JP4605612B2 - Equipment that controls the plant using preview control - Google Patents

Description

  The present invention relates to an apparatus for controlling a plant robustly against disturbance using predictive control, and more specifically, to an engine robustly against disturbance applied to an intake pipe using predictive control. The present invention relates to an apparatus for controlling an intake air amount.

  Conventionally, in order to achieve a predetermined engine torque, the amount of intake air to the engine has been controlled. According to one method, the target intake air amount is obtained by referring to a predetermined map based on the accelerator pedal opening, the vehicle speed, and the transmission selection. The opening degree of the throttle valve is adjusted according to the target intake air amount.

According to another method, the target engine output shaft torque is calculated according to the detected accelerator pedal opening and the rotational speed of the torque converter output shaft. Based on the actual engine speed and the target engine output shaft torque, a predetermined table is referred to determine the target throttle valve opening. An air amount corresponding to the target throttle valve opening is sucked into the engine (see, for example, Patent Document 1).
Japanese Patent No. 2780345

  However, in the conventional engine torque control, the disturbance applied to the intake pipe and the dead time from the throttle valve to the engine are not considered. As a result, the amount of intake air to the engine cannot be controlled with sufficient accuracy, which may cause a phenomenon that the engine torque changes in a vibrational manner.

  Therefore, there is a need for intake air amount control that is robust against disturbance applied to the intake pipe. There is also a need for intake air amount control that takes into account the dead time from the throttle valve to the engine.

  According to one aspect of the present invention, a control device that controls a modeled plant includes an estimator that estimates disturbance applied to the plant, and a predictive control algorithm so that the output of the plant converges to a target value. And a control unit for calculating an input to the plant. The control unit calculates an input to the plant so as to include a value obtained by multiplying the disturbance estimated value estimated by the estimator by a predetermined gain. According to this invention, since the input to the plant is calculated so as to include the estimated disturbance value, it is possible to quickly converge the deviation between the output of the plant and the target value generated when the disturbance is applied to the plant. it can.

  According to one embodiment of the invention, the disturbance estimator is an adaptive disturbance observer that identifies disturbances using a sequential identification algorithm. By the sequential identification algorithm, the disturbance estimated value can be identified quickly and stably. Further, even when noise or the like is mixed in the output of the plant, it is possible to prevent the estimated disturbance value from fluctuating due to the noise or the like due to the statistical processing effect of the sequential identification algorithm.

  According to an embodiment of the present invention, the control unit further calculates an input to the plant so as to include a value obtained by multiplying a target value of the output of the plant by a predetermined gain. Therefore, it is possible to improve the characteristic that the output of the plant follows the target value.

  According to another aspect of the present invention, the control device further includes a state predictor that predicts a value output from the plant based on the dead time and disturbance estimation value of the plant. The control unit calculates an input to the plant using a predictive control algorithm so that the predicted value converges to the target value of the plant output.

  In the conventional generalized predictive control, the gain needs to be set low considering the dead time. According to the present invention, since the dead time of the plant is compensated by introducing the state predictor, such a reduction in gain is not required, and the speed of control can be improved.

  According to another aspect of the present invention, the plant is an intake pipe communicated with the engine. The intake pipe is modeled so that the input to the plant is the target value of the opening of the valve that controls the amount of air flowing into the intake pipe, and the output of the plant is the amount of air drawn into the engine. The According to the present invention, the intake air amount to the engine can be converged to the target value with high accuracy, and therefore the engine torque of the vehicle is well controlled. In one embodiment, the input to the plant is a target value of the opening of a throttle valve provided in the intake pipe.

  According to one embodiment of the present invention, the control device includes a storage device that stores model parameters based on the engine speed and the opening degree of the throttle valve for the modeled plant. The control unit extracts model parameters based on the detected engine speed and the detected opening of the throttle valve from the storage device. The input to the plant is calculated using the extracted model parameters. According to the present invention, the engine torque can be controlled with high accuracy and high response under various engine operating conditions.

  According to another aspect of the present invention, the target value of the plant output is a target intake air amount. When the engine is idling or when the transmission is shifting, the intake air amount necessary to achieve the target engine speed is set as the target intake air amount. When the vehicle is in a normal driving state, the intake air amount necessary for realizing the target engine torque of the engine is set as the target intake air amount. According to the present invention, it is possible to realize engine torque characteristics according to the operating state of the engine.

Configuration of Internal Combustion Engine and Controller Next, an embodiment of the present invention will be described with reference to the drawings. FIG. 1 is an overall system configuration diagram of an internal combustion engine (hereinafter referred to as “engine”) and a control device thereof according to an embodiment of the present invention.

  An electronic control unit (hereinafter referred to as “ECU”) 1 includes an input interface 1a that receives data sent from each part of the vehicle, a CPU 1b that executes calculations for controlling each part of the vehicle, and a read-only memory (ROM) ) And a random access memory (RAM) 1c and an output interface 1d for sending control signals to various parts of the vehicle. The ROM of the memory 1c stores a program for controlling each part of the vehicle and various data. The ROM may be a rewritable ROM such as an EEPROM. The RAM is provided with a work area for calculation by the CPU 1b. Data sent from each part of the vehicle and control signals sent to each part of the vehicle are temporarily stored in the RAM.

  The engine 2 is an engine having, for example, four cylinders. The engine 2 is provided with an intake valve 5 for communicating the combustion chamber 7 with the intake pipe 3 and an exhaust valve 6 for communicating the combustion chamber 7 with the exhaust pipe 4 for each cylinder.

  A throttle valve 8 is provided on the upstream side of the intake pipe 3. The throttle valve 8 is an electronic control valve, and its opening degree is controlled by a control signal from the ECU 1. A throttle valve opening sensor (θTH) 9 connected to the throttle valve 8 supplies an electric signal corresponding to the opening of the throttle valve 8 to the ECU 1.

  An air flow sensor (AFS) 10 is provided upstream of the throttle valve 8. The air flow sensor 10 detects the amount of air Gth passing through the throttle valve 8 and sends it to the ECU 1. The airflow sensor 10 can be a vane airflow sensor, a Karman vortex airflow sensor, a hot wire airflow sensor, or the like.

