JP3233564B2 - Fuel injection control device for internal combustion engine - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a fuel injection control device for an internal combustion engine, and more specifically, in a fuel injection control using adaptive control, a gain matrix of an adaptive parameter is determined in accordance with an operating state, thereby controlling the control. It is related to one that improves the performance.

[0002]

2. Description of the Related Art In recent years, adaptive control theory has been introduced also for internal combustion engines, and a technique has been proposed for adaptively controlling the amount of fuel actually taken into a cylinder so as to match a target fuel amount. For example, a technique disclosed in JP-A-2-173,334 can be mentioned.

[0003]

The present applicant has also proposed a fuel injection control for an internal combustion engine using adaptive control in Japanese Patent Application No. 4-215665. By the way, in the above-described related art, in the calculation of the gain matrix for determining the convergence speed of the adaptive parameter, whether to optimize the gain matrix is not considered.

Accordingly, an object of the present invention is to solve the above-mentioned problems, and to determine the gain matrix for determining the convergence speed of the adaptive parameter in accordance with the operating state, thereby optimizing the convergence speed of the adaptive parameter and thereby improving the controllability. It is an object of the present invention to provide a fuel injection control device for an internal combustion engine which improves the fuel injection control.

[0005]

In order to achieve the above object, according to the present invention, there is provided a fuel injection amount control means for controlling a fuel injection amount of a multi-cylinder internal combustion engine; The detected air-fuel ratio and the adaptive
Beauty actual intake fuel amount of the adaptive controller for at least controlling the amount consisting of any <br/> target value to an Itasa, and at least the air-fuel ratio based on adaptive parameters used in the adaptive controller
The fuel injection control apparatus for a multi-cylinder internal combustion engine and an adaptive parameter-adjusting mechanism calculates Zui, the adaptive path
The gain matrix for determining the convergence speed of the adaptive parameter calculated by the parameter adjustment mechanism is determined according to the operating state of the multi-cylinder internal combustion engine.

According to a second aspect of the present invention, a gain matrix for determining a convergence speed of the adaptive parameter is determined according to a load state of the multi-cylinder internal combustion engine.

According to a third aspect of the present invention, the gain matrix for determining the convergence speed of the adaptive parameter is made different depending on whether or not the multi-cylinder internal combustion engine is in an idle operation state.

According to a fourth aspect of the present invention, the multi-cylinder internal combustion engine is provided with an exhaust gas recirculation mechanism, and a gain matrix for determining a convergence speed of the adaptive parameter is made different depending on whether or not exhaust gas recirculation is performed. did.

According to a fifth aspect of the present invention, the multi-cylinder internal combustion engine includes a canister-purge mechanism, and a gain matrix for determining a convergence speed of the adaptive parameter is made different from when the canister-purge is executed. It was configured as follows.

According to a sixth aspect of the present invention, the operating state is:
It was configured to be at atmospheric pressure.

In a preferred embodiment, the operating condition is
The engine cooling water temperature was configured.

In a preferred embodiment, the multi-cylinder internal combustion engine includes a variable valve timing mechanism, and the operating state is a valve timing of the variable valve timing mechanism.

[0013]

According to the first aspect of the present invention, since the gain matrix for determining the convergence speed of the adaptive parameter is determined according to the operating state of the multi-cylinder internal combustion engine, the convergence of the adaptive parameter is determined according to the operating state. Speed can be optimized, and controllability can be improved.

According to a second aspect of the present invention, the gain matrix for determining the convergence speed of the adaptive parameter is determined according to the load state of the multi-cylinder internal combustion engine. Even when the engine load suddenly changes, such as when the engine is decelerated, the convergence speed of the adaptive parameters can be optimized, and the controllability can be improved.

According to a third aspect of the present invention, the gain matrix for determining the convergence speed of the adaptive parameter is made different depending on whether or not the multi-cylinder internal combustion engine is in an idle operation state. In the above, the convergence speed of the adaptive parameter can be optimized, so that the controllability can be improved.

According to a fourth aspect of the present invention, the multi-cylinder internal combustion engine is provided with an exhaust gas recirculation mechanism, and a gain matrix for determining a convergence speed of the adaptive parameter is made different depending on whether or not exhaust gas recirculation is performed. Therefore, particularly in an operation state in which exhaust gas recirculation is performed, the convergence speed of the adaptive parameter can be optimized, and the controllability can be improved.

According to a fifth aspect of the present invention, the multi-cylinder internal combustion engine includes a canister-purge mechanism, and a gain matrix for determining a convergence speed of the adaptive parameter is made different from when the canister-purge is executed. With such a configuration, the convergence speed of the adaptive parameter can be optimized especially in an operation state in which canister purging is performed, and thus controllability can be improved.

According to a sixth aspect of the present invention, the operating state is:
Since the pressure is set to be the atmospheric pressure, the convergence speed of the adaptive parameter can be optimized especially in an operation state in which the atmospheric pressure is largely different from the normal pressure, and the controllability can be improved.

According to a seventh aspect of the present invention, the operating state is:
Since the configuration is such that the engine cooling water temperature is maintained, the convergence speed of the adaptive parameter can be optimized especially in an operation state in which the water temperature is significantly different from the normal temperature, so that the controllability can be improved.

According to the present invention, the multi-cylinder internal combustion engine is provided with a variable valve timing mechanism, and the operating state is the valve timing of the variable valve timing mechanism. However, the convergence speed of the adaptive parameter can be optimized, and the controllability can be improved.

[0021]

Embodiments of the present invention will be described below with reference to the accompanying drawings.

FIG. 1 is an overall view schematically showing a fuel injection control device for an internal combustion engine according to this application.

In the drawing, reference numeral 10 denotes an OHC in-line four-cylinder internal combustion engine. The intake air introduced from an air cleaner 14 disposed at the end of an intake pipe 12 is subjected to a surge while its flow rate is adjusted by a throttle valve 16. The gas flows into the first to fourth cylinders via two intake valves (not shown) via the tank 18 and the intake manifold 20. An injector 22 is provided near an intake valve (not shown) of each cylinder to inject fuel. The air-fuel mixture that has been injected and integrated with the intake air is first, third, and third spark plugs (not shown) in each cylinder.
The fourth and second cylinders are ignited in order and burn to drive a piston (not shown).

Exhaust gas after combustion is discharged to an exhaust manifold 24 through two exhaust valves (not shown), and is purified by a catalyst device (three-way catalyst) 28 through an exhaust pipe 26 to outside the engine. Is discharged. As described above, the throttle valve 16 is mechanically separated from the accelerator pedal (not shown), and is controlled via the pulse motor M to an opening degree corresponding to the depression amount of the accelerator pedal and the operating state. The intake pipe 12 is provided with a bypass passage 32 near the position of the throttle valve 16 for bypassing the throttle valve 16.

