Electronic control units and engine control systems

Abstract

The present invention is an electronic control device that controls an engine, wherein the engine comprises: an EGR system having an EGR pipe and an EGR valve arranged in the EGR pipe; an airflow sensor that is arranged on the intake pipe; a throttle valve that is on the downstream side of the airflow sensor; and an intake pipe pressure sensor that detects the intake pipe pressure. The electronic control device comprises: a state inference unit that infers the intake pipe pressure and the EGR rate based on at least the detection value of the airflow sensor and the EGR valve opening; and an inference value correction unit that corrects the EGR rate inference value of the state inference unit based on the detection value of the intake pipe pressure sensor and the intake pipe pressure inference value of the state inference unit.

Description

Technical Field

[0001] The invention relates to an electronic control device and an engine control system. Background Art

[0002] Conventionally, control technology is known for improving the combustion performance of internal combustion engines by recirculating (recirculating) a portion of the exhaust gas diverted from the engine's exhaust pipe back into the intake pipe. This control technology implements a system that controls the amount of air drawn into the engine and the ratio of exhaust gas recirculation using valve opening, while also controlling the fuel injection rate and ignition timing based on the relationship between the amount of fresh air detected by an intake air sensor and the exhaust gas recirculation ratio. (See, for example, Patent Document 1).

[0003] Prior art literature

[0004] Patent Literature

[0005] Patent Document 1: Japanese Patent Application Laid-Open No. 2011-169196 Summary of the Invention

[0006] Problems to be solved by the invention

[0007] Furthermore, increasing the exhaust gas recirculation rate (EGR ratio) can enhance the reduction of pumping losses caused by throttle tightness during part-load operation and improve knock reduction effects at high loads. However, excessively increasing the EGR ratio can lead to combustion instability and misfiring. Furthermore, if the EGR ratio falls short of the control target value, sufficient knock reduction can be achieved, leading to problems such as knocking and other abnormal combustion.

[0008] The control system described in Patent Document 1 discloses the following: It estimates intake manifold pressure based on the relationship between the intake air volume sensor, throttle opening, and EGR valve opening, and compares this estimated value with the measured value from an intake manifold pressure sensor installed in the intake manifold to determine whether there is an EGR system abnormality. Furthermore, the control system described in Patent Document 1 discloses diagnostic technology that combines the estimated intake manifold pressure with information from an air-fuel ratio sensor installed in the exhaust manifold to distinguish between fuel system abnormalities and EGR system abnormalities. However, in high-EGR combustion systems, high EGR control accuracy is required to prevent the aforementioned knock and misfire. Therefore, even EGR control errors within the normal range must be detected and appropriately controlled to maintain a high-precision EGR control state.

[0009] The present invention has been made in view of the above-mentioned situation, and an object of the present invention is to maintain EGR control accuracy at a high level and prevent combustion failure of an internal combustion engine caused by EGR control errors.

[0010] Technical means to solve the problem

[0011] To solve the above problems, an electronic control device in one form of the present invention is an electronic control device for controlling an engine, wherein the engine comprises: an EGR system, which has an EGR pipe for returning a portion of the exhaust gas of the internal combustion engine to the intake pipe and an EGR valve arranged in the EGR pipe; an air flow sensor, which detects the flow rate of air introduced into the intake pipe; a throttle valve, which is arranged on the downstream side of the air flow sensor; and an intake pipe pressure sensor, which is arranged on the downstream side of the throttle valve and on the downstream side of the connection part between the intake pipe and the EGR pipe, and detects the pressure downstream of the throttle valve in the intake pipe, that is, the intake pipe pressure. The electronic control device comprises: a state inference unit, which infers the intake pipe pressure and the EGR rate based on the detection value of the air flow sensor and the EGR valve opening; and an inference value correction unit, which corrects the EGR rate inference value inferred by the state inference unit based on the detection value of the intake pipe pressure sensor and the intake pipe pressure inference value inferred by the state inference unit.

[0012] Effects of the Invention

[0013] According to at least one aspect of the present invention, an estimated value correction unit corrects the EGR rate estimated value based on the detected value of the intake pipe pressure sensor and the intake pipe pressure estimated value. This maintains high EGR control accuracy and prevents internal combustion engine combustion defects (e.g., knocking or misfiring) caused by EGR control errors.

[0014] Other problems, structures, and effects than those described above will become clear from the following description of the embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 This is a schematic diagram of the overall configuration of an engine system that is a control target of an engine control system according to an embodiment of the present invention.

[0016] Figure 2 This is a block diagram showing an example of the hardware configuration of an ECU.

[0017] Figure 3 This is a graph showing the relationship between the EGR rate and the ignition advance angle.

[0018] Figure 4 This is a block diagram showing an example of the functional configuration of an ECU.

[0019] Figure 5 This figure explains a control block that executes throttle and EGR valve opening control to achieve target torque and target EGR rate.

[0020] Figure 6This figure explains a physical model that is considered when constructing a throttle and EGR valve opening control model for achieving a target torque and a target EGR rate.

[0021] Figure 7 This diagram uses a schematic diagram of a valve cross section to explain a method for calculating a target valve opening based on a valve flow rate calculation model.

[0022] Figure 8 A diagram illustrating a control block that estimates the state in the intake pipe using a Kalman filter based on information from an air flow sensor and an intake pipe pressure sensor, and corrects the throttle opening and EGR valve opening based on the estimated state in the intake pipe.

[0023] Figure 9 This is a conceptual diagram showing, in block form, the functions used when constructing a control model that implements internal state feedback control.

[0024] Figure 10 This is a block diagram showing an example of the internal structure of a Kalman filter.

[0025] Figure 11 is a flowchart showing an example of a Kalman filter algorithm.

[0026] Figure 12 A diagram illustrating a control block that performs learning of the throttle flow coefficient, the EGR valve flow coefficient, and the cylinder intake efficiency.

[0027] Figure 13 Flowchart showing an example of a recursive least squares algorithm for system identification.

[0028] Figure 14 This figure explains the control operation and its effect when carbon deposits are attached to the throttle valve in the control of the throttle opening and the EGR valve opening to achieve the target torque and the target EGR rate.

[0029] Figure 15 This diagram explains the control operation (internal state feedback based on Kalman filtering) and its effects when carbon deposits are attached to both the throttle valve and the EGR valve in controlling the throttle opening and the EGR valve opening to achieve the target torque and the target EGR rate.

[0030] Figure 16 This diagram explains the control action (internal state feedback and system identification based on Kalman filtering) and its effects when carbon deposits are attached to both the throttle valve and the EGR valve in controlling the throttle opening and the EGR valve opening to achieve the target torque and the target EGR rate.

[0031] Figure 17This is a flowchart showing an example of a procedure for executing throttle valve control and EGR valve control based on the detection values ​​of the air flow sensor and the detection values ​​of the intake pipe pressure sensor. DETAILED DESCRIPTION

[0032] In the following, examples of specific embodiments of the present invention will be described with reference to the accompanying drawings. In this specification and the accompanying drawings, components having substantially the same function or configuration are denoted by the same reference numerals, and repeated descriptions are omitted.

[0033] [Engine system schematic structure]

[0034] First, refer to Figure 1 , the overall configuration of an engine system as a control target of an engine control system according to an embodiment of the present invention will be described.

[0035] Figure 1 The following is a schematic diagram illustrating the overall configuration of an engine system controlled by an engine control system according to one embodiment of the present invention. The engine system includes an internal combustion engine 1, an accelerator pedal position sensor 2, an air flow sensor 3, a throttle valve 4, an intake manifold 5, a flow enhancement valve 7, an intake valve 8, an exhaust valve 10, a fuel injection valve 12, a spark plug 13, and a crank angle sensor 20. Furthermore, the engine system includes an air-fuel ratio sensor 14, an EGR (Exhausted Gas Recirculation) pipe 15, an EGR cooler 16, an EGR temperature sensor 17, an EGR valve upstream pressure sensor 18, an EGR valve 19, and an ECU (Electronic Control Unit) 21.

[0036] The throttle valve 4 is located upstream of the intake manifold 5 formed in the intake pipe 31. It controls the amount of intake air flowing into the cylinders of the internal combustion engine 1 by narrowing the intake flow path. The throttle valve 4 comprises an electrically controlled butterfly valve whose valve opening can be controlled independently of the driver's accelerator pedal depression. The intake manifold 5, equipped with an intake manifold pressure sensor 6, communicates downstream of the throttle valve 4.

[0037] The flow enhancement valve 7 is located downstream of the intake manifold 5 and biases the air drawn into the cylinder, thereby increasing the turbulence of the airflow inside the cylinder. When performing exhaust gas recirculation combustion, which will be described later, the flow enhancement valve 7 is closed to promote and stabilize turbulent combustion.

[0038] Internal combustion engine 1 is equipped with intake valves 8 and exhaust valves 10. Each of intake valves 8 and exhaust valves 10 has a variable valve mechanism that continuously changes the valve opening and closing phases. The variable valve mechanisms of intake valves 8 and exhaust valves 10 are equipped with intake valve position sensors 9 and exhaust valve position sensors 11, respectively, for detecting the valve opening and closing phases. Cylinders of internal combustion engine 1 are equipped with direct fuel injection valves 12 that inject fuel directly into the cylinders. Fuel injection valves 12 may also employ port injection, injecting fuel into the intake port.