  The intake pipe pressure (Pb) sensor 11 is provided on the downstream side of the throttle valve 8 in the intake pipe 3. The intake pipe pressure Pb detected by the Pb sensor 11 is sent to the ECU 1.

  The fuel injection valve 12 is provided for each cylinder slightly upstream of the intake valve 5 in the intake pipe 3. The fuel injection valve 12 is supplied with fuel from a fuel tank (not shown) and is driven in accordance with a control signal from the ECU 1.

  The rotation speed (Ne) sensor 13 is attached around the cam shaft or crank shaft (both not shown) of the engine 2. The Ne sensor 13 outputs a CRK signal pulse at a cycle of a crank angle (for example, 30 degrees) shorter than a cycle of a TDC signal pulse output at a crank angle related to the TDC position of the piston, for example. The CRK signal pulse is counted by the ECU 5, and the engine speed Ne is detected.

  The signal sent to the ECU 1 is passed to the input interface 1a and converted from analog to digital. The CPU 1b processes the converted digital signal according to a program stored in the memory 1c to generate a control signal. The output interface 1d sends these control signals to the fuel injection valve 12 and other actuators.

  The air sucked into the intake pipe 3 is filled into the chamber 14 via the throttle valve 8. When the intake valve 5 is opened, the air filled in the chamber 14 is supplied to the combustion chamber 7 of the engine 2. Fuel is supplied through the fuel injection valve 12, and the air-fuel mixture is ignited in the combustion chamber 7 by an ignition device.

Overall Block Diagram of Control Device FIG . 2 shows a general block diagram of a control device according to one embodiment of the present invention. The engine torque setting unit 20, the rotation speed feedback (FB) control unit 21, the switch 22, and the intake air amount feedback (FB) control unit 23 are typically realized by a computer program.

  The engine torque setting unit 20 determines a target engine torque with reference to a map stored in advance in the memory 1c based on the accelerator pedal opening, the vehicle speed, the transmission gear ratio, and the like. The engine torque setting unit 20 calculates the intake air amount necessary for the target engine torque as the target intake air amount Gcyl_cmd.

  The rotational speed FB control unit 21 performs feedback control of the engine rotational speed NE. The control target (hereinafter referred to as a plant) of the rotational speed FB control is the engine 2. The control input is the target intake air amount Gcyl_cmd, and the control output is the engine speed NE. The rotational speed FB control unit 21 calculates the target intake air amount Gcyl_cmd so that the engine rotational speed NE converges to the target value.

  The intake air amount FB control unit 23 is connected to the engine torque setting unit 20 by the switch 22 when the vehicle is traveling normally. In this case, the intake air amount FB control unit 23 uses the target intake air amount Gcyl_cmd calculated by the engine torque setting unit 20. The intake air amount FB control unit 23 is connected to the rotational speed FB control unit 21 when the vehicle is idling or when the transmission of the vehicle is shifting. In this case, the intake air amount FB control unit 23 uses the target intake air amount Gcyl_cmd calculated by the rotation speed FB control unit 21.

  The intake air amount FB control unit 23 performs feedback control on the air amount Gcyl sucked into the engine cylinder. The plant is an intake pipe 3. The control input is a target value THcmd of the throttle opening, and the control output is an air amount Gcyl taken into the engine. The intake air amount FB control unit 23 calculates the target throttle opening THcmd so that the intake air amount Gcyl converges to the target value Gcyl_cmd. The throttle valve 8 is controlled by the ECU 1 according to the target throttle opening THcmd.

  As described above, when the vehicle is in the idling operation state or when the transmission is shifting, the intake air amount for converging the rotational speed NE to the target value is set as the target intake air amount. Therefore, engine stall during idling can be suppressed. Also, the rotational speed during transmission shifting can be converged to the target value stably and at high speed.

  In the present specification, the intake air amount FB control will be described first, and then the rotational speed FB control will be described.

1. Intake amount feedback control
1.1 Modeling of intake dynamic characteristics First, a method for modeling the intake dynamic characteristics of the intake pipe 3 will be described. The intake pipe 3 is represented by a model in which the input is THcmd and the output is Gcyl.

The intake air amount Gcyl ′ to each cylinder in each cycle can be expressed as in equation (1) from a known gas state equation. Kηc ′ is the intake manifold filling efficiency (%), Pb is the pressure in the intake pipe (Pa), Vcyl is the cylinder volume (m ³ ), Tcyl is the temperature in the cylinder (K), R Is a gas constant (m ³ · Pa / g · K), and n is an identifier for identifying a sampling cycle.

  In the case of an in-line four-cylinder engine, since intake is performed twice every time the engine rotates once, the amount of air Gcyl flowing into the cylinder per unit time is expressed as shown in Equation (2). NE is an engine speed (rpm), and k is an identifier for identifying a sampling cycle. Fcyl is a function of the rotational speed NE.

  On the other hand, the amount of air ΔGb to be filled in the chamber 14 is expressed by Expression (3).

For chamber 14, equation (5) is derived from the gas state equation (4). Pb, Vb, and Tb are the pressure (Pa), volume (m ³ ), and temperature (K) of the intake pipe, respectively. R is the gas constant described above.

  Substituting equation (5) into equation (3) yields equation (6). Expression (6) represents the intake air amount Gcyl as a function of the intake pipe pressure Pb. T represents a sampling period.

  In order to express Pb in Expression (6) using Gcyl, Expression (2) is substituted into Expression (6) to derive Expression (7). Expression (7) represents an intake dynamic characteristic model with Gth as an input.

  On the other hand, the relationship between the amount of air Gth passing through the throttle valve and the throttle opening TH can be expressed by equation (8). Pc is the pressure (Pc) upstream of the throttle valve. Fth represents the flow rate per effective opening of the throttle valve (g / deg), and is determined according to the pressure Pb downstream of the throttle valve (that is, the intake pipe pressure) and the pressure Pc upstream of the throttle valve. Substituting equation (8) into equation (7) yields equation (9). Expression (9) represents an intake dynamic characteristic model using the throttle valve opening TH as an input.

