque of the engine 1), respectively. , The estimation accuracy is higher than using the linear model. Therefore, the calculation load of the estimator 104 can be suppressed and the estimation accuracy can be improved at the same time.

Since the estimation accuracy of the state quantity of the combustion chamber 17 is improved, the compression self-ignition type engine 1 can realize stable combustion with good fuel economy.

Further, since the estimator 104 is configured by the Kalman filter of the equation (4), the estimation accuracy is further improved while suppressing the calculation load of the estimator 104. Specifically, the estimator 104 can improve the estimation accuracy without increasing the computational load of the estimator 104 by performing the preliminary estimation by the discretized model. Further, by calculating the Kalman gain g _LPV from the first coefficient A _LPV , the second coefficient B _LPV , and the third coefficient C _LPV , the Kalman gain is obtained from the derivative of the discretized model while maintaining high estimation accuracy. The calculation load can be significantly reduced compared to the case where it is required.

In the above configuration, the estimator 104 estimates the state quantity of the valve closing timing of the intake valve 21. The estimator 104 may estimate, for example, the state quantity of the restart valve timing of the exhaust valve 22.

Further, in the above configuration, the first coefficient A _LPV , the second coefficient B _LPV , the third coefficient C _LPV , and the feedback gain K _LPV of the linear parameter fluctuation model are represented as functions of the target IMEP. The first coefficient A _LPV , the second coefficient B _LPV , the third coefficient C _LPV , and the feedback gain K _LPV of the linear parameter fluctuation model may be a function of the target IMEP and the boost pressure. For example, equation (6) exemplifies a functional equation of element A _{11 of the first coefficient A LPV.}

Here, Pibc is the boost pressure in the combustion chamber at the valve closing timing of the intake valve 21. α, β, γ, and δ are coefficients, respectively. The temperature Tivc in the combustion chamber 17 by making the first coefficient A _LPV , the second coefficient B _LPV , and the third coefficient C _LPV , and the feedback gain K _LPV a function of the target IMEP and the boost pressure. It is possible to improve the estimation and the estimation accuracy of the oxygen amount n_O2 in the combustion chamber.

Furthermore, the first coefficient A _LPV , the second coefficient B _LPV , the third coefficient C _LPV , and the feedback gain K _LPV of the linear parameter fluctuation model can be a function of the target IMEP, the boost pressure, and the engine speed. good. For example, equation (7) exemplifies a functional equation of element A _{11 of the first coefficient A LPV.}

Here, rpm is the engine speed. α, β, γ, δ, ε, and ξ are coefficients, respectively. The operating region of the engine 1 by making the first coefficient A _LPV , the second coefficient B _LPV , the third coefficient C _LPV , and the feedback gain K _LPV a function of the target IMEP, the boost pressure, and the engine speed. It is possible to improve the estimation accuracy of the state quantity in the combustion chamber 17 over a wide range of.

Further, the estimator 104 may be configured by a Kalman filter represented by the following equation (3).

Here, g _LPV is Kalman gain. Equation (3) is a linear Kalman filter, and the computational load of the estimator 104 is further reduced. On the other hand, since the first coefficient A _LPV , the second coefficient B _LPV , and the third coefficient C _LPV of the linear parameter fluctuation model are functions of the target IMEP, respectively, as described above, the state quantity in the combustion chamber 17 The estimation accuracy can be improved over that of a single linear model.

The technique disclosed herein is not limited to being applied to the engine 1 having the above-described configuration. As the configuration of the engine 1, various configurations can be adopted.

1 engine 10 ECU
100 Controller 104 Estimator 105 Feedforward controller 107 Feedback controller 17 Combustion chamber 231 Intake S-VT (state quantity setting device)
232 Intake CVVL (state quantity setting device)
241 Exhaust S-VT (state quantity setting device)
242 IDEVA (state quantity setting device)
43 Throttle valve (state quantity setting device)
49 Supercharging system 55 EGR system (state quantity setting device)
6 Injector SW1 Airflow sensor SW2 1st intake air temperature sensor SW3 1st pressure sensor SW4 2nd intake air temperature sensor SW5 2nd pressure sensor SW6 Finger pressure sensor SW7 Exhaust temperature sensor SW8 Linear O ₂ sensor SW9 Lambda O ₂ sensor SW10 Water temperature sensor SW11 Crank angle Sensor SW12 Accelerator opening sensor SW13 Intake cam angle sensor SW14 Exhaust cam angle sensor SW15 EGR differential pressure sensor SW16 Fuel pressure sensor

Claims (10)

An engine configured to self-ignite the air-fuel mixture in the combustion chamber at each cycle.
A state quantity setting device attached to the engine and configured to bring the inside of the combustion chamber into a desired state by adjusting the introduction of fresh air and burned gas into the combustion chamber.
An injector mounted on the engine and configured to inject fuel to be supplied into the combustion chamber.
A sensor attached to the engine and configured to detect parameters related to the pressure in the combustion chamber and output a detection signal.
A controller configured to operate the engine by connecting each of the state quantity setting device, the injector, and the sensor and outputting an operation signal to the state quantity setting device and the injector.
It is provided with an estimator configured to receive the operation signal of the controller and the detection signal of the sensor, estimate the state quantity in the combustion chamber, and output the state estimation signal to the controller. ,
The controller receives the detection signal of the sensor and the state estimation signal of the estimator, and the state quantity so that the combustion in the combustion chamber becomes the combustion corresponding to the target output of the engine. Set the operation amount of the device and the operation amount of the injector, respectively.
The estimator estimates the state quantity in the combustion chamber by using the linear parameter fluctuation model that models the state in the combustion chamber shown in the equation (1).
_{The first coefficient A LPV} and the second coefficient B _LPV in the linear parameter fluctuation model are control devices for a compression self-ignition engine that change according to the torque of the engine and the rotation speed of the engine, respectively.