Here, the internal combustion engine 10 is provided with an exhaust gas recirculation mechanism 100 for recirculating exhaust gas to the intake side.

Referring to FIG. 2, the exhaust gas recirculation path 121 of the exhaust gas recirculation mechanism 100 has one end 121a connected to the exhaust pipe 2.
The other end 121b communicates with the downstream side of the throttle valve 16 (not shown in FIG. 2) of the intake pipe 12 on the upstream side of the first catalyst device 28 (not shown in FIG. 2). This exhaust gas recirculation path 1
In the middle of 21, an exhaust gas recirculation valve (recirculation gas control valve) 122 for adjusting the amount of exhaust gas recirculation and a volume chamber 121c are provided. The exhaust gas recirculation valve 122 is an electromagnetic valve having a solenoid 122a. The solenoid 122a is connected to a control unit (ECU) 34, which will be described later.
The valve opening is changed linearly by the output from the controller.
The exhaust gas recirculation valve 122 is provided with a lift sensor 123 for detecting the opening degree of the valve, and the output is sent to the control unit 34.

Further, the connection between the intake system of the internal combustion engine 10 and the fuel tank 36 is established, and the canister-purge mechanism 200 is connected.
Is provided.

As shown in FIG. 3, the canister-purge mechanism 200 includes an upper portion of the sealed fuel tank 36 and the intake pipe 12.
Of the steam supply passage 221 and the adsorbent 231, which is formed between the downstream side of the throttle valve 16 and the canister 2.
23, and a purge passage 224. Steam supply passage 2
21, a two-way valve 222 is mounted, a purge control valve 225 is provided in the middle of the purge passage 224, a flow meter 226 for detecting a flow rate of a mixture containing fuel vapor flowing through the purge passage 224, Is provided with an HC concentration sensor 227 for detecting the HC concentration of the liquid crystal. The purge control valve (electromagnetic valve) 225 is connected to the control unit 34 as described later, and is controlled according to a signal from the control unit 34 to linearly change the valve opening amount.

According to this canister purge mechanism, the fuel vapor (fuel vapor) generated in the fuel tank 36 is
When a predetermined set amount is reached, the positive pressure valve of the two-way valve 222 is pushed open, flows into the canister 223, and adsorbent 2
Adsorbed by 31 and stored. When the purge control valve 225 is opened by an opening amount corresponding to the duty ratio of the on / off control signal from the control unit 34, the canister 22
The evaporative fuel temporarily stored in 3 is sucked into the intake pipe 12 through the purge control valve 225 together with the outside air sucked from the outside air intake port 232 by the negative pressure in the suction pipe 12 and sent to each cylinder. When the fuel tank 36 is cooled by the outside air and the negative pressure in the fuel tank increases, the two-way valve 22
The second negative pressure valve is opened, and the evaporated fuel temporarily stored in the canister 223 is returned to the fuel tank 36.

Further, the internal combustion engine 10 has a so-called variable valve timing mechanism 300 (shown as V / T in FIG. 1). The variable valve timing mechanism 300 is described in, for example, Japanese Patent Application Laid-Open No. 2-275043. FIG. 4 shows the valve timing V / T of the engine according to the operating state such as the engine speed Ne and the intake pressure Pb. Switching between two timing characteristics LoV / T and HiV / T. However, the mechanism itself is a well-known mechanism, and thus the description is omitted. Note that the switching of the valve timing characteristics includes an operation of stopping one of the two intake valves.

In FIG. 1, a crank angle sensor 40 for detecting a crank angle position of a piston (not shown) is provided in a distributor (not shown) of the internal combustion engine 10 and an opening of the throttle valve 16 is detected. A throttle opening sensor 42 and an absolute pressure sensor 44 for detecting the intake pressure Pb downstream of the throttle valve 16 as an absolute pressure are also provided. At an appropriate position of the internal combustion engine 10, an atmospheric pressure sensor 46 for detecting the atmospheric pressure Pa is provided.
Temperature sensor 4 for detecting the temperature of the intake air
8 is provided, and a water temperature sensor 50 for detecting an engine cooling water temperature is provided at an appropriate position of the engine. In addition, a valve timing (V / T) for detecting a valve timing characteristic selected by the variable valve timing mechanism 300 via a hydraulic pressure.
) A sensor 52 (not shown in FIG. 1) is also provided. Further, in the exhaust system, a wide-area air-fuel ratio sensor 54 is provided in an exhaust system collecting portion downstream of the exhaust manifold 24 and upstream of the catalyst device 28. These sensor outputs are sent to the control unit 34.

FIG. 5 is a block diagram showing details of the control unit 34. The output of the wide-range air-fuel ratio sensor 54 is input to a detection circuit 62, where appropriate linearization processing is performed to output a detection signal having a linear characteristic proportional to the oxygen concentration in the exhaust gas in a wide range from lean to rich. (Hereinafter, this wide-range air-fuel ratio sensor is referred to as a “LAF sensor”).

The output of the detection circuit 62 is
6 and input via the A / D conversion circuit 68 into the CPU. CPU is CPU core 70, ROM 72, RAM 7
4, the output of the detection circuit 62 is more specifically A / D-converted every predetermined crank angle (for example, 15 degrees).
The data is sequentially stored in one of the buffers in the RAM 74. 12
As shown in FIG. 53, 0 buffers to 11 buffers
No. up to Is appended. Similarly, the output of an analog sensor such as the throttle opening sensor 42 is also taken into the CPU via the multiplexer 66 and the A / D conversion circuit 68, and stored in the RAM 74.

The output of the crank angle sensor 40 is shaped by a waveform shaping circuit 76, the output value is counted by a counter 78, and the count value is input to the CPU.
In the CPU, the CPU core 70 calculates a control value according to a command stored in the ROM 72 as described later, and drives the injector 22 of each cylinder via the drive circuit 82. Further, the CPU core 70 is connected to the solenoid valve 90 (the bypass passage 32 for adjusting the secondary air amount) through the drive circuits 84, 86, 88.
Opening and closing), and the above-described exhaust recirculation control solenoid valve 122.
And drives the canister purge control solenoid valve 225. In FIG. 5, illustration of the lift sensor 123, the flow meter 226, and the HC concentration sensor 227 is omitted.

FIG. 6 is a flowchart showing the operation of the control device according to the present application.