  Further, the relationship between the target throttle opening THcmd and the actual opening TH of the electronically controlled throttle valve can be approximated as shown in Expression (10). Equation (10) is a first-order lag system having a dead time dth. The dead time dth is mainly caused by electrical communication required to operate the throttle valve. Substituting equation (10) into equation (9) yields equation (11).

  From formula (9), TH (k-1) can be represented by Gcyl (k-1) and Gcyl (k-2). Substituting this TH (k-1) into equation (11) yields equation (12). Equation (12) is a model equation of the intake dynamic characteristic having the target throttle opening THcmd as an input and the intake air amount Gcyl as an output.

  The model parameters Aair1, Aair2, and Bair1 include values Fcyl and Fth that vary according to the rotational speed NE, the intake pipe pressure Pb, and the pressure Pc upstream of the throttle valve. Model parameters based on the rotational speed NE and the throttle opening TH can be stored in advance in the memory 1c as a map. Alternatively, the controller may be provided with an identifier to identify the model parameter.

1.2 Problems of Application of Generalized Predictive Control (GPC) In the present invention, intake amount feedback control is realized using a predictive control algorithm. Generalized predictive control (hereinafter referred to as GPC) is known as a control method similar to predictive control (GPC may be included in the predictive control category). However, it is impossible to configure a realizable intake air amount feedback control unit 23 simply by using the conventional GPC. The reason will be explained first.

  The intake dynamic characteristic model expressed by Expression (12) is expressed as Expression (13). Here, it is assumed that the value of the dead time dth is “2”.

  When Expression (13) is expressed in state space expression, Expression (14) is obtained.

A differential operator Δ defined as Δ = 1−Z ⁻¹ is introduced to define an expanded system represented by Expression (15). In the expansion system, a one-line component is added in order to derive an integral term that suppresses the steady-state deviation.

GPC is a control method for converging the control amount Gcyl to the target value Gcyl_cmd in the M section from the time point k to the time point (k + M). Therefore, an evaluation function J _G as shown in Expression (16) is defined. H is a weight parameter (> 0).

The control input ΔTHcmd that minimizes the evaluation function J _G can be obtained using the principle of optimality. The control input ΔTHcmd is expressed as Equation (17) using the solution P of the Riccati equation of Equation (18).

  By defining an initial condition such as the equation (19), P and D can be obtained sequentially.

  When M = 1 (when the target value of one step ahead is usable), Expression (16) is expressed as Expression (20), and Expression (17) is expressed as Expression (21).

  Specifically, the feedback coefficients of X ′ (k) and Gcyl_cmd (k + 1) in Expression (21) are calculated.

  As described above, by executing the conventional GPC, if the first row component of the G and G ′ vectors is zero due to the existence of the dead time, the executable intake air amount FB control unit 23 can be configured. Can not.

1.3 Configuration of Intake Amount FB Control Unit FIG. 3 shows a configuration of the intake amount FB control unit 23 according to an embodiment of the present invention. In the present invention, the intake air amount feedback control using the predictive control is realized by configuring the intake air amount FB control unit 23 in this way.

  The intake air amount FB control unit 23 includes an adaptive disturbance observer 31, a state predictor 32, and a control unit 33. The adaptive disturbance observer 31 identifies an estimated value γ1 of the disturbance applied to the intake pipe 3. The state predictor 32 calculates the predicted value Pre_Gcyl of the output of the intake pipe 3 that is a plant using the disturbance estimated value γ1. The control unit 33 calculates a target throttle opening THcmd, which is a control input to the plant, by a predictive control algorithm using the predicted value Pre_Gcyl. The control input THcmd includes a value obtained by multiplying the estimated disturbance γ1 by a predetermined gain. By converging the predicted value Pre_Gcyl to the target value, the output Gcyl of the plant can be converged to the target value.

  The problem that the first row component of the G and G ′ vectors is zeroed can be prevented by introducing the state predictor 32. The state predictor 32 will be described.

  Since the value necessary to compensate for the dead time dth caused by the electronically controlled throttle valve is Gcyl (k + dth-1), the intake dynamic characteristic model equation (13) is set to the future by (dth-1) steps. shift.

  Since the equation (23) includes future values Gcyl (k + dth-2) and Gcyl (k + dth-3) that cannot be observed, these future values are deleted. Erasing can be obtained by recursive calculation as shown below. Expression (24) represents a prediction expression of the intake air amount Gcyl.

  GPC is a control theory that uses the principle of optimality, but is not designed to take into account modeling errors and prediction errors, and thus is not sufficiently robust. Therefore, the prediction equation (24) includes a modeling error and a disturbance estimated value γ1 for compensating for the prediction error. Expression (25) is an expression for obtaining the predicted value Pre_Gcyl, which is executed by the state predictor 32.

  By calculating the predicted value by the state predictor 32, the dead time is compensated, and hence the quick response of the intake air amount control can be improved. Further, by including the estimated disturbance value γ1 in the predicted value, it is possible to eliminate the steady deviation between the output Gcyl of the intake pipe being controlled and the predicted value Pre_Gcyl.

  The disturbance estimated value γ1 is identified by the adaptive disturbance observer 31. Expression (26) is an expression for obtaining the estimated disturbance value γ1 executed by the adaptive disturbance observer 31.

  As is apparent from the equation (26), the adaptive disturbance observer 31 calculates the predicted value Gcyl_hat (k) for the current cycle (the calculation method is the same as the prediction equation (25)). Further, the adaptive disturbance observer 31 calculates a deviation e_dov between the predicted value Gcyl_hat (k) and the actually detected value Gcyl (k). Thereafter, the disturbance estimated value γ1 is calculated using the sequential identification algorithm so as to eliminate the deviation e_dov. By using the sequential identification algorithm, the estimated disturbance value can be identified quickly and stably. Further, even when noise is mixed in the output of the intake pipe 3 that is the control target, it is possible to prevent the estimated disturbance value from fluctuating due to the noise due to the statistical processing effect of the sequential identification algorithm.

λ ₁ and λ ₂ are weight parameters. For lambda ₁ = 1 and lambda ₂ = 1, least squares method, lambda ₁ <1 and lambda ₂ = 1 Weighted least square method when a fixed gain method in the case of lambda ₁ = 1 and lambda ₂ = 0 , Λ ₁ 1 and λ ₂ <1, the gradually decreasing gain method is used.