[0039] A spark plug 13 is mounted on the cylinder of the internal combustion engine 1. Its electrode portion is exposed within the cylinder, and it ignites the combustible mixture via a spark. A crank angle sensor 20 is mounted on the crankshaft and outputs a signal corresponding to the crankshaft's rotational angle as a signal indicating the engine speed to the ECU 21. An air-fuel ratio sensor 14 is mounted on the exhaust pipe 32 and outputs a signal indicating the detected exhaust gas composition, or air-fuel ratio, to the ECU 21.

[0040] An EGR system is constructed including an EGR pipe 15 and an EGR valve 19 disposed in the EGR pipe 15. The EGR pipe 15 connects the exhaust flow path (intake pipe 31) with the intake flow path (exhaust pipe 32), diverts the exhaust gas from the exhaust flow path and refluxes (recirculates) it downstream of the throttle valve 4. The EGR cooler 16 provided on the EGR pipe 15 cools the exhaust gas. The EGR valve 19 is provided downstream of the EGR cooler 16 to control the flow rate of the exhaust gas. The EGR pipe 15 is provided with an EGR temperature sensor 17 for detecting the temperature of the exhaust gas flowing upstream of the EGR valve 19 and an EGR valve upstream pressure sensor 18 for detecting the pressure upstream of the EGR valve 19.

[0041] ECU 21 is an example of an electronic control unit that controls the various components of the engine system and performs various data processing. The engine system and ECU 21 constitute an engine control system. The various sensors and actuators described above are connected to ECU 21 in a communicative manner. ECU 21 controls the operation of actuators such as the throttle valve 4, fuel injection valve 12, intake valve 8, exhaust valve 10, and EGR valve 19. Furthermore, ECU 21 detects the operating state of internal combustion engine 1 based on signals input from various sensors and ignites spark plug 13 at a time determined by the operating state. Furthermore, if an abnormality or malfunction is detected in the engine system including internal combustion engine 1, ECU 21 illuminates the corresponding warning indicator lamp 22 (MIL).

[0042] [ECU hardware configuration]

[0043] Figure 2This is a block diagram showing an example of the hardware configuration of the ECU 21. The ECU 21 includes a control unit 23, a storage unit 24, and an input / output interface 25, which are interconnected via a system bus 26. The control unit 23 is composed of a CPU (central processing unit) 23a, a ROM (Read Only Memory) 23b, and a RAM (Random Access Memory) 23c. The CPU 23a loads the control program stored in the ROM 23b into the RAM 23c and executes it, thereby realizing the various functions of the ECU 21. The storage unit 24, which serves as an auxiliary storage device composed of a semiconductor memory or the like, records state-space models, parameters, data obtained by executing the control program, and the like. The storage unit 24 may also store the control program.

[0044] The input / output interface 25 is an interface for communicating signals and data with various sensors and actuators. The ECU 21 includes an A / D (analog / digital) converter (not shown) and a drive circuit, etc., for processing the input and output signals of each sensor. The input / output interface 25 can also serve as the A / D converter. Furthermore, while a CPU is used as the processor, another processor such as an MPU (microprocessing unit) can also be used.

[0045] [Relationship between EGR rate and ignition advance angle]

[0046] Here, reference Figure 3 , the relationship between the EGR rate and the ignition advance angle is explained.

[0047] Figure 3 This is a graph showing the relationship between the EGR rate and the ignition advance angle. The horizontal axis represents the EGR rate, and the vertical axis represents the ignition advance angle. The EGR rate represents the proportion of exhaust gas that flows back from the EGR pipe 15 relative to the intake air (new air) of the intake pipe 31. The EGR rate and the ignition advance angle are controlled in such a way that the operating point does not deviate from the area sandwiched by the knock limit curve and the misfire limit curve. The higher the EGR rate, the narrower the area sandwiched by the knock limit curve and the misfire limit curve, and the smaller the allowable error of the EGR control accuracy shown by the single-dot chain line. Therefore, the higher the EGR rate, the more likely it is that problems such as combustion instability or misfire will occur, and thus high EGR control accuracy is required. That is, especially in a high EGR rate combustion system, it is required to maintain the EGR control accuracy at a high level to prevent knock or misfire of the internal combustion engine 1 caused by EGR control errors.

[0048] [ECU functional configuration]

[0049] Next, refer to Figure 4 , the functional structure of ECU 21 is described.

[0050] Figure 4 : is a block diagram showing an example of the functional configuration of the ECU 21. The ECU 21 is configured to execute EGR valve control and EGR valve characteristic learning using a Kalman filter. This block diagram schematically shows the functional configuration of the ECU 21 according to this embodiment.

[0051] The ECU 21 includes a state-space model setting unit 410 , a Kalman filter 420 , a combustion control unit 430 , a valve correction amount calculation unit 440 , and a learner (system identification) 450 .

[0052] The state-space model setting unit 410 sets a state-space model representing the internal state of the intake pipe 31 based on information such as the detection value (AFS) of the air flow sensor 3, which detects the flow rate of air introduced into the intake pipe 31, the throttle opening (Th / V), and the EGR valve opening (EGR / V). The state-space model setting unit 410 then outputs the settings of the state-space model to the Kalman filter 420.

[0053] The Kalman filter 420 uses the state-space model set in the state-space model setting unit 410 to estimate the internal state of the intake pipe 31 (hereinafter referred to as the "intake pipe internal state") and then corrects (corrects) the estimated intake pipe internal state. The Kalman filter 420 includes an intake pipe internal state estimation unit 421 and a state observer 422. The intake pipe internal state estimation unit 421 and the state observer 422 are generalized Kalman filters, and the state observer 422 has the correction function that is the essence of the Kalman filter.

[0054] The intake pipe state estimation unit 421 estimates the intake pipe state using the state-space model set in the state-space model setting unit 410 and outputs information about the estimated intake pipe state to the state observer 422. The state observer 422 corrects (amends) the intake pipe state estimated by the intake pipe state estimation unit 421 based on the detection value of the intake pipe pressure sensor 6. The state observer 422 outputs the corrected intake pipe pressure estimate and the corrected EGR rate estimate as the intake pipe state information.

[0055] The combustion control unit 430 performs ignition timing control and / or fuel injection control using the information on the state in the intake pipe (the corrected intake pipe pressure estimated value and the corrected EGR rate estimated value) output from the state observer 422 of the Kalman filter 420 .

[0056] The valve correction amount calculation unit 440 calculates the valve correction amount (e.g., EGR valve opening correction amount) using the output of the state observer 422 of the Kalman filter 420 (e.g., the corrected EGR rate estimation value), and outputs the calculation result to the learner 450. Figure 8 As shown, an EGR valve opening correction amount calculation unit 805 and a throttle opening correction amount calculation unit 806 are provided as the valve correction amount calculation unit 440 .

[0057] The learner (system identification) 450 learns the valve characteristics of the controlled object using the output of the valve correction amount calculation unit 440 (eg, the EGR valve opening correction amount), and outputs the learning result to the intake pipe state estimation unit 421 of the Kalman filter 420 . Figure 4 In the embodiment, the learner 450 shows an example of learning the EGR valve characteristics, but the valve to be learned is not limited to this example. Figure 12 As shown, a throttle flow coefficient system identification unit 1204 , an EGR valve flow coefficient system identification unit 1205 , and a cylinder intake efficiency system identification unit 1206 are provided as the learner 450 .

[0058] [Control block that controls the opening of the throttle and EGR valve]

[0059] Next, refer to Figure 5 , the opening control of the throttle and EGR valve to achieve the target torque and target EGR rate is described.

[0060] Figure 5 A control block that controls the openings of the throttle and EGR valves to achieve target torque and target EGR rate is shown.

[0061] The target torque calculation unit 501 calculates the target torque of the internal combustion engine 1 based on, for example, the rotational speed of the internal combustion engine 1, the amount of accelerator pedal depression, and the externally requested torque. The externally requested torque is a torque that is independent of the driver's intention (e.g., the amount of accelerator pedal depression) and is determined based on the vehicle interior air conditioning state and the power generation state of the AC generator.

[0062] The target charging efficiency calculation unit 502 calculates a target charging efficiency related to the amount of fresh air intake into the cylinder required to achieve the target torque, taking into account the current engine speed and target torque. The charging efficiency is expressed as a percentage of the amount of intake air relative to the cylinder volume.

[0063] The target throttle opening calculation unit 503 calculates the throttle opening command value for achieving the target intake air amount by taking into account the current speed and the target charging efficiency. Here, the throttle flow coefficient system identification unit 1204 (see Figure 12The ECU 21 controls the throttle valve 4 based on the throttle opening command value and the throttle opening correction amount calculated by the throttle opening correction amount calculation unit 806 ("B806" in the figure).

[0064] Meanwhile, the target EGR rate calculation unit 504 calculates the target EGR rate based on the current engine speed and target torque. Exhaust gas recirculation (EGR) is performed at low and medium loads to reduce pumping loss due to throttle tightness, and at high loads to reduce knocking.