However, x _k is one or both of the amount of oxygen in the combustion chamber and the amount of fuel in the combustion chamber as the state amount at a predetermined timing in the current cycle, and in the combustion chamber. The temperature of. x _{k + 1} is one or both of the amount of oxygen in the combustion chamber and the amount of fuel in the combustion chamber, and the temperature in the combustion chamber as the state quantity at a predetermined timing in the next cycle. be. u _k is as the operation amount in the current cycle, a EGR rate and fuel injection amount. An engine configured to self-ignite the air-fuel mixture in the combustion chamber at each cycle.
A state quantity setting device attached to the engine and configured to bring the inside of the combustion chamber into a desired state by adjusting the introduction of fresh air and burned gas into the combustion chamber.
An injector mounted on the engine and configured to inject fuel to be supplied into the combustion chamber.
A sensor attached to the engine and configured to detect parameters related to the pressure in the combustion chamber and output a detection signal.
A controller configured to operate the engine by connecting each of the state quantity setting device, the injector, and the sensor and outputting an operation signal to the state quantity setting device and the injector.
It is provided with an estimator configured to receive the operation signal of the controller and the detection signal of the sensor, estimate the state quantity in the combustion chamber, and output the state estimation signal to the controller. ,
The controller receives the detection signal of the sensor and the state estimation signal of the estimator, and the state quantity so that the combustion in the combustion chamber becomes the combustion corresponding to the target output of the engine. Set the operation amount of the device and the operation amount of the injector, respectively.
The estimator estimates the state quantity in the combustion chamber by using the linear parameter fluctuation model that models the state in the combustion chamber shown in the equations (1) and (2).
The first coefficient A _LPV , the second coefficient B _LPV , and the third coefficient C _LPV in the linear parameter fluctuation model are each a control device for a compression self-ignition engine that changes according to the torque of the engine.

However, x _k is one or both of the amount of oxygen in the combustion chamber and the amount of fuel in the combustion chamber as the state amount at a predetermined timing in the current cycle, and in the combustion chamber. The temperature of. x _{k + 1} is one or both of the amount of oxygen in the combustion chamber and the amount of fuel in the combustion chamber, and the temperature in the combustion chamber as the state quantity at a predetermined timing in the next cycle. be. u _k is as the operation amount in the current cycle, a EGR rate and fuel injection amount. y _k is the indicated mean effective pressure and the maximum pressure increase rate (dP / dθ) as the output of the engine in the current cycle. In the control device for the compression self-ignition engine according to claim 2.
The estimator is configured by a Kalman filter represented by Eq. (3) using the linear parameter variation model.
The first coefficient A _LPV , the second coefficient B _LPV , and the third coefficient C _LPV are each a control device for a compression self-ignition engine set according to the target torque of the engine for each cycle.

However, g _LPV is Kalman gain. In the control device for the compression self-ignition engine according to claim 2.
The estimator is a control device for a compression self-ignition engine configured by a Kalman filter represented by the equation (4) using the linear parameter fluctuation model.

However, f is a discretized model that models the state in the combustion chamber, and the linear parameter fluctuation model is obtained from the discretized model. The g _LPV is a Kalman gain calculated from the first coefficient A _LPV , the second coefficient B _LPV , and the third coefficient C _{LPV , and the g LPV} corresponds to the target torque of the engine for each cycle. Is set. y _k is the actual output of the engine obtained based on the detection signal of the sensor. In the control device for the compression self-ignition engine according to any one of claims 2 to 4.
It comprises a supercharging system attached to the engine and configured to supercharge the gas introduced into the combustion chamber.
The first coefficient A _LPV , the second coefficient B _LPV , and the third coefficient C _LPV are each a control device for a compression self-ignition engine that changes according to the torque and boost pressure of the engine. In the control device for the compression self-ignition engine according to any one of claims 2 to 5.
The first coefficient A _LPV , the second coefficient B _LPV , and the third coefficient C _LPV are each a control device for a compression self-ignition engine that further changes according to the rotation speed of the engine. In the control device for the compression self-ignition engine according to any one of claims 2 to 6.
The first coefficient A _LPV , the second coefficient B _LPV , and the third coefficient C _LPV are control devices for a compression self-ignition engine represented by a linear function expression or a quadratic function expression, respectively. In the control device for the compression self-ignition engine according to any one of claims 2 to 7.
The controller has a feedback controller.
The feedback controller receives the deviation between the target output of the engine and the output of the engine based on the detection signal of the sensor, and the state estimation signal of the estimator, and inputs the feedback operation amounts u _{FB, k} . Output and
The control rule of the feedback controller is the control device of the compression self-ignition engine according to the equation (5).

However, x _{i and k} are cumulative terms of the deviation. K _LPV is the feedback gain. In the control device for the compression self-ignition engine according to claim 8.
The feedback gain K _LPV is set based on the linear parameter fluctuation models of the equations (1) and (2), and the feedback gain K _LPV is set for each cycle according to the target torque of the engine. The control device of the compression self-igniting engine to be. In the control device for the compression self-ignition engine according to claim 8 or 9.
The controller further has a feedforward controller.
The feedforward controller is a control device for a compression self-ignition engine configured to receive the target output and the state estimation signal of the estimator and output a feedforward operation amount.

Images