First, the engine speed Ne and the intake pressure Pb detected in S10 are read out, and the program proceeds to S12 to judge whether or not cranking is performed. If not, the program proceeds to S14 to determine whether or not fuel cut is performed. to decide. The fuel cut is performed in a predetermined operating state, for example, when the throttle valve opening is at the fully closed position and the engine speed is equal to or higher than a predetermined value, the fuel supply is stopped, and the injection amount is controlled in an open loop. You.

When it is determined in S14 that the fuel cut is not performed, the process proceeds to S16, in which a map is retrieved from the detected engine speed Ne and the intake pressure Pb to determine the basic fuel injection amount T.
Calculate im. Then, the process proceeds to S18, where the LAF sensor 54
It is determined whether or not activation has been completed. This is, for example, L
This is performed by comparing the difference between the output voltage of the AF sensor 54 and its center voltage with a predetermined value (for example, 0.4 V), and determining that activation has been completed when the difference is smaller than the predetermined value. If it is determined that the activation has been completed, the process proceeds to S20, and it is determined whether or not the activation is in the feedback control region. When the operating state changes due to high rotation, full opening increase, or high water temperature, the injection amount is controlled in an open loop. When it is determined in S20 that the region is the feedback control region, the process proceeds to S22,
The detected value of the LAF sensor is read, and the routine proceeds to S24, where a detected air-fuel ratio KACT (k) (k: sampling time in a discrete system; the same applies hereinafter) is obtained from the detected value. Then, the program proceeds to S26, in which a feedback correction coefficient KLAF (k) based on the PID control law is calculated.

The feedback correction coefficient KLAF based on the PID control law is calculated as follows.

First, the control deviation DKAF between the target air-fuel ratio KCMD and the detected air-fuel ratio KACT is determined as DKAF (k) = KCMD (k-d ')-KACT (k). In the above, KCMD (k-d '): target air-fuel ratio (where d' indicates a dead time until KCMD is reflected in KACT,
Therefore, it means the target air-fuel ratio before the dead time control cycle),
KACT (k): Indicates the detected air-fuel ratio (of the current control cycle). still,
In this specification, both the target value KCMD and the detected value KACT are actually shown as equivalent ratios, that is, Mst / M = 1 / λ (Mst:
Theoretical air-fuel ratio, M = A / F (A: air consumption, F: fuel consumption, λ: excess air ratio)).

Next, the P term KL is multiplied by a predetermined coefficient.
AFP (k), I-term KLAFI (k), and D-term KLAFD (k) are converted to P-term: KLAFP (k) = DKAF (k) × KP I-term: KLAFI (k) = KLAFI (k−1) + DKAF ( k) × KI D term: KLAFD (k) = (DKAF (k) −DKAF (k-1)) × KD.

As described above, the P term is obtained by multiplying the deviation by the proportional gain KP, and the I term is obtained by adding the value obtained by multiplying the deviation by the integral gain KI to the previous value KLAFI (k-1) of the feedback correction coefficient. The D term is the deviation DKAF (k) and the previous value DKAF (k-1)
Multiplied by the differential gain KD. In addition, each gain KP, K
I and KD are obtained in accordance with the engine speed and the engine load. More specifically, the engine speed Ne and the intake pressure P are determined using a map.
b is set so that it can be searched. Finally, the value thus obtained is added to KLAF (k) = KLAFP (k) + KLAFI (k) + KLAFD (k) to obtain the present value KLAF (k) of the feedback correction coefficient based on the PID control law. In this case, in order to obtain a feedback correction coefficient by multiplication correction, it is assumed that the offset value of 1.0 is included in the I term KLAFI (k) (that is, the initial value of KLAFI (k) is 1. 0).

In the flow chart of FIG.
Proceeding to 28, feedback correction coefficient by adaptive control law
Calculate KSTR (k). The feedback correction coefficient KSTR (k) based on the adaptive control law will be described later in detail.

Then, the program proceeds to S30, in which the obtained basic fuel injection amount Tim is multiplied by a target air-fuel ratio correction coefficient KCMDM (k) and another correction coefficient KTOTAL (integrated value of various correction coefficients performed by multiplication such as water temperature correction). The required fuel injection amount Tcyl (k) required by the internal combustion engine is determined. In this control, the target air-fuel ratio is actually obtained as the equivalent ratio as described above,
This is also used as a correction coefficient for the fuel injection amount. In addition, since the charging efficiency of the intake air differs depending on the heat of vaporization,
A filling efficiency correction is performed on the target air-fuel ratio with appropriate characteristics to obtain a target air-fuel ratio correction coefficient KCMDM.

Then, the program proceeds to S32, in which the required fuel injection amount Tcy is calculated.
l (k) is multiplied by one of the feedback correction coefficient KLAF (k) or KSTR (k) obtained in S26 or S28, and an addition term TTOTAL is added to the product, and the output fuel injection amount T
out (k) is determined. Here, the addition term TTOTAL indicates the total value of the correction coefficient performed by the addition value such as the atmospheric pressure correction. (However, the invalid time of the injector is not included because it is separately added when the output fuel injection amount Tout is output.) ).

Then, the program proceeds to S34, in which the output fuel injection amount Tout (k) is corrected using the adhesion coefficient obtained by searching the adhesion coefficient map from the engine cooling water temperature or the like, and the output fuel injection amount Tout (k). The intake pipe wall adhesion correction of k) is performed (the value after the adhesion correction is Tout-F (k)). Incidentally, since the suction pipe wall surface adhesion correction itself has no direct relation to the gist of the present invention, the description is omitted. Then, the program proceeds to S36, in which the output fuel injection amount Tout-F (k) corrected for adhesion is outputted, and the process ends.

If the result in S18 or S20 is negative, the program proceeds to S38, in which the basic fuel injection amount Tim (k) is multiplied by the target air-fuel ratio correction coefficient KCMDM (k) and various correction coefficients KTOTAL, and the product is corrected and added. The term TTOTAL is added to calculate the output fuel injection amount Tout, and the routine proceeds to S34 and thereafter. If it is determined in step S12 that cranking is to be performed, the process proceeds to step S40 to search for a fuel injection amount Ticr at the time of cranking. The process proceeds to step S42 to calculate the output fuel injection amount Tout by the equation of the start mode, and to perform fuel cut in step S. When it is determined that this is the case, the routine proceeds to S44, where the output fuel injection amount Tout is set to zero.

Next, the feedback correction coefficient KSTR using the adaptive control law mentioned in S28 of the flow chart of FIG.
The operation of (k) will be described.