[0065] The target EGR valve opening calculation unit 505 calculates the target EGR valve flow rate based on the current speed, target charging efficiency and target EGR rate, and then calculates the target EGR valve opening instruction value to achieve the target EGR valve flow rate. Here, the EGR valve flow coefficient system identification unit 1205 (see the following) described later is taken into account in the calculation of the EGR valve opening instruction value. Figure 12 The ECU 21 controls the EGR valve 19 based on the EGR valve opening instruction value and the EGR valve opening correction amount calculated by the EGR valve opening correction amount calculation unit 805 ("B805" in the figure).

[0066] [Physical model considered when constructing the opening control model]

[0067] Figure 6 This figure illustrates the physical model considered when constructing the throttle and EGR valve opening control model to achieve the target torque and target EGR rate. The pressure in the intake pipe 31 (e.g., the intake manifold 5) (hereinafter referred to as "intake pipe pressure") p is defined as a state quantity in the intake pipe. m and the EGR rate ξ in the intake pipe 31 m , can be obtained by the following formulas (1) and (2) respectively.

[0068] [Formula 1]

[0069]

[0070] [Formula 2]

[0071]

[0072] Here, the m with a superscript dot th The flow rate through the throttle valve is m, which is marked with a dot above. egr The flow rate through the EGR valve is m, which is marked with a dot. cyl is the cylinder suction flow, κ is the polytropic index, R is the gas constant, V m is the intake manifold volume, T atm is the atmospheric temperature, Tegr is the EGR temperature, T m is the temperature in the intake pipe. The superscript dot symbol indicates the first-order differential based on time.

[0073] Throttle flow rate (m with superscript dot mark) th ) can be obtained by the following formula (3). Furthermore, the throttle valve flow rate is roughly equivalent to the detection value of the air flow sensor 3 (mafs with a superscript dot mark).

[0074] [Formula 3]

[0075]

[0076] Here, ρ atm is the atmospheric density, μ th is the throttle flow coefficient, D th is the outer diameter of the throttle valve, φ th is the throttle opening, φ th0 is the minimum throttle opening, p atm is atmospheric pressure.

[0077] EGR valve flow rate (m egr ) can be obtained by the following formula (4).

[0078] [Formula 4]

[0079]

[0080] Here, ρ e is the EGR density (recirculated exhaust gas density), μ egr is the EGR valve flow coefficient, D egr is the outer diameter of the EGR valve, φ egr is the EGR valve opening, φ egr0 is the minimum opening of the EGR valve. cyl ) is obtained from the following formula (5).

[0081] [Formula 5]

[0082]

[0083] Here, N e is the speed of the internal combustion engine 1 (speed per minute), η in is the intake efficiency, V d is the total stroke volume of the internal combustion engine 1. The intake efficiency is a value representing the ratio of the mass of gas actually drawn into the cylinder to the mass of gas in the intake manifold equivalent to the stroke volume of all cylinders (eg, four cylinders) as a reference (1.0).

[0084] Filling efficiency η of fresh air sucked into the cylinder ch It is defined by the following formula (6).

[0085] [Formula 6]

[0086]

[0087] Here, p0 and T0 are the temperature and pressure of the atmosphere under standard conditions (for example, 25° C. and 101.325 kPa).

[0088] The net mean effective pressure, which is an indicator of torque, is obtained by the following equation (7).

[0089] [Formula 7]

[0090]

[0091] Here, H L is the low calorific value of the fuel, η ite is the thermal efficiency shown in the figure, φ is the equivalence ratio, L0 is the theoretical air-fuel ratio, p f The friction mean effective pressure is related to the friction torque. Friction torque is the torque that resists the motion of contacting objects due to friction.

[0092] [Target valve opening calculation based on valve flow calculation model]

[0093] Here, reference Figure 7 , the target valve opening calculation based on the valve flow calculation model is explained.

[0094] Figure 7 This figure uses a valve cross-sectional diagram to illustrate the method of calculating the target valve opening based on the valve flow calculation model. The dotted m in the figure is the valve flow, and p up is the pressure of the gas on the upstream side of the valve (in), p down is the pressure of the gas on the downstream side of the valve (out), ρ up is the density of the gas upstream of the valve, D is the outer diameter of the valve, and φ is the valve opening. The shaded area represents the flow path of the gas through the valve. The cross-sectional area of ​​this flow path, or the opening area S, is expressed by the following equation (8).

[0095] [Formula 8]

[0096]

[0097] In the case where the above valve is a throttle valve, Figure 7 p up Equivalent to atmospheric pressure patm, p down Equivalent to the intake manifold pressure p m , ρup Equivalent to the atmospheric density ρ atm , D is equivalent to the outer diameter of the throttle valve D th The throttle flow rate formula of formula (3) is transformed and the throttle flow rate formula is calculated in reverse as in formula (9) below to obtain the target air volume (m) specified by the target torque and speed. th,d ) throttle opening φ th .

[0098] [Formula 9]

[0099]

[0100] The above formula (9) can be moved to the table calculation of throttle opening and opening area and used for Figure 5 The target throttle opening calculation unit 503 calculates the target throttle opening.

[0101] Similarly, when the valve is an EGR valve, Figure 7 p up Equivalent to the EGR valve upstream pressure p egr , p down Equivalent to the intake manifold pressure p m , ρ up Equivalent to EGR density ρ egr , D is equivalent to the outer diameter of the EGR valve D egr The EGR valve flow rate formula of formula (4) is transformed and the EGR valve flow rate formula is calculated in reverse as in formula (10) below to obtain the target EGR flow rate (m marked with a superscript dot) specified by the target torque and speed. egr,d ) of the EGR valve opening φ egr .

[0102] [Formula 10]

[0103]

[0104] The above formula (10) can be moved to the table calculation of EGR valve opening and opening area and used for Figure 5 The target EGR valve opening is calculated by the target EGR rate calculation unit 504.

[0105] [Correction of throttle opening and EGR valve opening]

[0106] Figure 8 A control block is shown which estimates the state in the intake pipe based on the output information of the air flow sensor 3 and the intake pipe pressure sensor 6 by means of the Kalman filter 420 and corrects the throttle opening and the EGR valve opening based on the estimated state in the intake pipe.

[0107] In the state-space model setting unit 410, the state in the intake pipe defined by equations (1) and (2) is described using the state-space model (Equations (12) and (13)) described later. The matrix, state vector, input vector, and output vector (Equations (14) to (16)) are defined based on the input information. In this embodiment, the state-space model setting unit 410 receives the detection value of the airflow sensor 3, the calculation result of the EGR valve flow rate calculation unit 803, the calculation result of the cylinder intake flow rate calculation unit 804, and the state in the intake pipe output by the Kalman filter 420. The state-space model setting unit 410 redefines each element of the state-space model for each discrete time step.

[0108] In the Kalman filter 420, based on the matrix, state vector, input vector and output vector defined in the state space model setting unit 410, the detection value of the intake pipe pressure sensor 6 is calculated according to the Kalman filter algorithm described later (refer to Figure 11 ) updates (corrects) the state vector representing the state within the intake pipe. Here, the state vector is a vector consisting of the intake pipe pressure (Equation (1)) and the intake pipe EGR rate (Equation (2)). That is, Kalman filter 420 outputs the updated (corrected) estimated intake pipe pressure and estimated EGR rate as the state within the intake pipe. Furthermore, unless otherwise specified, the EGR rate in this specification refers to the intake pipe EGR rate.

[0109] The EGR valve flow rate calculation unit 803 calculates the EGR valve flow rate based on the intake pipe pressure estimate value updated by the Kalman filter 420, the EGR valve opening, the EGR valve upstream state, and the EGR valve flow coefficient. The EGR valve upstream state refers to the detection values ​​of the EGR temperature sensor 17 and the EGR valve upstream pressure sensor 18, both located upstream of the EGR valve 19. By using the intake pipe pressure estimate value updated by the Kalman filter 420, the EGR valve flow rate can be accurately estimated, taking into account the actual measured intake pipe pressure value.

[0110] The cylinder intake flow calculation unit 804 calculates the cylinder intake flow based on the intake pipe pressure estimation value and EGR rate estimation value updated by the Kalman filter 420, the rotational speed, the variable valve state, and the intake efficiency. The cylinder intake flow is the flow of the intake air running from the intake manifold 5 to the cylinder. Here, the so-called variable valve state is the detection value (phase) of the intake valve position sensor 9 and the exhaust valve position sensor 11. In this way, by using the intake pipe pressure estimation value and the EGR rate estimation value updated by the Kalman filter 420, the cylinder intake flow, the new air filling efficiency, and the cylinder EGR rate can be estimated with high precision taking into account the actual measured value of the intake pipe pressure. The cylinder EGR rate is the ratio of the new air to the exhaust gas in the cylinder. Then, in the combustion control unit 430 ( Figure 4 ), this information is used for ignition timing control and fuel injection control, thereby appropriately reflecting the current state and implementing ignition timing control and fuel injection control with high precision and robustness.