  Next, the control unit 33 will be described. When the prediction formula (25) is shifted to the future by one step and further converted to include the future value, the formula (27) is obtained. The conversion including the future value may be performed by reversing the conversion process from Expression (23) to Expression (24).

A differential operator Δ defined as Δ = 1−Z ⁻¹ is introduced to define an expansion system as shown in Equation (28). Egc is a deviation between the actual intake air amount Gcyl and the target value Gcyl_cmd. The reason why the one-line component is added in the expansion system is to derive an integral term that suppresses the steady-state deviation. Further, it is assumed that the disturbance fluctuation is constant (ie, Δγ1 (k) = Δγ1 (k + 1)).

Define the evaluation function _JSG . When the target prediction stage number is represented by Nr and the disturbance prediction stage number is represented by Nd, the evaluation function _JSG is defined using N = Max (Nr, Nd). The target prediction stage number Nr is the same as that of the section M described above, and defines which section of the future value is used for the target value Gcyl_cmd. The number of disturbance prediction stages Nd defines which section of the future value is used for the estimated disturbance value γ1 calculated by the adaptive disturbance observer 31. In this example, Nd is zero and Nr is one. Therefore, N is 1. The evaluation function J _SG is shown in Expression (29).

The control input ΔTHcmd that minimizes the evaluation function J _SG can be obtained using the principle of optimality. Using the solution of the Riccati equation represented by equation (31), the control input ΔTHcmd is represented as equation (30).

  Equation (30) is solved based on the initial condition of equation (32). If N = 1 (ie, Nr = 1 and Nd = 0), equation (33) is obtained.

  Specifically, the feedback coefficients Fx and Fd and the feedforward coefficient Fr in Expression (33) are calculated.

  The control input ΔTHcmd when N = 1 can be calculated based on the equations (33) and (34).

  Expression (35) is an expression for calculating the difference ΔTHcmd. The control input THcmd is calculated by integrating the equation (35).

  If the initial values of Gcyl (0 + dth-1) to Gcyl (0), Gcyl_cmd (0 + dth) to Gcyl_cmd (0), γ1 (0), and THcmd (0) are zero, Equation (36) 37).

  Equation (37) includes future values Gcyl (k + dth-1) and Gcyl (k + dth-2) that are unobservable at the present time k. Instead of these values, the predicted values Pre_Gcyl (k) and Pre_Gcyl (k−1) calculated by the state predictor 32 are used. Expression (38) is an expression executed by the control unit 33. In this way, the control input THcmd (k) is generated by the control unit 33.

  Since the control input THcmd includes the feedback term of the disturbance estimated value γ1, the deviation between the intake air amount Gcyl and the target value Gcyl_cmd generated by applying the disturbance can be quickly converged. Further, since the control input THcmd includes the target value feedforward term Gcyl_cmd (k + dth), the follow-up speed of the intake air amount Gcyl with respect to the target value Gcyl_cmd can be improved.

1.4 Simulation Result of Intake Air Volume FB Control FIG. 4 shows a virtual control target model used in the simulation of the intake air volume FB control according to an embodiment of the present invention. The virtual control target has a structure based on the model formula (13). The control input is a target throttle opening THcmd (k-dth) delayed by time dth. The control output is the intake air amount Gcyl (k). The intake air amount Gcyl (k-1) one cycle before and the intake air amount Gcyl (k-2) two cycles before are fed back.

  The simulation is configured so that three disturbances can be applied to the virtual control target. FIG. 4 shows an input disturbance d1, a state quantity disturbance d2, and an output disturbance d3. The input disturbance d1 includes, for example, variations in the behavior of the throttle valve. The state quantity disturbance d2 includes, for example, a modeling error. The output disturbance d3 includes, for example, sensor noise.

Table 1 shows the conditions of cases G-1 to G-5 implemented in the simulation.

  In case G-1, no disturbance is applied. In the state predictor 32 and the control unit 33, the estimated value Pre_Gcyl and the target throttle opening THcmd are respectively calculated using the estimated disturbance value γ1. FIG. 5 shows a simulation result in case G-1. Since there is no disturbance, there is no deviation between the predicted value Pre_Gcyl and the actual intake air amount Gcyl. The control unit 33 can cause the intake air amount Gcyl to follow the target value Gcyl_cmd without causing overshoot.

  In case G-2, disturbances d1 to d3 are added, and the estimated disturbance value γ1 is not used in both the state predictor 32 and the control unit 33. FIG. 6 shows a simulation result in case G-2. Due to the disturbance, a steady deviation occurs between the predicted value Pre_Gcyl and the actual intake air amount Gcyl. Since the control unit 33 calculates the control input THcmd based on the predicted value Pre_Gcyl, the intake air amount Gcyl cannot be converged to the target value Gcyl_cmd.

  In case G-3, disturbances d1 to d3 are added, and the estimated disturbance value γ1 is used in the predictor 32, but the estimated disturbance value γ1 is not used in the control unit 33. FIG. 7 shows a simulation result in case G-3. The steady-state deviation between Pre_Gcyl and Gcyl caused by disturbance is eliminated by the predictor 32. Therefore, the controller 33 can converge the intake air amount Gcyl to the target value Gcyl_cmd. However, since the feedback term based on the disturbance estimated value γ1 is not included in the control input THcmd, the convergence speed is relatively slow.

  In case G-4, disturbances d1 to d3 are added, and the estimated disturbance value γ1 is used in both the predictor 32 and the control unit 33. Case G-4 corresponds to the preferred embodiment of the present invention described with reference to FIG. FIG. 8 shows a simulation result in case G-4. As is clear from FIG. 7, the time required for the deviation between the intake air amount Gcyl and the target value Gcyl_cmd to converge is dramatically shortened.