FIG. 7 is a block diagram showing the operation more functionally.

The illustrated device is based on the adaptive control technique proposed earlier by the present applicant. It comprises an adaptive controller consisting of a STR (self-tuning regulator) controller and an adaptive (control) parameter adjusting mechanism (also referred to as a "parameter adjusting mechanism" in this specification) for adjusting its adaptive (control) parameter (vector). The STR controller receives a target value and a control amount (plant output) of a feedback system for fuel injection amount control, receives a coefficient vector identified by an adaptive parameter adjustment mechanism, and calculates an output.

In such adaptive control, I.I. D. There is a parameter adjustment rule proposed by Landau et al. This method converts the adaptive control system into an equivalent feedback system consisting of linear and nonlinear blocks, and adjusts the nonlinear blocks so that Popov's integral inequality for input and output is satisfied and the linear blocks are strongly positive. This is a method that guarantees the stability of an adaptive control system by determining a rule.
In other words, in the parameter adjustment rule proposed by Landau et al., The adjustment rule (adaptive rule) expressed in a recurrence form is stabilized by using at least one of the above-mentioned Popov's hyperstability theory or Lyapunov's direct method. Is guaranteed.

This method is described in, for example, “Computer Roll” (Corona Corp.) No. 27, 28-41, or "Automatic Control Handbook" (Ohm), pages 703-707, "A Survey of Model Reference Adaptive
Techniques-Theory and Ap-plication "ID LANDAU
"Automatica" Vol. 10, pp. 353-379, 1974, "Unifi-c
ation of Discrete Time Explicit Model Reference A
daptive ControlDesigns "IDLANDAU and others" Automatic
a "Vol. 17, No. 4, pp. 593-611, 1981, and" Comb
ining Model Reference Adaptive Controllers and Sto
chasticSelf-tuning Regulators "ID LANDAU" Autom
atica ", Vol. 18, No. 1, pp. 77-84, 1982.

In the adaptive control technique of the illustrated example, the adjustment law of Landau et al. Is used. To explain below, Landau et al.'S adjustment rule states that the transfer function B (Z ^-1 ) / A
When the polynomial of the denominator and numerator of (Z ^-1 ) is represented by Equations 1 and 2, the adaptive parameter θ hat (k) identified by the parameter adjustment mechanism is represented by a vector (transposed vector) as shown by Equation 3. . Input to parameter adjustment mechanism
(k) is defined as in Equation 4. Here, m = 1,
A case where n = 1 and d = 3, that is, a plant having a dead time of three control cycles in the primary system is taken as an example.

[0053]

(Equation 1)

[0054]

(Equation 2)

[0055]

(Equation 3)

[0056]

(Equation 4)

Here, the adaptive parameter θ shown in Expression 3
The hat is a scalar quantity b0 that determines the gain hat- ¹ (k)
, The control element BR hat (Z ⁻¹ ,
k) and a control element S (Z ⁻¹ ,
k), and are expressed as Equations 5 to 7, respectively.

[0058]

(Equation 5)

[0059]

(Equation 6)

[0060]

(Equation 7)

The parameter adjustment mechanism identifies and estimates the scalar amount and each coefficient of the control element, and sends them to the STR controller as the adaptive parameter θ hat shown in the above equation (3). The parameter adjustment mechanism controls the operation amount u of the plant.
(I) and control amount y (j) (i and j include past values)
Is used to calculate the adaptive parameter θ hat so that the deviation between the target value and the control amount becomes zero. The adaptive parameter θ hat is specifically calculated as shown in Expression 8. In equation 8, Γ
(k) is a gain matrix (m + n + d order) that determines the identification / estimation speed of the adaptive parameter, and e asterisk (k) is the identification / estimation speed.
A signal indicating the estimation error, which is represented by a recurrence formula as shown in Expressions 9 and 10, respectively.

[0062]

(Equation 8)

[0063]

(Equation 9)

[0064]

(Equation 10)

Various specific algorithms are given depending on how to select λ1 (k) and λ2 (k) in equation (9). For example, assuming that λ1 (k) = 1, λ2 (k) = λ (0 <λ <2), a gradually decreasing gain algorithm (least square method when λ = 1), λ1 (k) = λ1 (0 <λ1 <1), λ2 (k) =
If λ2 (0 <λ2 <λ), a variable gain algorithm (weighted least squares method when λ2 = 1), λ1 (k)
When / λ2 (k) = σ and λ3 is expressed as in Equation 11, setting λ1 (k) = λ3 results in a fixed trace algorithm. When λ1 (k) = 1 and λ2 (k) = 0, the fixed gain algorithm is used. In this case, as is apparent from Equation 9, Γ (k) = Γ (k−1), and thus Γ (k) = Γ
Is fixed. For time-varying plants such as fuel injection or air-fuel ratios, any of the progressive gain algorithm, the variable gain algorithm, the fixed gain algorithm, and the fixed trace algorithm are suitable.

[0066]

[Equation 11]

Here, in FIG. 7, the aforementioned STR
The controller (adaptive controller) and the adaptive parameter adjustment mechanism are located outside the fuel injection amount calculation system, and the detected air-fuel ratio KACT
(k) is operated so that it adaptively matches the target air-fuel ratio KCMD (k-d ') (where d' is the dead time until KCMD is reflected in KACT as described above), and the feedback correction coefficient KSTR
(k) is calculated. That is, the STR controller receives the coefficient vector θ hat (k) adaptively identified by the adaptive parameter adjustment mechanism, and receives the target air-fuel ratio KCMD (k−d ′).
The feedback compensator is formed to match The calculated feedback correction coefficient KSTR (k) is multiplied by the required fuel injection amount Tcyl (k), and the corrected fuel injection amount is output as the output fuel injection amount Tout (k) via the adhesion correction compensator to the control plant (internal combustion engine). Institutions).

As described above, the feedback correction coefficient KSTR
(k) and the detected air-fuel ratio KACT (k) are obtained and input to the adaptive parameter adjusting mechanism, where the adaptive parameter θ hat (k) is calculated and input to the STR controller.
The target air-fuel ratio KCMD (k) is input to the STR controller as input.
And the detected air-fuel ratio KACT (k) becomes the target air-fuel ratio KCMD (k-
The feedback correction coefficient KSTR (k) is calculated using the recurrence formula so as to match d ′).

The feedback correction coefficient KSTR (k) is specifically obtained as shown in Expression 12.