[0111] The EGR valve opening correction amount calculation unit 805 uses the equation (28) described later to calculate the EGR valve opening correction amount (equation (28) described later) based on the difference between the EGR rate estimated value updated by the Kalman filter 420 and the target EGR rate. By calculating the EGR valve opening correction amount using the EGR rate estimated value updated by the Kalman filter 420, the EGR rate can be controlled with high accuracy, taking into account the actual measured value of the intake pipe pressure.

[0112] In throttle opening correction amount calculation unit 806, a throttle opening correction amount is calculated based on the difference between the intake pipe pressure estimate value updated by Kalman filter 420 and the target intake pipe pressure defined by the target torque and the target EGR rate, using equation (27) described later. By calculating the throttle opening correction amount using the intake pipe pressure estimate value updated by Kalman filter 420, torque can be controlled with high precision while taking into account the actual measured value of the intake pipe pressure.

[0113] Furthermore, the target charging efficiency correction amount can be calculated based on the difference between the fresh air charging efficiency (Equation (6)) taking into account the corrected EGR rate estimated value and the target charging efficiency output by the target charging efficiency calculation unit 502, and the throttle opening correction amount can be calculated based on the target charging efficiency correction amount.

[0114] [Control Model]

[0115] Next, functions used when constructing a control model that implements internal state feedback control will be described.

[0116] Figure 9This is a conceptual block diagram representing the functions used when building a control model. Inference 910 calculates output variables based on input variables, internal state variables, model constants, and the static and dynamic characteristics defined by these model constants. This allows inference of output behavior and internal state behavior corresponding to the inputs.

[0117] Inference 910 describes the controlled object as a forward problem. The control model, in contrast, solves an inverse problem. Specifically, a controller takes an output variable as a target value and outputs the input variable (controlled variable) needed to achieve that target value. To derive this controller, the following functional blocks are defined.

[0118] First, the input-output relationship of inference unit 910 is changed to construct observer 920. Observer 920 is constructed by setting input variables, output variables, and model constants to block inputs and outputting state variables. One method for implementing observer 920 is the Kalman filter used in this embodiment.

[0119] Furthermore, learner 930 can be constructed by changing the input-output relationship of inference unit 910. Learner 930 sets input variables and output variables as training data to the block input and outputs model constants. In this embodiment, learner 930 (system identification) is implemented using a recursive least squares algorithm.

[0120] [State space model of the state in the intake manifold]

[0121] Next, a method of describing the state in the intake pipe using a state space model is described. According to Euler's first-order forward difference, a discrete expression of the time differential of the state variable is defined as the following equation (11).

[0122] [Formula 11]

[0123]

[0124] Here, the subscript k above and to the right of x represents the current value when discretized on the time axis. Although first-order forward differencing is used here, the present invention is not limited to this. Discretizing equations (1) and (2) according to equation (11) allows them to be expressed as the state space model of equations (12) and (13).

[0125] [Formula 12]

[0126] x k+1 =Ax k +Bu k ····(12)

[0127] [Formula 13]

[0128] y k=Cx k ····(13)

[0129] Here, in equations (12) and (13), A, B, and C represented by Latin letters are matrices. k is the state vector, u k is the input vector, y k is the output vector, which is given by the following equations (14), (15), and (16) respectively.

[0130] [Formula 14]

[0131]

[0132] [Formula 15]

[0133]

[0134] [Formula 16]

[0135]

[0136] In addition, matrices A, B, and C are given by the following equations (17), (18), and (19), respectively.

[0137] [Formula 17]

[0138]

[0139] [Formula 18]

[0140]

[0141] [Formula 19]

[0142]

[0143] exist Figure 8 In the state space model setting unit 410, the state equation is sorted out using the state space model, and the matrices and vectors defined by equations (14) to (16) and equations (17) to (19) are handed over to the Kalman filter processing executed in the Kalman filter 420.

[0144] [Internal structure of Kalman filter and Kalman filter algorithm]

[0145] Next, refer to Figure 10 and Figure 11 , the internal structure of the Kalman filter 420 and the Kalman filter algorithm are explained.

[0146] Figure 10 4 is a block diagram showing an example of the internal structure of the Kalman filter 420. Figure 11is a flow chart representing an example of a Kalman filter algorithm executed in the Kalman filter 420 .

[0147] While the system of this embodiment employs a linear Kalman filter algorithm, the present invention is not limited thereto. In other words, similar effects can be achieved by employing nonlinear Kalman filters such as the extended Kalman filter or the ensemble Kalman filter. The following describes the algorithm and its application in this control system, specifically Kalman filter 420, a component of throttle control and EGR valve control.

[0148] Kalman filter 420 describes the state within the intake pipe, which is the target of control, using a state equation. Sensor measurement information is assigned to the output variables of the state equation. Kalman filter 420 then infers state variables representing the state within the intake pipe (hereinafter referred to as "internal state variables") that cannot be directly measured, based on the sensor measurement information assigned to the output variables.

[0149] When executing the Kalman filter 420, the control unit 23 of the ECU 21 determines whether the Kalman filter 420 can be executed (S1101). The sensor state and the prediction range of the state equation used as a prerequisite are considered as indicators for determining whether the Kalman filter 420 can be executed. For example, if the sensor state is determined through diagnosis to be unable to obtain sensor output due to a sensor failure or disconnection, or if sensor degradation has caused an unacceptable error in the sensor output, the Kalman filter 420 will not be executed because accurate filtering processing cannot be achieved. The state quantities used as the subject in the state equation are the intake manifold pressure and the intake manifold EGR rate. If any variable in the state equation exceeds the prediction range (the theoretical / physical range that can be taken in the state equation), that is, if it is within the area not covered by the state equation, the Kalman filter 420 will not be executed because accurate filtering processing cannot be achieved.

[0150] If the control unit 23 detects these conditions, it determines that the Kalman filter 420 cannot be executed (No in S1101) and ends the process of this flowchart. If it determines that the Kalman filter 420 can be executed (Yes in S1101), the process proceeds to step S1102. Furthermore, if the control unit 23 determines that the Kalman filter 420 cannot be executed, it may also set a fail-safe processing flag for the engine system.

[0151] The calculation formulas executed in steps S1102 to S1106 are specifically described below. In the Kalman filter 420, the state equation including the system noise Q and the observation noise R defined by the following equations (20) and (21) is assumed.

[0152] [Formula 20]

[0153] x(k+1)=Ax(k)+Bu(k)+Q····(20)

[0154] [Formula 21]

[0155] y(k)=Cx(k)+R ····(21)

[0156] Here, "k" means the current value of discrete time. The processing of Kalman filter 420 is divided into a prediction step and a filtering step. In the prediction step, the intake pipe state estimation unit 421 ( Figure 4 ) In the Kalman filter 420, the internal state variable vector x and the covariance matrix P (S1102, S1103) are calculated (updated) based on the input variables and the system noise Q using the following equations (22) and (23).

[0157] [Formula 22]

[0158] x(k|k-1)=Ax(k-1|k-1)+Bu(k)····(22)

[0159] [Formula 23]

[0160] P(k|k-1)=AP(k-1|k-1)A T +Q ····(23)

[0161] Next, in the filtering step, the state observer 422 ( Figure 4 ) The Kalman gain K defined by the updated covariance matrix P and the observation noise R is calculated using the following equation (24) (S1104). Furthermore, the state observer 422 may be configured to calculate the covariance matrix P in step S1103.

[0162] [Formula 24]

[0163]

[0164] Furthermore, the state observer 422 uses the Kalman gain K and the observation data (the detection value of the intake pipe pressure sensor 6) to update the internal state variable vector x and the covariance matrix P again (S1105, S1106) using the following equations (25) and (26). The Latin letter "I" in equation (26) represents the identity matrix. After the processing of steps 1105 and S1106 is completed, the processing sequence returns to step S1101. In this way, the internal state variable vector x and the covariance matrix P are corrected using the actual observation data y(k) (the detection value of the intake pipe pressure sensor 6).

[0165] [Formula 25]

[0166] x(k|k)=x(k|k-1)+K(k)(y(k)-c(k)x(k|k-1))····(25)

[0167] [Formula 26]

[0168] P(k|k)=[IK(k)C(k)]P(k|k-1)····(26)

[0169] The above calculations allow the behavior of the EGR rate, one of the internal state variables x(k|k) that is difficult to measure directly, to be inferred based on the output information of the intake pipe pressure sensor 6, which can be measured. In this embodiment, the information on the intake pipe state (intake pipe pressure, EGR rate) output from the Kalman filter 420 is used as training data for internal state feedback control and system identification (learning).

[0170] [Internal state feedback control]

[0171] Then, Figure 8 The following describes an internal state feedback control method performed in calculating the throttle opening correction amount by the throttle opening correction amount calculation unit 806 and the EGR valve opening correction amount by the EGR valve opening correction amount calculation unit 805. The internal state feedback control uses PID control composed of a proportional term, an integral term, and a differential term, but is not limited to this example.

[0172] (Throttle opening correction calculation)

[0173] The throttle opening correction amount calculation unit 806 calculates the throttle opening correction amount δφ based on the difference between the intake pipe pressure estimation value corrected (updated) by the state observer 422 of the Kalman filter 420 and the target intake pipe pressure specified by the target torque and the target EGR rate based on the following equation (27): th .