  In case G-5, disturbances d1 to d3 are added, and the estimated disturbance value γ1 is used in both the predictor 32 and the control unit 33. However, the control input THcmd does not include the target value feedforward term Gcyl_cmd (k + dth). FIG. 9 shows a simulation result in case G-5. As apparent from the comparison with FIG. 8, the follow-up speed of the intake air amount Gcyl with respect to the target value Gcyl_cmd is slow. This is because the term of the deviation (Gcyl and Gcyl_cmd) included in the control input THcmd is only the integral term (Pre_Egc term). In this way, by including the target value feedforward term in the control input, it is possible to improve the followability of the intake air amount Gcyl to the target value Gcyl_cmd.

  Now consider the case where the control model has no dead time. In this case, the state predictor 32 may not be included, and the intake dynamic characteristic model for controlling the intake air amount Gcyl can be expressed as in Expression (39).

  Since there is no dead time, Expression (26) executed by the adaptive disturbance observer 31 is expressed by Expression (40).

  Since there is no dead time, the equation (38) executed by the control unit 33 is expressed by the equation (41).

A simulation as shown in Table 2 was performed for a case where such a model does not include a dead time.

  In case G-6, disturbances d1 to d3 are applied, and the control unit 33 calculates the target throttle opening THcmd using the estimated disturbance value γ1. The control input THcmd includes a target value feedforward term Gcyl_cmd (k + dth). FIG. 10 shows a simulation result in case G-6.

  In case G-7, disturbances d1 to d3 are added, and the control unit 33 calculates the target throttle opening THcmd without using the disturbance estimated value γ1. The control input THcmd includes a target value feedforward term Gcyl_cmd (k + dth). FIG. 11 shows a simulation result in case G-7.

  In FIG. 12, the behavior of Gcyl in FIG. 10 (Case G-6) is compared with the behavior of Gcyl in FIG. 11 (Case G-7). It can be seen that the former has better convergence than the latter. As described above, the control unit 33 calculates the control input THcmd using the estimated disturbance value γ1, thereby improving the convergence of the control amount Gcyl with respect to the target value Gcyl_cmd.

  In the embodiment described above, an adaptive disturbance observer using a sequential identification algorithm is used to estimate the disturbance. Alternatively, other suitable estimators that estimate the disturbance by referring to a predetermined map or the like may be used.

  In the above-described embodiment, a throttle valve is used as a valve for controlling the intake air amount. Alternatively, another valve that controls the intake air amount, such as a bypass valve, may be used.

2. Speed feedback control
2.1 Engine Modeling First, a method for modeling the engine 2 will be described. The engine 2 is represented by a model in which the input is the intake air amount Gcyl and the output is the rotational speed NE.

An equation of motion of the inertia system of the engine is expressed by Equation (42). Ieng is the moment of inertia of the engine (kgm ² ), Kne is the friction coefficient of the engine, and NE is the engine speed (rad / sec). Teng is the engine torque (Nm), Tload is the equipment drive torque (Nm) for driving electrical components such as air conditioners and generators mounted on the vehicle, and Tdrv is the vehicle drive system. This is the torque (Nm) that is allocated to drive the vehicle. t indicates time.

  The engine torque Teng is expressed as in equation (43). Ktrq is a torque coefficient, and is determined according to the engine speed NE, the engine ignition timing IG, and the equivalence ratio (reciprocal of the air-fuel ratio) λ. Gcyl is the amount of air (g) taken into the engine.

  Substituting equation (43) into equation (42) yields equation (44). Expression (44) represents a first-order lag system of the rotational speed NE with the intake air amount Gcyl as an input. “-(Tload + Tdrv) / Ieng” is added as a disturbance term.

  Equation (44) is converted to a discrete time system to obtain Equation (45). T indicates a sampling period, and k is an identifier for identifying the sampling cycle. Expression (45) is a model expression of the engine inertia system.

  The model parameters Ane, Bne, and Cne vary according to the rotational speed NE and the throttle opening TH. Model parameters based on the rotational speed NE and the throttle opening TH can be stored in advance in the memory 1c as a map. Alternatively, the controller may be provided with an identifier to identify the model parameter.

2.2 Configuration of Rotational Speed FB Control Unit FIG. 13 shows a block diagram of the rotational speed FB control unit 21 according to an embodiment of the present invention. The rotational speed FB control unit 21 includes an adaptive disturbance observer 41, a state predictor 42, and a control unit 43. The adaptive disturbance observer 41 and the state predictor 42 have the same configuration as that shown in FIG. 3 for the intake air amount FB control unit 21.

The adaptive disturbance observer 41 identifies an estimated value δne of disturbance applied to the engine 2. The state predictor 42 calculates a predicted value Pre_NE of the output of the engine that is the plant (that is, the engine speed) based on the estimated disturbance value δne. The control unit 43 calculates a target intake air amount Gcyl_cmd, which is a control input to the plant, by response specifying control using the predicted value Pre_NE. The control input Gcyl_cmd includes a value obtained by multiplying the estimated disturbance Δne by a predetermined gain. By converging the predicted value Pre_NE to the target value, the output NE of the plant can be converged to the target value. First, the state predictor 42 will be described. As described with reference to FIG. 2, the rotational speed FB control unit 21 is positioned upstream of the intake air amount FB control unit 23. The intake dynamic characteristics of the intake pipe 3 include a dead time. If this is compensated by both the rotational speed FB control and the intake air amount FB control, interference occurs. Accordingly, the dead time of the intake pipe 3 is compensated by the intake air amount FB control unit 23, and the intake pipe 3 is regarded as a dead time element from the rotational speed FB control unit 21. As a result, it can be seen from the rotation speed FB control unit 21 that Gcyl_cmd (k-dth) = Gcyl (k). In other words, from the rotational speed FB control unit 21, it appears that the intake air amount Gcyl is sucked into the engine after the dead time dth from the calculation of Gcyl_cmd. Therefore, the model expression of the engine inertia system of Expression (45) can be expressed as Expression (46). Here, the disturbance Td indicates the sum of Tload and Tdrv.

  In order to compensate for the dead time dth, it is necessary to predict the control output NE (k + dth). Therefore, the equation (46) is shifted to the future by (dth-1) steps.

  Since equation (47) includes future values NE (k + dth-1) and Td (k + dth-1) that cannot be observed, these future values are deleted. The erasure can be performed in the same manner as the erasure of the future value from the equation (23) described above.