[0070]

(Equation 12)

On the other hand, the detected air-fuel ratio KACT (k) and the target air-fuel ratio KC
The MD (k) is also input to the controller (shown as PID in the figure) based on the PID control law described earlier in S26 of the flow chart of FIG. The second feedback correction coefficient KLAF (k) is calculated based on the PID control law in order to eliminate the deviation. The feedback correction coefficient KSTR based on the adaptive control law and the feedback correction coefficient KLAF based on the PID control law are determined by the switching mechanism 40 shown in FIG.
Either one is used for calculating the fuel injection amount via 0. Then, as described later, when the operation of the adaptive control system (STR controller) is determined to be unstable, or when the adaptive control system is out of the adaptive region, the feedback correction coefficient KSTR (k) based on the adaptive control law is used instead of The feedback correction coefficient KLAF (k) based on the PID control law is used.

By the way, when controlling the fuel injection amount of the internal combustion engine, as shown in FIG. 57, the injection amount is calculated, and a certain time is required until the calculated fuel is compressed, exploded and exhausted in the cylinder. It costs. Furthermore, considering the time required for the exhaust gas to reach the LAF sensor, the detection delay of the sensor itself, and the time required for calculating the amount of fuel actually sucked into the cylinder from the detected value, this time is further increased. Become.
As described above, dead time is inevitably involved in controlling the fuel injection amount of the internal combustion engine. Assuming that the dead time is, for example, three times in the combustion cycle as described above by focusing on one cylinder, when the internal combustion engine has four cylinders, as shown in FIG.
C. Here, the "combustion cycle" is a four-stroke process consisting of suction, compression, explosion, and exhaust. In this embodiment, it corresponds to 4TDC.

In the adaptive controller (STR controller) described above, the number of elements of the adaptive parameter θ hat (k) is m + n + d, as is apparent from Equation 3, and is proportional to the dead time d. Assuming that the dead time is 3 as in the above example, when operating the STR controller and the adaptive parameter adjusting mechanism in TDC synchronization in order to cope with an ever-changing operation state, the number of elements of the adaptive parameter θ hat (k) is , M = n = 1, as shown in FIG.
(3 combustion cycles × 4 TDC), and m + n + d = 1
It becomes 4. As a result, the calculation of the gain matrix Γ becomes 14 × 14
The amount of calculation increases, the load on the on-board computer increases, and in the performance of a conventional on-board computer, it becomes difficult to complete the operation within 1 TDC with an increase in the engine speed. As described above, an increase in the number of times of dead time causes deterioration of controllability.

Therefore, the fuel injection control device for an internal combustion engine shown in the figure can cope with the ever-changing operation state as much as possible, and also reduces the amount of matrix calculation to reduce the load on the on-vehicle computer. Specifically, as shown in FIG. 9, the control plant output is synchronized with the combustion cycle, more specifically, only a predetermined crank angle (such as TDC) of a specific cylinder (such as the first cylinder). Then, the above-mentioned adaptive parameter θ hat is calculated.

Here, the calculation of the adaptive parameter θ hat is performed at a predetermined crank angle (such as TDC) of all cylinders, as is apparent from FIG. Note that the fact that the STR controller operates in synchronization with a predetermined crank angle (TDC or the like) of all cylinders to calculate the feedback correction coefficient is not different from the configuration shown in FIG.

Thus, for example, the combustion cycle, ie,
When the operation is performed in synchronization with only a predetermined crank angle of a specific cylinder, d = 3, the number of elements of the adaptive parameter θ hat is m + n + d = 5, and the calculation of the gain matrix Γ is 1
The matrix operation is reduced from 4 × 14 to 5 × 5, and the load on the in-vehicle computer is reduced, so that the operation can be processed within 1 TDC. As described above, the large dead time of the controlled object generally impairs controllability as compared with the case where the dead time is small, and particularly becomes remarkable in adaptive control. Can be greatly reduced, and controllability can be improved.

Specifically, the above can be realized by taking the control cycles k of the formulas 1 to 12 for each cylinder. More specifically, in the case of a four-cylinder internal combustion engine, equation (4) is changed to equation (13), equation (8) to equation (14), equation (9) to equation (15), and equations (10) to (12) as equations (16) to (18). Just do it.

[0078]

(Equation 13)

[0079]

[Equation 14]

[0080]

(Equation 15)

[0081]

(Equation 16)

[0082]

[Equation 17]

[0083]

(Equation 18)

Thus, even in the configuration shown in FIG.
As in the configuration shown in FIG. 8, it is possible to reduce the order of the matrix and vector used in the calculation while taking the control cycle for each TDC of all the cylinders, that is, calculating the adaptive parameters in synchronization with the TDC of all the cylinders. Become. Of course, the control cycle is taken for each cylinder, and
It goes without saying that the same operation is performed even if a configuration having internal variables for each cylinder by setting = cylinder number × k. still,
Here, K indicates the number of combustion cycles, and k indicates TDC. FIG. 10 is a diagram in which the configuration of FIG. 8 is rewritten focusing on the STR controller and the parameter adjustment mechanism. In FIG. 10, the operation cycle m × TDC of the STR controller and the operation cycle n × TDC of the parameter adjustment mechanism are each represented by m.
If n = 1, the configuration shown in FIGS. 8 and 9 is obtained. Here, if the input cycle of the parameter adjustment mechanism is synchronized with TDC and the dead time is d = 12 , the configuration shown in FIG. 8 is obtained.
On the other hand, if the input cycle of the parameter adjustment mechanism is synchronized with the combustion cycle and the dead time is d = 3, the configuration shown in FIG. 9 is obtained.

However, inputting the plant output to the parameter adjusting mechanism in synchronization with the combustion cycle to perform the operation (operation) means that the operation is performed in synchronization with the predetermined crank angle of the specific cylinder. The exhaust gas air-fuel ratio. As a result, when controlling the stoichiometric air-fuel ratio, for example, when the exhaust gas air-fuel ratio of the specific cylinder is in the lean direction and that of the remaining cylinders is in the rich direction, the adaptive controller (STR controller) reduces the operation amount. In some cases, the air-fuel ratio of the remaining cylinders is further increased in the rich direction by adjusting the air-fuel ratio in the rich direction so as to match the target value.

For that purpose, in the illustrated apparatus, as will be described later, the plant output is input to the parameter adjusting mechanism in synchronization with the combustion cycle to operate, thereby reducing the number of elements of the adaptive parameter and reducing the matrix operation amount. In addition to reducing the impact, the influence of the exhaust gas air-fuel ratio of the specific cylinder has been reduced. In order to realize this, the operation is performed as follows.