[0174] [Formula 27]

[0175]

[0176] Here, C P,t 、C I,t and C D,t is the PID control parameter. Figure 5 As shown, the control unit 23 controls the throttle opening by adding a throttle opening correction value to the throttle opening command value. By using the intake pipe pressure (estimated value) updated by the Kalman filter 420 to determine the throttle opening correction value, the torque can be controlled with high precision while taking into account the actual measured value of the intake pipe pressure.

[0177] (EGR valve opening correction calculation)

[0178] The EGR valve opening correction amount calculation unit 805 calculates the EGR valve opening correction amount δφ based on the difference between the EGR rate estimated value corrected (updated) by the state observer 422 of the Kalman filter 420 and the target EGR rate using the following equation (28): egr .

[0179] [Formula 28]

[0180]

[0181] Here, C p,e 、C I,e and C D,e is the PID control parameter. Figure 5 As shown, the control unit 23 controls the EGR valve opening by adding the EGR valve opening correction value to the EGR valve opening command value. By using the EGR rate (estimated value) updated by the Kalman filter 420 to calculate the EGR valve opening correction value, the EGR rate can be controlled with high precision, taking into account the actual value of the intake pipe pressure. Furthermore, while PID control is used here, the present invention is not limited to this. Similar effects can be achieved using any one of the proportional term, integral term, and differential term, or a combination of these terms.

[0182] [Study on throttle flow coefficient, EGR valve flow coefficient and cylinder intake efficiency]

[0183] Figure 12 The control block that performs learning of the throttle flow coefficient, EGR valve flow coefficient, and cylinder intake efficiency is shown.

[0184] The throttle flow coefficient calculation unit 1201 calculates the throttle flow coefficient based on the detection value of the airflow sensor 3, the throttle opening, the throttle upstream state, and the intake pipe pressure estimation value as the output of the Kalman filter 420. The throttle flow coefficient is input to the throttle flow coefficient system identification unit 1204. Here, the so-called throttle upstream state refers to the temperature and pressure of the intake air upstream of the throttle 4, that is, the atmospheric pressure. Figure 1 The description of the temperature sensor and pressure sensor for measuring the state upstream of the throttle valve is omitted.

[0185] The EGR valve flow coefficient calculation unit 1202 calculates the EGR valve flow coefficient based on the EGR valve opening, the EGR valve upstream state, the estimated EGR rate output from the Kalman filter 420, and the cylinder air intake efficiency calculated by the cylinder air intake efficiency calculation unit 1203. The EGR valve flow coefficient is input to the EGR valve flow coefficient system identification unit 1205. Here, the EGR valve flow rate is calculated based on the cylinder air intake efficiency, the engine speed, and the estimated EGR rate. The EGR valve flow coefficient is then calculated based on the EGR valve flow rate, the EGR valve opening, and the EGR valve upstream state.

[0186] Furthermore, the EGR valve flow coefficient calculation unit 1202 utilizes the cylinder intake efficiency learning results from the cylinder intake efficiency system identification unit 1206. The airflow sensor 3 detects the flow rate of intake air in the intake pipe 31, but this detection value is affected by the throttle valve 4. Therefore, as the state downstream of the throttle valve 4, the cylinder intake efficiency learning results output by the cylinder intake efficiency system identification unit 1206 are more reliable than the detection value of the airflow sensor 3. Therefore, by utilizing the cylinder intake efficiency learning results output by the cylinder intake efficiency system identification unit 1206 in the calculation of the EGR valve flow coefficient, the learning efficiency of the EGR valve flow coefficient is improved. Of course, the detection value of the airflow sensor 3 can also be used to calculate the EGR valve flow coefficient.

[0187] Cylinder air intake efficiency calculation unit 1203 calculates cylinder air intake efficiency based on the estimated intake pipe pressure and EGR rate values ​​output by Kalman filter 420, the detection value of airflow sensor 3, the engine speed, and the variable valve state. The cylinder air intake efficiency is input to cylinder air intake efficiency system identification unit 1206. The detection value of airflow sensor 3 is the value output by airflow sensor 3 when exhaust gas from EGR pipe 15 is not recirculating to intake pipe 31 (steady state).

[0188] The throttle flow coefficient system identification unit 1204 , the EGR valve flow coefficient system identification unit 1205 , and the cylinder intake efficiency system identification unit 1206 each correspond to the learner 450 .

[0189] The throttle flow coefficient system identification unit 1204 uses the current speed as a basis and uses the system identification algorithm described later (see Figure 17 ) sequentially learns the throttle flow coefficient calculated by the throttle flow coefficient calculation unit 1201. Specifically, the throttle flow coefficient system identification unit 1204 (learner) is configured to learn the relationship between the throttle flow coefficient, the detection value of the airflow sensor 3, and the throttle opening. The throttle flow coefficient is calculated based on the intake pipe pressure estimated value corrected by the state observer 422 (estimated value correction unit) of the Kalman filter 420.

[0190] Furthermore, the EGR valve flow coefficient system identification unit 1205 sequentially learns the EGR valve flow coefficient calculated by the EGR valve flow coefficient calculation unit 1202 based on the current engine speed using a system identification algorithm described later. Specifically, the EGR valve flow coefficient system identification unit 1205 (learner) is configured to learn the relationship between the EGR valve flow coefficient and the EGR valve opening. The EGR valve flow coefficient is calculated based on the EGR rate estimated value corrected by the state observer 422 (estimated value correction unit) of the Kalman filter 420.

[0191] Furthermore, cylinder air intake efficiency system identification unit 1206 sequentially learns the cylinder air intake efficiency calculated by cylinder air intake efficiency calculation unit 1203 based on the current engine speed using a system identification algorithm described later. Specifically, cylinder air intake efficiency system identification unit 1206 (learner) is configured to learn the relationship between cylinder air intake efficiency, engine speed, and variable valve state. The cylinder air intake efficiency is calculated based on the estimated intake pipe pressure and EGR rate values, as corrected by state observer 422 (estimated value correction unit) of Kalman filter 420, and the engine speed.

[0192] The learning results (model constants) of the throttle flow coefficient, the EGR valve flow coefficient, and the cylinder intake efficiency are input to the Kalman filter 420 ( Figure 4 ). As a result, the model constants (adjustment parameters) of the state space model used in the intake pipe state estimation unit 421 of the Kalman filter 420 are updated. The model constants are, for example, partial regression coefficients of a polynomial obtained by a recursive least squares algorithm.

[0193] This configuration enables sequential learning of the effects of temporal changes in flow characteristics caused by deposits on the throttle valve 4 or EGR valve 19, opening and closing phase deviations due to elongation of the intake valve timing chain, and other factors, allowing these effects to be appropriately reflected in internal state feedback control. Deposits are fuel and oil combustion products, namely oxides and carbides, that accumulate on combustion chamber walls and valve interiors. In this specification, deposits are simply referred to as "carbon deposits."

[0194] Abnormality diagnosis unit 1207 diagnoses normality / abnormality based on the learning results of the throttle flow coefficient, EGR valve flow coefficient, and cylinder intake efficiency. If any of the learned values ​​exceeds the threshold for determining an abnormal state, abnormality diagnosis unit 1207 determines an abnormal state, notifies the outside world by lighting warning indicator 22, and stores the abnormal value in the memory (storage unit 24) of ECU 21.

[0195] By comparing the throttle flow coefficient with a threshold, it is possible to diagnose whether throttle valve 4 has any abnormalities. For example, it is possible to determine whether the relationship between the corrected intake manifold pressure (the estimated intake manifold pressure value output by Kalman filter 420), the actual value measured by airflow sensor 3, and throttle opening deviates from the pre-defined relationship (whether it is within a normal range). Furthermore, based on the comparison of the EGR valve flow coefficient with a threshold, it is possible to diagnose whether the EGR valve 19 (EGR system) has any abnormalities. For example, it is possible to determine whether the relationship between the corrected EGR rate (the estimated EGR rate output by Kalman filter 420) and the EGR valve opening deviates from the pre-defined relationship.

[0196] Furthermore, abnormality diagnosis unit 1207 may also predict (predictive diagnosis) the period until the learned values ​​reach threshold values ​​set for determining an abnormal state based on the temporal changes (e.g., the amount of change per predetermined time) in the learned values ​​of the throttle flow coefficient, EGR valve flow coefficient, and cylinder intake efficiency output by learner 450. Abnormality diagnosis unit 1207 outputs the predicted period to warning indicator 22, etc., and stores the predicted period in storage unit 24. Furthermore, in the predictive diagnosis, if the learned value exceeds a second threshold value that is lower than the aforementioned threshold value (the first threshold value), the period until the learned value reaches the first threshold value may be predicted.

[0197] [Function approximation used in learning]

[0198] Then, Figure 12 The function approximation method used in learning the throttle flow coefficient, EGR valve flow coefficient, and cylinder intake efficiency shown in FIG.

[0199] The relationship between the output variable and the input variable is approximated by the following quadratic polynomial.