  Since Td (k + dth-1) to Td (k) in equation (48) vary depending on the driving operation and driving conditions of the driver, it is difficult to predict them. Therefore, it is assumed that the disturbance Td is constant as shown in the equation (49). Based on this assumption, Expression (48) is expressed by Expression (50).

  The disturbance estimated value δne is introduced into the equation (50). The disturbance estimated value Δne includes not only the estimation error of the disturbance Td but also other disturbances applied to the plant. Expression (51) is an expression for calculating the predicted value Pre_NE of the rotation speed, and is executed by the state predictor 42.

  By calculating the predicted value by the state predictor 42, the dead time is compensated, and therefore the speed response of the rotation speed control can be improved. Further, since the predicted value Pre_NE is calculated based on the estimated disturbance value Δne, it is possible to eliminate the steady deviation between the output NE of the engine that is the control target and the predicted value Pre_NE.

  The disturbance estimated value δne is identified by the adaptive disturbance observer 41. Expression (52) is an expression for obtaining the estimated disturbance value δne executed by the adaptive disturbance observer 41.

  As apparent from the equation (52), the adaptive disturbance observer 41 calculates an estimated value NE_hat (k) for the current cycle (this is calculated by shifting the prediction equation (51) by dth steps in the past). ) Here, it is assumed that the estimated disturbance value δne is constant. That is, Δne (k−dth) = Δne (k−1). Further, the adaptive disturbance observer 41 calculates a deviation e_dne between the predicted value NE_hat (k) and the actually detected value NE (k). Thereafter, the disturbance estimated value δne is calculated using a sequential identification algorithm so as to eliminate the deviation e_dne.

By using the sequential identification algorithm, the estimated disturbance value δne can be estimated quickly and stably. As described above, λ ₁ and λ ₂ are weight parameters and are determined according to the type of the sequential identification algorithm.

  Next, the control unit 43 will be described. When the prediction formula (51) is shifted to the future by one step and further converted to include the future value, the formula (53) is obtained. The conversion including the future value may be performed by reversing the conversion from Expression (47) to Expression (48). Here, it is assumed that the fluctuation of the future value of the disturbance Td and the disturbance estimated value Δne is constant. That is, Td (k + dth) = Td (k) and Δne (k + dth) = Δne (k).

  A switching function σne is defined in order to implement response assignment type control. The convergence function with respect to the target value NE_cmd of the actual rotational speed NE can be defined by the switching function σne. E_ne is a deviation between the actual rotational speed NE and the target value NE_cmd.

  The control input is determined so that the switching function σne becomes zero.

  Equation (55) shows a first-order lag system without input. That is, the control unit 43 operates so as to constrain the control amount E_ne to the first-order lag system represented by Expression (55).

  FIG. 14 shows a phase plane with E_ne (k) on the vertical axis and E_ne (k−1) on the horizontal axis. In the phase plane, a switching line 61 expressed by Expression (55) is shown. Assuming that the point 62 is an initial value of the state quantity (E_ne (k-1), E_ne (k)), the control unit 43 places the state quantity on the switching line 61 and restrains it on the switching line 61. . In this way, the state quantity is constrained to a first-order lag system with no input. Therefore, the state quantity is automatically set to the origin of the phase plane (ie, E_ne (k), E_ne (k-1) = 0) over time. Converge to. By constraining the state quantity on the switching line 61, the state quantity can be converged to the origin without being affected by disturbance.

  The setting parameter S_ne in Expression (55) is set so as to satisfy −1 <S_ne <1. Preferably, it is set to satisfy -1 <S_ne <0. This is because if S_ne has a positive value, the first-order lag system of Equation (55) becomes a vibration stable system.

  The setting parameter S_ne is a parameter that defines the convergence speed of the deviation E_ne. In FIG. 15, convergence speeds when S_ne = −1, −0.8, and −0.5 are shown by graphs 63, 64, and 65, respectively. As the absolute value of the setting parameter S_ne decreases, the convergence speed of the deviation E_ne increases.

  The control unit 43 obtains the control input Upas according to the equation (56). The equivalent control input Ueq is an input for constraining the state quantity on the switching line. The reaching law input Urch is an input for placing the state quantity on the switching line.

  A method for obtaining the equivalent control input Ueq will be described. The equivalent control input Ueq has a function of holding the state quantity at an arbitrary place on the phase plane. Therefore, it is necessary to satisfy Expression (57).

  From equation (54), equation (57) is represented by equation (58).

  Substituting equation (53) into equation (58) yields equation (59).

  The control input Ueq (k) is calculated by the equation (60).

  Equation (60) includes future values NE (k + dth) and NE (k + dth-1) that are unobservable at the present time k. Instead of these values, the prediction values Pre_NE (k) and Pre_NE (k−1) calculated by the state predictor 42 are used. The control unit 43 executes the equation (61) to generate the equivalent control input Ueq (k).

  Thus, the equivalent control input Ueq includes the disturbance feedback term Δne and the disturbance feedforward term Td. Therefore, the deviation between the rotational speed NE and the target value NE_cmd generated by applying a disturbance to the engine 2 to be controlled can be quickly converged.

  The control unit 43 further executes Expression (62) to generate the reaching law input Urch. F represents a reaching law gain.

2.3 Simulation Result of Rotational Speed FB Control FIG. 16 shows a virtual controlled object model used in the simulation of the rotational speed FB control according to an embodiment of the present invention. The virtual control target has a structure based on the engine model formula (46). The control input is a target intake air amount Gcyl_cmd delayed by time dth. The control output is the rotational speed NE. Further, the driving torque Td is input as a disturbance to the controlled object. The rotation speed NE one cycle before is fed back.

  The simulation is configured so that three disturbances can be applied to the virtual control target. In the figure, places where the input disturbance L1, the state quantity disturbance L2, and the output disturbance L3 are applied are shown. The input disturbance L1 includes, for example, an estimation error Td_error of the drive torque Td. The state quantity disturbance L2 includes, for example, a modeling error. The output disturbance L3 includes, for example, sensor noise.

Table 3 shows the conditions of Case N-1 to Case N-5 implemented in the simulation.