The parameter adjusting mechanism operates in synchronization with the combustion cycle, ie, operates in synchronization with a predetermined crank angle of a specific cylinder among the four cylinders.
(k) is determined as a predetermined crank angle of each cylinder during a combustion cycle, for example, as an average value of the detected air-fuel ratio KACT (k) for each TDC, for example, a simple average value, and is input to the parameter adjusting mechanism, thereby obtaining the specific cylinder. So that it is not greatly affected by the exhaust gas air-fuel ratio.

Further, for each predetermined crank angle, an average value is also obtained for the adaptive parameter θ hat calculated by the parameter adjustment mechanism, and an average value is also obtained for the feedback correction coefficient KSTR (k) calculated by the STR controller. Thereby, the exhaust gas air-fuel ratio of the specific cylinder is not greatly affected.

FIG. 11 is a subroutine flowchart showing the calculation operation.

Referring to the figure, first, at S100, it is determined whether or not the engine is in a predetermined operating range. here,
The predetermined operation region is a low rotation region including idle.
If it is determined in S100 that the vehicle is not in the predetermined operating range,
The process proceeds to step 102, where the currently calculated air-fuel ratio KACT (k) calculated for the cylinder in S24 of FIG. 6, the previously calculated air-fuel ratio KACT (k-1) for the previous combustion cylinder, and the previous and previous calculated air for the previous and previous combustion cylinders The average value KACTAVE of the fuel ratio KACT (k-2) and the previous / last calculated air-fuel ratio KACT (k-3) for the previous / previous previous combustion cylinder is determined, and is set as a control amount y (k) which is the plant output.
That is, the control cycle is traced back three times, and a simple average value of the air-fuel ratios calculated during one combustion cycle for the four cylinders including the cylinder is determined as a control amount y (k). With this method, the influence of the exhaust gas air-fuel ratio of the specific cylinder can be reduced.

Then, the process proceeds to S104, where as shown at the end of FIG. 7, the adaptive parameter θ hat (k) is calculated from the control amount y (k) just obtained by the parameter adjustment mechanism in accordance with the equation (3) and input to the STR controller. I do.

Then, the process proceeds to S106, in which the calculated values up to three control cycles before including the adaptive parameter θ hat (k) calculated this time, that is, θ hat (k), θ hat (k− 1) Calculate an average value of θ hat (k-2) and θ hat (k-3), for example, a simple average value AVE-θ hat (k).
That is, for the adaptive parameter θ hat (k) on the output side, not on the input side of the parameter adjustment mechanism, the average value of θ hats for four control cycles (one combustion cycle) corresponding to that of the four cylinders is obtained and sent to the STR controller. input. Even if this method is used, even if the average value of the θ hat (k) of the four cylinders is input to the STR controller, the object of reducing the effect of the exhaust gas air-fuel ratio of the specific cylinder can be achieved. It should be noted that since the θ hat is obtained as a vector as shown in Equation 3, its average value is calculated for each element s of the vector.
It is calculated by calculating the average value of 0, r1, r2, r3, b0. Note that an average value may be obtained for any one of the elements, a change amount may be obtained for the other elements in proportion to the average value, and the average value of the θ hat may be calculated from them. In S106,
The expression for obtaining the average value of θ hat including the meaning is schematically shown.

Subsequently, the flow proceeds to S108, where the STR controller calculates a feedback correction coefficient KSTR (k) according to the equation (12) based on the input value. Then, the flow proceeds to S110, where the feedback correction coefficient KSTR (k) calculated this time is calculated.
Calculated values up to three control cycles before, including KSTR (k), KSTR (k-1), KSTR (k-2) and KSTR (k-3) during one combustion cycle
, For example, a simple average value AVEKSTR (k) is calculated. That is, the STR controller which outputs the control input KSTR (k) which is the feedback correction coefficient of the fuel calculation system, not the parameter adjustment mechanism,
This is because even if the average value of the KSTR for the control cycle (one combustion cycle) is obtained, the object of reducing the influence of the exhaust gas air-fuel ratio of the specific cylinder can be achieved.

On the other hand, if it is determined in S100 that the engine is in the predetermined operation range, the flow proceeds to S112 to calculate y (k), that is, the presently calculated equivalent ratio KACT (k) obtained in S24 of FIG. Is directly used as a control amount (plant output). Then, the process proceeds to S114 to calculate the adaptive parameter θ hat (k) in the same manner as in S104, and proceeds to S116 to calculate the feedback correction coefficient KSTR (k) in the same manner as in S108.

As described above, since the average value of the air-fuel ratios of all the cylinders is obtained and input to the parameter adjusting mechanism as the control amount y (k), the equivalent ratio of the specific cylinder (for example, the first cylinder), more specifically, Is not greatly affected by the exhaust gas air-fuel ratio. Further, as for the output of the STR controller, a signal vector ζ is obtained by using values for four control cycles including the latest value u (k) = KSTR (k), and is input to the parameter adjustment mechanism. The effect of the gas air-fuel ratio is further reduced.

Further, regarding the adaptive parameter θ hat (k) not on the input side but on the output side of the parameter adjusting mechanism, the average value of θ hats for four control cycles (one combustion cycle) corresponding to that of four cylinders is obtained. Since the input is made to the STR controller, the object of reducing the influence of the exhaust gas air-fuel ratio of the specific cylinder can be achieved also by smoothing. Furthermore, KSTR, which is a feedback correction coefficient of the fuel calculation system, is not on the parameter adjustment mechanism side.
The STR controller that outputs (k) also has a KS of four control cycles (one combustion cycle) corresponding to that of the four cylinders.
Since the average value of TR is obtained, the influence of the exhaust gas air-fuel ratio of the specific cylinder can be similarly reduced.

On the other hand, in S100, the engine operates in a predetermined operating region,
Specifically, it is determined whether or not the vehicle is in a low rotation region including idle, and when the result is affirmative, the average value is not calculated, so that no inconvenience occurs. That is, since the control cycle becomes longer at low rotation, the response delay of the LAF sensor can be ignored. Conversely, since the phase of the detected air-fuel ratio KACT (k) and its average value KACTAVE are shifted as shown in FIG. 12, the same phenomenon as the change in the dead time of the control system occurs. Therefore, if adaptive control is performed using KACTAVE (k) having a phase shift, adverse effects such as hunting may occur. For this reason, when the engine is in a low rotation state such as during an idling operation and is affected by this, smoothing is stopped.