[0200] [Formula 29]

[0201]

[0202] Here, y is the output variable, x1 and x2 are input variables, and β0 to β5 are partial regression coefficients. By setting the square term and the interaction term, it is possible to consider the nonlinearity between the input variables for learning. Although the structure here is set to use a quadratic polynomial of two variables, it is not limited to this. Linear expressions, multidimensional polynomials of three or more dimensions, polynomials with higher-order terms of three or more degrees, polynomials with three-dimensional / quadratic interaction terms, and polynomials composed of these elements can also be used. Furthermore, the same effect can be achieved by using an approximation method based on the superposition of radial basis functions (radial basis function network approximation) or mapping, or updating of table values.

[0203] If the above polynomial is arranged using the partial regression coefficient vector θ and the input variable vector φ, it can be expressed as the following equation (30).

[0204] [Formula 30]

[0205]

[0206] For example, in learning the throttle flow coefficient, as shown in formula (31), the throttle flow coefficient μ is set for the output variable y. th , set the speed N for input variables x1 and x2 e and throttle opening φ th When carbon deposits are attached to each valve, the throttle flow coefficient μ th and EGR valve flow coefficient μ egr The value of becomes smaller. Thus, the throttle flow coefficient reflecting the current operating state can be learned. In addition, the input variable (operating state) is only an ideal example and is not limited to this example.

[0207] [Formula 31]

[0208]

[0209] For example, in learning the EGR valve flow coefficient, as shown in equation (32), the EGR valve flow coefficient μ is set for the output variable y. egr , set the speed N for input variables x1 and x2 e and EGR valve opening φ egr This allows learning of the EGR valve flow coefficient that reflects the current operating state. Note that the input variable (operating state) is merely an ideal example and is not limited to this example.

[0210] [Formula 32]

[0211]

[0212] For example, in the learning of cylinder intake efficiency, as shown in formula (33), the cylinder intake efficiency η is set for the output variable y in , set the speed N for input variables x1 and x2 e and intake manifold pressure p m This allows learning of the cylinder intake efficiency that reflects the current operating state. Note that the examples of input variables (operating state) are merely ideal examples and are not limited to these examples.

[0213] [Formula 33]

[0214]

[0215] The following shows a method for sequentially updating the partial regression coefficient vector θ based on the relationship between input and output. Figure 13 In equations (34) and (36), the symbol ^ is marked above θ to represent the partial regression coefficient vector.

[0216] [System Identification]

[0217] Figure 13 This is a flowchart showing an example of a recursive least squares algorithm for system identification performed by the learner 450 (the throttle flow coefficient system identification unit 1204 , the EGR valve flow coefficient system identification unit 1205 , and the cylinder intake efficiency system identification unit 1206 ).

[0218] When executing system identification (recursive least squares algorithm) using learner 450, control unit 23 of ECU 21 determines whether system identification can be performed (S1301). Similar to step S1101, the sensor state and the prediction range of the state equation used as the prerequisite are considered as indicators for determining whether system identification can be performed. If control unit 23 determines that system identification cannot be performed (No in S1301), the process of this flowchart ends. If it determines that system identification can be performed (Yes in S1301), the process proceeds to step S1302. The calculation formulas executed in steps S1302 to S1306 are described in detail below.

[0219] The intake system used as the object here is a time-varying system. In order to cope with it, a sequential identification algorithm with a variable forgetting element is adopted. The so-called forgetting element is a function that reduces the influence of past data exponentially according to the degree of oldness. By adopting the forgetting element, the influence of the latest state of the engine system can be properly considered on the partial regression coefficient vector (adjustment parameter). Furthermore, by setting it as variable forgetting, on the one hand, past data can be forgotten in a transient state, and on the other hand, the forgetting element can be made close to 1 in a steady state, thereby actively utilizing past data. The recursive least squares algorithm with a variable forgetting element is shown below. First, the learner 450 ( Figure 4 ) For each output variable, the difference between the polynomial and the output value is calculated as error ε(k) using the following equation (34) (S1302). Error ε(k) is the difference between the polynomial approximation and the actual value (the output value of each operation unit 1201 to 1203). Hereinafter, the partial regression coefficient vector θ(k) is updated so that this difference becomes zero (the symbol ^ is indicated above θ in the figure and the equation).

[0220] [Formula 34]

[0221]

[0222] Next, learner 450 uses the covariance matrix P(k-1), input vector φ(k), and forgetting factor λ(k) to calculate L(k) according to the following equation (35) (S1303). Then, learner 450 updates the partial regression coefficient vector θ(k) as needed based on L(k) and error ε(k) according to the following equation (36) (the symbol ^ is marked above θ in the figure and in the equation) (S1304).

[0223] [Formula 35]

[0224]

[0225] [Formula 36]

[0226]

[0227] At this time, the learner 450 calculates the forgetting factor λ(k) and the covariance matrix P(k) using the following equations (37) and (38), respectively (S1305, S1306).

[0228] [Formula 37]

[0229]

[0230] [Formula 38]

[0231]

[0232] Here, σ in equation (37) is the adjustment parameter for the forgetting factor λ(k) during learning. The Latin letter "I" in equation (38) represents the identity matrix, a matrix whose number of rows and columns is equal to the number of state variables (here, the intake pipe pressure and the intake pipe EGR rate). After completing steps 1305 and 1306, learner 450 returns to step S1301 and repeats the series of steps at predetermined intervals.

[0233] Furthermore, although the parameter identification algorithm of this embodiment adopts the recursive least squares algorithm, the present invention is not limited thereto. In other words, the use of other optimization methods such as gradient method and genetic algorithm as the parameter identification algorithm can also achieve the same or similar effects.

[0234] [Control of throttle opening and EGR valve opening and their effects]

[0235] Next, refer to Figures 14 to 16 , the control actions and effects of the throttle opening and EGR valve opening to achieve the target torque and target EGR rate are explained.

[0236] (Carbon deposits on the throttle valve, Kalman filter)

[0237] Figure 14The following describes a control operation and its effect when carbon deposits are attached to the throttle valve 4 in controlling the throttle opening and the EGR valve opening to achieve the target torque and the target EGR rate. Figure 14 The first to fourth layers of graphs respectively show the increase and decrease of torque (net mean effective pressure) and EGR rate and the changes of throttle opening and EGR valve opening relative to the increase and decrease of torque and EGR rate. Figure 15 and Figure 16 Same here.

[0238] In the absence of carbon deposits, when the target torque is increased and decreased like a rectangular wave (the first solid line) while the target EGR rate is fixed (the second solid line), the throttle opening and EGR valve opening increase as the target torque increases (the third solid line), and the EGR valve opening increases (the fourth solid line). On the other hand, when carbon deposits accumulate on the throttle valve 4, the flow of air through the throttle valve 4 is obstructed, reducing the amount of air. This, in turn, causes a relative increase in exhaust gas within the EGR pipe 15, increasing the EGR rate (the second dashed line).

[0239] In contrast, when internal state feedback based on Kalman filter 420, based on the values ​​measured by airflow sensor 3 and intake manifold pressure sensor 6, is implemented, the throttle opening and EGR valve opening are corrected toward the increasing side (the third-layer dashed line), allowing both torque and EGR rate to be controlled to target values ​​with high precision (the first and second-layer solid lines). Implementing internal state feedback based on Kalman filter 420 achieves robust control of torque and EGR rate to target values ​​with high precision, even in the presence of disturbances such as carbon deposits on throttle valve 4 within intake manifold 31.

[0240] (Carbon deposits on the throttle and EGR valve, Kalman filter)

[0241] Figure 15 This section demonstrates the control operation (based on internal state feedback from the Kalman filter 420) and its effects when carbon deposits are deposited on both the throttle valve 4 and the EGR valve 19 during the control of the throttle opening and EGR valve opening to achieve the target torque and target EGR rate. The section also demonstrates the transitions in the throttle opening and EGR valve opening when the target torque is increased and decreased like a rectangular wave while the target EGR rate is fixed.

[0242] When carbon deposits adhere to the throttle valve 4 and the EGR valve 19, the throttle opening and the EGR valve opening are corrected toward the increasing side (the single-dotted lines of the third and fourth layers) when the internal state feedback of the Kalman filter 420 is implemented based on the measurement values ​​of the airflow sensor 3 and the intake pipe pressure sensor 6, and the torque can be controlled to be near the target value (increasing side) (the dotted line of the first layer).

[0243] On the other hand, the EGR rate shifts significantly toward the decreasing side with an error (dashed line in the second layer). The internal state feedback of Kalman filter 420 cannot appropriately correct the EGR rate for the disturbance caused by carbon deposits on EGR valve 19. This is because the EGR valve flow coefficient in the EGR valve flow equation (Equation (4)), which is a premise of Kalman filter 420, differs from the actual value (when the influence of carbon deposits is not considered). Therefore, the EGR valve flow coefficient in the EGR valve flow equation must be updated to the actual value or a value approximately equal to the actual value.

[0244] (Carbon deposits on the throttle and EGR valve, Kalman filtering, and system identification)

[0245] Figure 16 This diagram illustrates the control operation (based on internal state feedback and system identification using Kalman filter 420) and its effects when carbon deposits are deposited on both the throttle valve 4 and the EGR valve 19 in controlling the throttle opening and EGR valve opening to achieve target torque and a target EGR rate. The diagram shows the transitions in throttle opening and EGR valve opening when the target torque is increased and decreased like a rectangular wave while the target EGR rate is fixed.