  In case N-1, disturbances L1 to L3 are applied. Both the predictor 42 and the control unit 43 use the estimated disturbance value δne and the driving torque Td. Case N-1 is a preferable case based on the rotational speed FB control of the present invention in FIG. FIG. 17 shows the simulation result of case N-1. With the disturbance applied, the rotational speed NE can be converged to the target value NE_cmd without a steady deviation. When the target value NE_cmd fluctuates, the following characteristic of the rotational speed NE to the target value NE_cmd is also good.

  In case N-2, the predictor 42 and the control unit 43 do not use the estimated disturbance value δne and the drive torque Td. FIG. 18 shows a simulation result of case N-2. Due to the disturbance, a steady deviation occurs between the actual rotational speed NE and the predicted value Pre_NE. Since the control unit 43 performs response specifying control based on the predicted value Pre_NE, the actual rotational speed NE cannot be converged to the target value NE_cmd.

  In case N-3, the predictor 42 and the control unit 43 use the drive torque Td, but do not use the estimated disturbance value δne. FIG. 19 shows the simulation result of case N-3. At time t1, an estimated error Td_error of the drive torque Td is input as a step. In response to this, the estimated disturbance value δne also increases. Since the feedforward term of the driving torque Td is used, the deviation between the actual rotational speed NE and the predicted value Pre_NE does not change.

  At time t2, the drive torque Td varies. At this time, other disturbances L2 and L3 are also applied to the plant. Even if the drive torque Td varies due to the feedforward term of the drive torque Td, the deviation between the actual rotational speed NE and the target value NE_cmd does not change. However, since the estimated disturbance value δne is not used, the steady deviation between the actual rotational speed NE and the predicted value Pre_NE is not eliminated, and therefore the rotational speed NE cannot be converged to the target value NE_cmd.

  Case N-4 shows a case where the estimated disturbance δne and the drive torque Td are not used in the predictor 42 and the control unit 43, but the adaptive law input Uadp is added to the control input in the control unit 43. The adaptive law input Uadp is expressed by Equation (63). G represents the gain of the adaptive law input.

  FIG. 20 shows a simulation result of case N-4. The predicted value Pre_NE converges to the target value NE_cmd by the control unit 43. However, since the predicted value Pre_NE is not calculated based on the estimated disturbance value δne, the steady-state deviation between the actual rotational speed NE and the predicted value Pre_NE is not eliminated, and therefore the rotational speed NE cannot be converged to the target value NE_cmd.

  In case N-5, the predictor 42 uses the drive torque Td and the estimated disturbance value δne. In the control unit 43, the driving torque Td is used, but the disturbance estimated value δne is not used. Instead, in the control unit 43, the adaptive law input Uadp is added to the control input.

  FIG. 21 shows a simulation result of case N-5. The convergence time of the rotational speed NE can be shortened by increasing the gain G of the adaptive law input Uadp. However, as apparent from the comparison with FIG. 17, when the gain G is increased, an integral overshoot occurs.

  Now consider the case where the control model has no dead time. In this case, the state predictor 42 may not be included, and the model of the engine inertia system for controlling the engine speed NE can be expressed as in Expression (64).

  Since there is no dead time, Expression (52) executed by the adaptive disturbance observer 41 is expressed by Expression (65).

  Since there is no dead time, equations (61) and (62) executed by the control unit 43 are expressed by equations (66) and (67).

Simulations as shown in Table 4 were performed for cases N-6 to N-9 in which the model does not include time delay characteristics.

  In case N-6, disturbances L1 to L3 are applied. In the control unit 43, the estimated disturbance value δne and the driving torque Td are used. Case N-6 is a preferable case based on the rotational speed FB control of the present invention as described above. FIG. 22 shows the simulation result of case N-6. With the disturbance applied, the rotational speed NE can be converged to the target value NE_cmd without a steady deviation. When the target value NE_cmd fluctuates, the following characteristic of the rotational speed NE to the target value NE_cmd is also good.

  In case N-7, the driving torque Td is used in the control unit 43, but the estimated disturbance value δne is not used. FIG. 23 shows the simulation result of case N-7. It can be seen that the deviation between the actual rotational speed NE and the target value NE_cmd increases each time a disturbance is applied. Since the estimated disturbance value δne is not used, the actual rotational speed NE cannot be converged to the target value NE_cmd.

  Case N-8 shows a case where the disturbance estimation value δne is not used in the control unit 43 but the adaptive law input Uadp is added to the control input. In case N-8, a gain G having a relatively small value is used. FIG. 24 shows the simulation result of case N-8. In case N-9, the control unit 43 uses a gain G having a relatively large value. FIG. 25 shows the simulation result of case N-9. FIG. 26 is a diagram comparing the behavior of the rotational speed NE of FIG. 24 (case N-8) with the behavior of the rotational speed NE of FIG. 25 (case N-9). It can be seen that by increasing the gain G of the adaptive law input Uadp, the convergence time of the rotational speed NE can be shortened, but an integral overshoot occurs. On the other hand, when the disturbance estimated value δne corresponding to case N-6 is used, the rotational speed NE can be converged to the target value NE_cmd without causing an integral overshoot.

  In the embodiment described above, an adaptive disturbance observer using a sequential identification algorithm is used to estimate the disturbance. Alternatively, other suitable estimators that obtain disturbances by referring to a predetermined map or the like may be used.

3 Operation Flow FIG. 27 shows a flowchart of the rotational speed FB control and the intake air amount FB control according to the embodiment of the present invention shown in FIG. This flowchart can be applied to any vehicle of manual transmission (MT), automatic manual transmission (automatic MT), and automatic transmission (AT).

  In step S1, it is determined whether the vehicle is idling or whether the transmission is being changed. If the determination in step S1 is Yes, the process proceeds to step S2 in order to carry out the rotation speed FB control described above.

  In step S2, a target engine speed NE_cmd is obtained. For example, when the vehicle is in the idling operation state, a value corresponding to the running condition, the warm-up state, etc. is set as the target value NE_cmd. When the gear is being changed, a value corresponding to the vehicle speed and the gear ratio is set as the target value NE_cmd.