In the above description, the average value AVE-θ hat (k) of the adaptive parameter θ hat calculated in S106 is not used for calculating the identification error signal e asterisk shown in Expression 10. That is, since the identification error signal e asterisk is a function for evaluating the magnitude of the error between the detected air-fuel ratio and the target air-fuel ratio, the AVE-θ hat (k) obtained as described above is expressed by the following equation (10).
If the AVE-θ hat (k) is used only for the calculation of Equation 8, it is useful to provide an operating area that is not used for the calculation of Equation 10, because the error may be inaccurate.

In the above, S102, S10
6, In S110, the average values of the air-fuel ratio, θ hat (k), and KSTR (k) are all used, but it goes without saying that any one or two suitable values may be used. In addition, in the calculation of the average value at the time of starting the engine or restarting the calculation of the STR controller, if there is no past value, it goes without saying that an appropriate predetermined value is used.

When the average value of the adaptive parameter θ hat (k) and the feedback correction coefficient KSTR (k) is obtained,
It is not always necessary to input those values to the parameter adjustment mechanism. This is the adaptive parameter θ hat
This is because the feedback correction coefficient KSTR (k) calculated by the STR controller using the average value of (k) is already a value that is not greatly affected by the exhaust gas air-fuel ratio of the specific cylinder. Similarly, the average value of the feedback correction coefficient KSTR (k) calculated by the STR controller is a value that is not significantly affected by the exhaust gas air-fuel ratio of the specific cylinder.

The selection of the feedback correction coefficient shown in S32 of the flow chart of FIG. 6 will be described.

FIG. 13 is a subroutine flowchart showing the operation.

Referring to the figure, first, in S200, it is determined whether or not it is in the applicable area of the adaptive control system. For example,
In the unstable combustion operation region such as extremely low water temperature region, the accurate calculated air-fuel ratio KACT (k) is not obtained, so it is out of the applicable region.
In that case, the process proceeds to S210, where the output fuel injection amount Tout (k) is calculated using the feedback correction coefficient KLAF (k) obtained by the PID control law. If it is determined that the adaptive control system is in the applicable area, the process advances to step S202 to determine the stability of the adaptive control system using each element of the adaptive parameter θ hat.

Specifically, the transfer characteristic of the feedback correction coefficient KSTR (k) calculated by the STR controller is given by
9 is represented.

[0105]

[Equation 19]

Here, assuming that the adhesion correction is correct and there is no disturbance in the fuel calculation system, KSTR (k) and KACT
The transfer characteristic of (k) is as shown in Expression 20.

[0107]

(Equation 20)

The transfer function from KCMD (k) to correction coefficient KSTR (k) is as shown in Equation 21.

[0109]

(Equation 21)

[0110] Here, b0 because a scalar quantity determining the gain, zero or the negative and not be, the denominator function f of the transfer function of the number ^{21 (z) = b0Z 3 +} r1Z 2 + r
2Z + r3 + s0 is one of the functions shown in FIG. Therefore, it is determined whether the real root is within the unit circle,
That is, as shown in FIG. 15, f (−1) <0 to f (−1)
(1) If it is determined whether or not> 0, the affirmative determination means that the real root is within the unit circle, so that it can be easily determined whether or not the system is stable.

Then, the process proceeds to S204, where it is determined whether or not the adaptive control system is unstable. If the result is affirmative, the process proceeds to S206, where the adaptive parameter vector θ hat is returned to the initial value. Thereby, the stability of the system can be restored. Then, the process proceeds to S208, where the gain matrix Γ is corrected. Since the gain matrix Γ determines the change (convergence) speed of the parameter adjustment mechanism, this correction is performed so as to reduce the convergence speed. Here, each element of the gain matrix Γ is replaced with a small value. Thereby, the stability of the system can be restored similarly. Then, the process proceeds to S210, where the adaptive control system is unstable, as shown in the figure, so that the feedback correction coefficient KLAF (k) based on the PID control law is used as the feedback correction coefficient.
Is multiplied by the required fuel injection amount Tcyl (k), and an addition term TTOTAL is added to the product to obtain an output fuel injection amount Tcyl (k).
out (k) is determined.

If it is determined in step S204 that the adaptive control system is not unstable, the flow advances to step S212, and as shown in FIG.
The output fuel injection amount Tout (k) is calculated using the correction coefficient KSTR (k) based on the adaptive control law as the feedback correction coefficient. At this time, if the average value of the feedback correction coefficient KSTR is obtained in S110 of the flow chart of FIG. 11, it goes without saying that the average value is used.

In the block diagram of FIG. 7, the switching mechanism 4
The output u (k) of 00 is input to the STR controller and the parameter adjustment mechanism. This is to enable the calculation of the feedback correction coefficient KSTR by the adaptive control law even when the feedback correction coefficient KLAF by the PID control law is selected.

In this embodiment, as a result of the above-described configuration, the number of elements of the adaptive parameter is 5 although the parameter adjustment mechanism operates for each cylinder TDC.
The matrix operation is reduced to 5 × 5, and the load on the on-board computer is reduced.
The calculation can be completed between TDCs. On the other hand, S
The TR controller also calculates the feedback correction coefficient KSTR for every cylinder TDC, and changes the change to all cylinders TD.
By performing this for each C, it is possible to respond to changes in the operating state as much as possible. In addition, by greatly reducing the dead time,
Controllability can be improved.

Further, when the parameter adjusting mechanism operates on a cylinder-by-cylinder basis, it operates every combustion cycle. As a result, the cylinder always operates at a predetermined crank angle of a specific cylinder, for example, the first cylinder. The average value of the detected air-fuel ratios (control amounts) for all the groups including the group is obtained, and the average value is input to the parameter adjustment mechanism, or the average value of the adaptive parameter θ hat is obtained, or STR
Since the average value of the feedback correction coefficient KSTR, which is the output of the controller, is obtained and used, there is no inconvenience of strongly reflecting only the combustion state of the specific cylinder.

That is, if the feedback correction coefficient KSTR is determined based on the control amount for a specific cylinder, for example, when the air-fuel ratio of the first cylinder is rich and that of the other cylinders is lean, the feedback correction coefficient KSTR becomes empty. The fuel ratio is determined to be corrected in the lean direction, and the leaning of the air-fuel ratio of the other cylinders is spurred. However, such an inconvenience does not occur as a result of the average value of all the cylinders.

For further simplification, as shown in FIG. 16, the adaptive parameter .theta. Is not synchronized for every cylinder TDC, but synchronized with the combustion cycle of a specific cylinder, that is,
The calculation may be performed once every 4 TDC, and the STR controller may use the same value as the adaptive parameter θ hat several times in the cylinder (corresponding to the case where m = 1 and n = 4 in FIG. 10).