[0246] In the learner 450( Figure 12 The EGR valve flow coefficient system identification unit 1205 learns the changes in the EGR valve flow coefficient associated with carbon deposits deposited on the EGR valve 19 and reflects this information in the EGR valve flow rate equation (Equation (4)) in the Kalman filter 420. This improves the accuracy of the information on the intake pipe state (estimated intake pipe pressure and estimated EGR rate) fed back from the Kalman filter 420, enabling more accurate calculation of the EGR valve opening correction amount. Consequently, the EGR valve opening is appropriately corrected toward the increasing side (the single-dot chain line in the fourth layer), allowing both torque and EGR rate to be accurately controlled to their target values ​​(the solid lines in the first and second layers).

[0247] According to the above, when the internal state feedback based on the Kalman filter 420 and the learning based on the system identification are implemented, it is possible to achieve robustness in which the torque and the EGR rate can be controlled to the target values ​​with high accuracy even when there is a disturbance such as carbon deposits on both the throttle valve 4 and the EGR valve 19. On the other hand, by learning the change in the EGR valve flow coefficient associated with the carbon deposits on the throttle valve 4 and the EGR valve 19 and taking the learning result into account in the feedforward control calculations of the throttle opening and the EGR valve opening ( Figure 5 ), the control responsiveness can also be improved. In addition, by pre-determining the relationship between the valve opening correction amount and the carbon deposit amount, the carbon deposit amount can be estimated based on the valve opening correction amount.

[0248] [Throttle control and EGR valve control]

[0249] Figure 17 This is a flowchart showing an example of a procedure for executing throttle valve control and EGR valve control based on the detection value of the air flow sensor 3 and the detection value of the intake pipe pressure sensor 6.

[0250] First, the control unit 23 (target torque calculation unit 501) of the ECU 21 calculates the target torque of the internal combustion engine 1 based on the driver's accelerator pedal depression amount, the rotational speed of the internal combustion engine 1, the externally required torque, and other conditions (S1701).

[0251] Then, the control unit 23 (target EGR rate calculation unit 504) calculates the target EGR rate based on the rotational speed and target torque of the internal combustion engine 1 (S1702).

[0252] Then, the control unit 23 (target throttle opening calculation unit 503 ) calculates a feedforward control amount of the target throttle opening based on the target torque obtained in step S1701 ( S1703 ).

[0253] Then, the control unit 23 (target EGR valve opening calculation unit 505) calculates a feedforward control amount of the target EGR valve opening based on the target EGR rate obtained in step S1702 (S1704).

[0254] Then, the control unit 23 detects the amount of air sucked into the internal combustion engine 1 by the air flow sensor 3 (S1705). Then, the control unit 23 detects the pressure in the intake manifold 5 by the intake pipe pressure sensor 6 (S1706).

[0255] Then, the control unit 23 (state-space model setting unit 410) sets a state-space model (matrix, input / output / state vector) based on the physical model describing the state in the intake pipe (S1707).

[0256] Then, the control unit 23 determines whether the internal state feedback control based on the Kalman filter 420 can be executed based on the sensor state and the state in the intake pipe (S1708). The control unit 23 sets the execution permission flag to "1" (on) when it is determined to be executable, and sets the execution permission flag to "0" (off) when it is determined to be not executable. Then, when the execution permission flag is "1" (yes in S1708), the control unit 23 executes the processing of the Kalman filter 420 (the state in the intake pipe inference unit 421, the state observer 422) to infer the intake pipe pressure and EGR rate representing the state in the intake pipe (S1709). As described above, the processing of the Kalman filter 420 is executed based on the detection value of the airflow sensor 3, the detection value of the intake pipe pressure sensor 6, and the state space model (matrix, input / output / state vector) (refer to Figure 8 ).

[0257] On the other hand, if the execution permission flag is "0" (No in S1708), the control unit 23 executes the state-space model (intake pipe state estimation unit 421) to estimate the intake pipe pressure and EGR rate, which represent the state within the intake pipe. These estimated intake pipe pressure and EGR rate values ​​are values ​​that have not been corrected (corrected) by the state observer 422 of the Kalman filter 420.

[0258] Then, after the negative judgment of step S1708 or the processing of step S1709, the control unit 23 (valve correction amount calculation unit 440) corrects the throttle opening and the EGR valve opening based on the intake pipe pressure and the estimated value of the EGR rate indicating the state in the intake pipe (S1710, S1711). That is, the throttle opening correction amount calculation unit 806 calculates the throttle opening correction amount based on the estimated value of the intake pipe pressure and the target intake pipe pressure, and the EGR valve opening correction amount calculation unit 805 calculates the EGR valve opening correction amount based on the estimated value of the EGR rate and the target EGR rate ( Figure 8 ).

[0259] Then, the control unit 23 determines whether the system identification under the recursive least squares algorithm can be performed based on the sensor status and the status in the intake pipe (S1712). When the control unit 23 determines that the system identification can be performed (Yes in S1712), the system identification permission flag is set to "1" (on); when the control unit 23 determines that the system identification cannot be performed (No in S1712), the control unit 23 sets the system identification permission flag to "0" (off). Then, when the system identification permission flag is "1" (Yes in S1712), the control unit 23 performs system identification with the help of the learner 450 (S1713). Here, as the learner 450, the throttle flow coefficient system identification unit 1204, the EGR valve flow coefficient system identification unit 1205 and the cylinder intake efficiency system identification unit 1206 ( Figure 12) respectively perform system identification. As a result, the throttle flow coefficient, EGR valve flow coefficient, and cylinder intake efficiency are updated to their latest values. If the system identification permission flag is "0" (No in S1712), the control unit 23 does not perform system identification and transfers to step S1724.

[0260] Learner 450 (each system identification unit) determines whether system identification is complete based on the update status of the state-space model's adjustment parameters (the time variation of the adjustment parameters is below a specified value). If identification is not yet complete, the adjustment parameters change when the inputs and outputs change. Therefore, if the adjustment parameters change only slightly even when the inputs and outputs change, identification can be determined to be complete. Therefore, if learner 450 determines that system identification is not yet complete, it continues the system identification process. Then, when system identification is complete, learner 450 updates the state-space model's adjustment parameters (model constants).

[0261] Then, after the negative judgment of step S1712 or the processing of step S1713, the control unit 23 (abnormality diagnosis unit 1207) makes an abnormality judgment (S1714) on the throttle 4 and the EGR valve 19 (EGR system) based on whether the latest values ​​of the above-mentioned throttle flow coefficient, EGR valve flow coefficient and cylinder intake efficiency exceed the threshold values ​​set for each.

[0262] If the abnormality diagnosis unit 1207 determines that the throttle valve 4 or EGR valve 19 is abnormal (YES in S1714), the control unit 23 notifies the external device (e.g., the warning indicator 22 (MIL)) of the abnormality of the throttle valve 4 or EGR valve 19 as a diagnosis result. Alternatively, the control unit 23 notifies the external device (e.g., the warning indicator 22 (MIL)) that the latest value of the throttle flow coefficient, EGR valve flow coefficient, or cylinder intake efficiency is approaching a threshold value (a pre-diagnosis result) (S1715). The control unit 23 stores information related to the abnormal state in the storage unit 24 of the ECU 21.

[0263] Then, when it is determined in step S1714 that there is no abnormality in the throttle valve 4 or the EGR valve 19 (No in S1714) or after the processing of step S1715 is completed, the control unit 23 ends the processing of this flowchart.

[0264] By repeatedly executing steps S1701 to S1715 at set time intervals, it is possible to achieve robust control of the torque and EGR rate to target values ​​with high accuracy even when disturbances such as carbon deposits on both the throttle valve 4 and the EGR valve 19 occur.

[0265] As described above, the electronic control unit (ECU 21) of this embodiment is an electronic control unit for controlling an engine equipped with an EGR system including an EGR pipe (EGR pipe 15) for recirculating a portion of the exhaust gas from the internal combustion engine to the intake pipe, and an EGR valve (EGR valve 19) disposed within the EGR pipe; an airflow sensor (airflow sensor 3) for detecting the flow rate of air introduced into the intake pipe; a throttle valve (throttle valve 4) disposed downstream of the airflow sensor; and an intake pipe pressure sensor (intake pipe pressure sensor 6) disposed downstream of the throttle valve and downstream of the connection between the intake pipe and the EGR pipe, for detecting the pressure within the intake pipe downstream of the throttle valve, i.e., the intake pipe pressure. The electronic control unit (ECU 21) includes a state estimation unit (intake pipe state estimation unit 421) for determining the intake pipe pressure based on the detected value of the airflow sensor and the EGR valve opening (φ). egr ) to infer the intake manifold pressure (p m ) and EGR rate (ξ m ) and an estimated value correction unit (state observer 422) that corrects the EGR rate estimated value estimated by the state inference unit based on the detection value of the intake pipe pressure sensor and the intake pipe pressure estimated value estimated by the state inference unit.

[0266] Furthermore, in the present embodiment, a Kalman filter (Kalman filter 420 ) is applied at least in the estimated value correction unit (state observer 422 ).