  In step S3, model parameters Ane and Bne are extracted with reference to the map stored in the memory 1c of the ECU 1 based on the detected actual rotational speed NE. In step S4, the device driving torque Tload and the vehicle driving torque Tdrv are obtained. The device driving torque Tload can be calculated in accordance with the on / off state of the electrical component mounted on the vehicle. Further, the vehicle driving torque Tdrv can be calculated according to the running resistance, the state of the clutch, and the like. Tload and Tdrv are added to calculate the driving torque Td as a disturbance. In step S5, the above-described rotation speed FB control is performed to calculate the target intake air amount Gcyl_cmd.

  On the other hand, when the vehicle is in normal running, the process proceeds to step S6, and the target engine torque is obtained. The target engine torque can be calculated according to the accelerator pedal opening, the vehicle speed, the transmission gear ratio, the traveling environment, and the like. In step S7, an engine intake air amount Gcyl_cmd necessary for realizing the target engine torque is calculated. For example, the target intake air amount Gcyl_cmd can be obtained by referring to a predetermined map based on the air-fuel ratio and the ignition timing.

  In step S8, the intake air amount Gcyl is estimated. The intake air amount Gcyl can be estimated based on output values from the air flow meter 10 and the Pb sensor 11. In step S9, model parameters Aair1, Aair2 and Bair1 are extracted with reference to a predetermined map based on the engine speed NE and the throttle opening TH. Instead of the throttle opening, the air amount Gth passing through the throttle or the intake air amount Gcyl calculated in step S8 may be used.

  In step S10, the above-described intake air amount FB control is performed to calculate the target throttle opening THcmd.

  The present invention can also be applied to a marine vessel propulsion engine such as an outboard motor having a vertical crankshaft.

1 schematically shows an internal combustion engine and a control device according to one embodiment of the present invention. FIG. The figure which shows the outline of the control apparatus according to one Example of this invention. The figure which shows the structure of intake air amount feedback control according to one Example of this invention. The figure which shows the virtual control object in the simulation of intake air amount feedback control according to one Example of this invention. The figure which shows the result of case G-1 of the intake air amount control simulation according to one Example of this invention. The figure which shows the result of case G-2 of the intake air amount control simulation according to one Example of this invention. The figure which shows the result of case G-3 of the intake air amount control simulation according to one Example of this invention. The figure which shows the result of case G-4 of the intake air amount control simulation according to one Example of this invention. The figure which shows the result of case G-5 of the intake air amount control simulation according to one Example of this invention. The figure which shows the result of case G-6 of the intake air amount control simulation according to one Example of this invention. The figure which shows the result of case G-7 of the intake air amount control simulation according to one Example of this invention. The figure for comparing case G-7 and case G-8 of the intake air amount control simulation according to one embodiment of the present invention. The figure which shows the structure of rotation speed feedback control according to one Example of this invention. The figure which shows the switching line of response designation | designated type control according to one Example of this invention. The figure which shows the relationship between the value of the setting parameter of the switching function of response designation | designated type control, and convergence speed according to one Example of this invention. The figure which shows the virtual control object in the simulation of rotation speed feedback control according to one Example of this invention. The figure which shows the result of case N-1 of the rotation speed control simulation according to one Example of this invention. The figure which shows the result of case N-2 of the rotation speed control simulation according to one Example of this invention. The figure which shows the result of case N-3 of the rotation speed control simulation according to one Example of this invention. The figure which shows the result of case N-4 of the rotation speed control simulation according to one Example of this invention. The figure which shows the result of case N-5 of the rotation speed control simulation according to one Example of this invention. The figure which shows the result of case N-6 of the rotation speed control simulation according to one Example of this invention. The figure which shows the result of case N-7 of the rotation speed control simulation according to one Example of this invention. The figure which shows the result of case N-8 of the rotation speed control simulation according to one Example of this invention. The figure which shows the result of case N-9 of the rotation speed control simulation according to one Example of this invention. The figure for comparing case N-8 and case N-9 of rotation speed control simulation according to one example of this invention. The flowchart of rotation speed feedback control and intake air amount feedback control according to one embodiment of the present invention.

Explanation of symbols

1 ECU
2 Engine 3 Intake pipe

Claims (7)

A control device for controlling a modeled plant,
A disturbance estimator that estimates a disturbance applied to the plant and calculates a disturbance estimated value, calculates an output of the model using inputs and outputs to the plant, and outputs the calculated model and actual A disturbance estimator for calculating the disturbance estimated value based on an integral value of the deviation from the detected plant output ;
A state predictor that predicts a future value of the output of the plant using the disturbance estimated value calculated by the disturbance estimator;
An input that configures the input to the plant to minimize the square value of the deviation between the future value predicted by the state predictor using a predictive control algorithm and the future value of the target value of the plant output A control unit that sequentially calculates the gain of the term and calculates the input to the plant ,
The control unit further calculates an input to the plant so as to include a value obtained by multiplying the disturbance estimated value by a predetermined gain.
Control device.   The control device according to claim 1, wherein the disturbance estimator is an adaptive disturbance observer that identifies the disturbance using a sequential identification algorithm.   The control device according to claim 1, wherein the control unit calculates an input to the plant so as to include a value obtained by multiplying a target value of the output of the plant by a predetermined gain. The plant is an intake pipe communicated with an engine,
In the intake pipe, the input to the plant is a target value of the opening of a valve that controls the amount of air flowing into the intake pipe, and the output of the plant is the amount of air taken into the engine The control device according to claim 1, which is modeled as follows.   The control device according to claim 4, wherein the valve that controls the amount of air sucked into the intake pipe is a throttle valve provided in the intake pipe. The modeled plant further comprises a storage device for storing model parameters based on the engine speed and the throttle valve opening,
The control unit extracts a model parameter based on the detected engine speed and the detected opening of the throttle valve from the storage device, and calculates an input to the plant using the extracted model parameter.
The control device according to claim 4 or 5. The target value of the plant output is a target intake air amount,
The control unit further comprises:
When the engine is idling or when the transmission is shifting, the intake air amount required to achieve the target engine speed is set to the target intake air amount;
When the engine is in a normal operating state, the intake air amount necessary to achieve the target engine torque of the engine is set to the target intake air amount;
The control device according to claim 4.

Images