[0267] According to the present embodiment constructed as described above, the estimated value correction unit corrects the EGR rate estimated value based on the intake pipe pressure sensor's detection value and the intake pipe pressure estimated value. This maintains high EGR control accuracy, preventing combustion defects such as engine knock and misfiring caused by EGR control errors. Furthermore, this prevents the target EGR rate from fluctuating downward, thereby improving fuel efficiency.

[0268] Furthermore, the present invention is not limited to the above-described embodiments. Various other applications and modifications are possible without departing from the spirit of the present invention as set forth in the claims. For example, the above-described embodiments provide detailed and specific descriptions of the electronic control device and engine control system for the purpose of easily understanding the present invention, and are not necessarily limited to including all of the described components. Furthermore, other components may be added to, substituted for, or deleted from some of the components of the above-described embodiments.

[0269] In the above embodiment, the present invention is applied to an engine system without a supercharger. However, the present invention is not limited to this example. For example, if a control model for an engine system equipped with a supercharger is created, the present invention can be applied to an engine system equipped with a supercharger.

[0270] Furthermore, the functions of the Kalman filter 420 and the learner 450 may be configured as a single unified function, or the Kalman filter 420 and the learner 450 may be integrated into one unit. Furthermore, the various components, functions, and processing units of the above-described embodiments may be partially or entirely implemented in hardware, for example, by designing them using integrated circuits. Examples of the hardware include FPGAs (Field Programmable Gate Arrays) and ASICs (Application Specific Integrated Circuits).

[0271] In addition, Figure 11 、 Figure 13 as well as Figure 17 In the flowchart shown, a plurality of processes may be executed in parallel or the order of the processes may be changed within a range that does not affect the process results.

[0272] Explanation of symbols

[0273] 1...Internal combustion engine, 3...Air flow sensor, 4...Throttle valve, 5...Intake manifold, 6...Intake pipe pressure sensor, 12...Fuel injection valve, 13...Spark plug, 15...EGR pipe, 17...EGR temperature sensor, 18...EGR valve upstream pressure sensor, 19...EGR valve, 21...ECU, 22...Warning indicator light, 23...Control unit, 24...Storage unit, 31...Intake pipe, 32...Exhaust pipe, 410...State space model setting unit, 420...Kalman filter, 421...State in intake pipe Inference unit, 422…state observer, 430…combustion control unit, 440…valve correction amount calculation unit, 450…learner (system identification), 803…EGR valve flow calculation unit, 804…cylinder intake flow calculation unit, 805…EGR valve opening correction amount calculation unit, 806…throttle opening correction amount calculation unit, 1204…throttle flow coefficient system identification unit, 1205…EGR valve flow coefficient identification unit, 1206…cylinder intake efficiency system identification unit, 1207…abnormal diagnosis unit.

Claims

1. An electronic control device for controlling an internal combustion engine, the internal combustion engine comprising: an EGR system including an EGR pipe for recirculating a portion of exhaust gas from the internal combustion engine to an intake pipe, and an EGR valve disposed in the EGR pipe; an air flow sensor for detecting a flow rate of air introduced into the intake pipe; a throttle valve disposed downstream of the air flow sensor; and an intake pipe pressure sensor disposed downstream of the throttle valve and in the intake manifold downstream of a connection between the intake pipe and the EGR pipe, for detecting a pressure in the intake manifold downstream of the throttle valve, i.e., an intake pipe pressure. The electronic control device is characterized by comprising: a state-space model representing an internal state of the intake pipe, the state-space model being set based on at least information about a detection value of the airflow sensor, a throttle opening, and an EGR valve opening; a state estimating unit that uses the state space model to estimate an estimated value of the intake pipe pressure and an estimated value of the EGR rate in the intake manifold based on the detection value of the airflow sensor, the throttle opening, and the EGR valve opening; an estimated value correction unit that utilizes a Kalman filter to update model constants of the state-space model based on a difference between a detection value of the intake pipe pressure sensor and the estimated intake pipe pressure value estimated by the state estimation unit, the estimated value correction unit inputting the detection value of the airflow sensor, the throttle opening, and the EGR valve opening into the state-space model in which the model constants have been updated by the Kalman filter to estimate a corrected estimated intake pipe pressure value and a corrected estimated EGR rate value; and The combustion control unit performs ignition timing control and / or fuel injection control of the engine using the corrected intake pipe pressure estimated value and the corrected EGR rate estimated value.

2. The electronic control device according to claim 1, characterized in that An EGR valve opening correction amount calculation unit is provided, wherein the EGR valve opening correction amount calculation unit calculates the correction amount of the EGR valve opening based on the corrected EGR rate estimated value and the target EGR rate. The EGR valve opening correction amount calculated in the EGR valve opening correction amount calculation unit is added to the EGR valve opening instruction value to control the EGR valve opening.

3. The electronic control device according to claim 2, characterized in that A learner is provided for learning the relationship between the EGR valve flow coefficient obtained based on the corrected EGR rate estimated value and the EGR valve opening; Information on the learned relationship between the EGR valve flow rate coefficient and the EGR valve opening is input to the state estimation unit, and model constants of the state space model used in the state estimation unit are updated.

4. The electronic control device according to claim 3, characterized in that The learner uses a recursive least squares algorithm for identifying a partial regression coefficient as the model constant included in a polynomial whose output variable is the EGR valve flow coefficient and whose input variable is the operating state of the internal combustion engine.

5. The electronic control device according to claim 4, characterized in that: The operating state of the internal combustion engine as the input variable is at least the rotational speed of the internal combustion engine and the EGR valve opening.

6. The electronic control device according to claim 1, characterized in that A throttle opening correction amount calculation unit is provided for calculating a correction amount for the throttle opening based on a difference between the corrected intake pipe pressure estimate value and a target intake pipe pressure defined by a target torque and a target EGR rate. The throttle opening is controlled by adding the throttle opening correction amount calculated by the throttle opening correction amount calculation unit to a throttle opening command value.

7. The electronic control device according to claim 6, characterized in that: A learner is provided for learning a relationship between a throttle flow coefficient obtained based on the corrected intake pipe pressure estimated value, a detection value of the air flow sensor, and a throttle opening; Information on the learned relationship between the throttle flow coefficient, the detection value of the airflow sensor, and the throttle opening is input to the state estimation unit, and model constants of the state space model used by the state estimation unit are updated.

8. The electronic control device according to claim 7, characterized in that: The learner uses a recursive least squares algorithm that identifies a partial regression coefficient as the model constant contained in a polynomial whose output variable is the throttle flow coefficient and whose input variable is the operating state of the internal combustion engine.

9. The electronic control device according to claim 8, characterized in that: The operating state of the internal combustion engine as the input variable is at least the rotational speed of the internal combustion engine and the throttle opening.

10. The electronic control device according to claim 3, characterized in that An abnormality diagnosis unit is provided that compares the EGR valve flow coefficient with a threshold value for determining an abnormal state and diagnoses abnormality of the EGR system based on the comparison result.

11. The electronic control device according to claim 7, characterized in that: An abnormality diagnosis unit is provided for comparing the throttle valve flow coefficient with a threshold value for determining an abnormal state and diagnosing an abnormality of the throttle valve based on the comparison result.

12. The electronic control device according to claim 3, characterized in that An abnormality diagnosis unit is provided for predicting a period until a learned value reaches an abnormal state based on a temporal change in the EGR valve flow rate coefficient and a threshold value for determining an abnormal state.

13. The electronic control device according to claim 7, characterized in that: An abnormality diagnosis unit is provided for predicting a period until a learned value reaches an abnormal state based on a temporal change in the throttle flow coefficient and a threshold value for determining an abnormal state.

14. An engine control system comprising an engine and an electronic control unit, the engine comprising: an EGR system including an EGR pipe for returning a portion of the exhaust gas of the internal combustion engine to an intake pipe, and an EGR valve disposed in the EGR pipe; an air flow sensor for detecting the flow rate of air introduced into the intake pipe; a throttle valve disposed downstream of the air flow sensor; and an intake pipe pressure sensor disposed downstream of the throttle valve and in the intake manifold downstream of a connection between the intake pipe and the EGR pipe, for detecting the pressure in the intake manifold downstream of the throttle valve, i.e., the intake pipe pressure. The engine control system is characterized in that: The electronic control device comprises: a state-space model representing an internal state of the intake pipe, the state-space model being set based on at least information about a detection value of the airflow sensor, a throttle opening, and an EGR valve opening; a state estimating unit that uses the state space model to estimate an estimated value of the intake pipe pressure and an estimated value of the EGR rate in the intake manifold based on a detection value of the airflow sensor, a throttle opening, and an EGR valve opening as information on the internal state of the intake pipe; and an estimated value correction unit that utilizes a Kalman filter to update model constants of the state-space model based on a difference between a detection value of the intake pipe pressure sensor and the estimated intake pipe pressure value estimated by the state estimation unit, the estimated value correction unit inputting the detection value of the airflow sensor, the throttle opening, and the EGR valve opening into the state-space model in which the model constants have been updated by the Kalman filter to estimate a corrected estimated intake pipe pressure value and a corrected estimated EGR rate value; and The combustion control unit performs ignition timing control and / or fuel injection control of the engine using the corrected intake pipe pressure estimated value and the corrected EGR rate estimated value.

Images