This method is particularly effective when the operable time decreases with an increase in the engine speed. Since the variation of the adaptive parameter θ hat required for each cylinder at the time of high rotation is reduced, even if the adaptive parameter θ hat of a specific cylinder is used for all cylinders including other cylinders, there is little deterioration in controllability. Without compromising controllability
The calculation time can be reduced.

Further, as shown in FIG. 17, if the STR controller is operated only once every 4 TDC in synchronization with the combustion cycle, the configuration can be further simplified. Although the control accuracy is reduced, some effects can be obtained with this configuration (m = n in FIG. 10).
= 4).

FIG. 18 is a flow chart showing a second embodiment of the apparatus according to the present application, and relates to the setting of a gain matrix 用 い る used for calculating the feedback correction coefficient KSTR.

In the operation of the feedback correction coefficient KSTR, the gain matrix 前述
(k) is required. In the second embodiment, λ ₁ = 1 and λ ₂ = 0 in Equation 9, that is, when the fixed gain algorithm is used, all off-diagonal elements of the gain matrix Γ (k) are set to 0. As a result, the calculation time was shortened and the setting was facilitated.

For the sake of explanation, the internal variable Γζ (k
Consider the case where the operation of -d) is performed. Gain matrix Γ is 5 × 5
In the case of the first embodiment, the operation of is performed as shown in Expression 22, which requires 25 multiplications and 20 additions.

[0123]

(Equation 22)

This is done by setting all off-diagonal elements of the gain matrix ０ to 0.
In other words, it can be expressed as in Expression 23, and the operation can be reduced to five multiplications.

[0125]

(Equation 23)

Further, all off-diagonal elements of the gain matrix Γ are set to 0
In the case where the calculation of the adaptive parameter θ hat (k) is performed, Equation 24 is obtained.

[0127]

(Equation 24)

As a result, the matrix elements g ₁₁ , g ₂₂ , g ₃₃ , g
₄₄ and g ₅₅ can independently set the changing speed of each element of the adaptive parameter θ hat (k) with a value corresponding to only one element of ζ (k). If the off-diagonal element of the gain matrix で な け れ ば is not 0, as can be seen from Expressions 22 and 24, the operation of the adaptive parameter θ hat (k) is as shown in Expression 25, and one of the θ hat (k) To determine the rate of change of the element, it is necessary to consider five variables corresponding to all elements of ζ (kd), which makes setting difficult. By setting all off-diagonal elements of the gain matrix ０ to 0, it is possible to shorten the operation time and facilitate the setting.

[0129]

(Equation 25)

Further, when the inventors conducted a test,
In the Γ matrix, if some of the five setting elements g _{11 to} g ₅₅ are set to the same value, the rate of change of each element of the adaptive parameter θ hat (k) becomes appropriate, and the controllability is the best. It turned out to be. For example, g ₁₁ = g
_{22 = g 33 = g 44 =} g is the case far. With such placing, it is possible to reduce the setting element into two g and g _55, it is possible to reduce the man-hours for setting, for example, the internal variable zeta ^T of (kd) Γζ (kd) The operation is as shown in Equation 26, and the multiplication is performed 12 times.

[0131]

(Equation 26)

On the other hand, when g ₁₁ to g ₄₄ take different values, the above operation becomes as shown in Expression 27, and the number of multiplications increases to 15 times.

[0133]

[Equation 27]

As described above, by setting some of g ₁₁ to g ₅₅ to the same value, the number of setting elements can be reduced, and the calculation time can be further reduced. Further, since the ratio of the change speed of each element of the adaptive parameter θ hat (k) can be made appropriate, the controllability is also improved.
At this time, if g ₁₁ = g ₂₂ = g ₃₃ = g ₄₄ = g ₅₅ , it goes without saying that the effect is most apparent.

Further, for example, in an operation region where the combustion is unstable and the plant output is also unstable, the above g
By reducing ₅₅ , hunting of so (k) can be suppressed. As described above, setting the off-diagonal element of the gain matrix ０ to 0 has a great merit that the setting of the control characteristics becomes easy. Further, by changing the gain matrix Γ depending on the operation region, it is possible to always obtain optimal controllability for the engine.

In this case, g _{11 to} g ₅₅ are stored in the RAM 74 in the control unit 34 according to the operation state. More specifically, in addition to operating conditions, canister purge,
The information is stored according to the operating state of the engine control device such as the exhaust gas recirculation. At this time, g _{11 to} g ₅₅ may be all the same value, all different values, or some same values. In this case, if there is a margin in the capacity of the RAM 74 or the operation time, a non-diagonal element of the gain matrix Γ may be used.

Based on the above, a second embodiment of the apparatus according to the present application will be described with reference to the flow chart of FIG.

First, at S300, the engine speed Ne,
The engine operating parameters such as the intake pressure Pb and the operating state of the exhaust gas recirculation mechanism or the canister / purge mechanism described above are read, and the program proceeds to S302 to judge whether or not the engine is in an idling region. Search the Γ map. On the other hand, when it is determined in S302 that the variable valve timing mechanism is not in the idle region, the process proceeds to S306, and it is determined whether the variable valve timing mechanism is operating with the Hi valve timing characteristic. In addition to searching for a Γ map for Lo valve timing, when the result is negative, the process proceeds to S310 to search for a Γ map for Lo valve timing.

FIG. 19 shows the characteristics of the Γ map for Lo valve timing. This map as shown, to search for the matrix element g ₁₁ to g ₅₅ from the engine speed Ne and the manifold pressure Pb. Note that the Γ maps for idle and Hi valve timing have similar characteristics. Further, in this map, since the value of the gain matrix Γ is searched based on the intake pressure Pb indicating the engine load, it is possible to obtain the optimum value of the gain matrix even in a deceleration operation state where the engine load fluctuates rapidly. Can be.

Then, the program proceeds to S312, in which it is determined whether an EGR (exhaust gas recirculation mechanism) is operating. If the result is affirmative, the program proceeds to S314, in which the fuel correction coefficient KEGR for the exhaust gas recirculation rate is determined.
Modify the gain matrix Γ according to N. More specifically, a correction coefficient KΓEGR is obtained from the fuel correction coefficient KEGRN for the exhaust gas recirculation rate by searching a table showing the characteristics in FIG. 20, and the obtained correction coefficient KΓEGR is multiplied by a gain matrix 補正 for correction. The reason for modifying the gain matrix according to the fuel correction coefficient KEGRN for the exhaust gas recirculation rate is that, as shown in the figure, the correction coefficient KΓEGR increases as the fuel correction coefficient KEGRN for the exhaust gas recirculation rate decreases as the exhaust gas recirculation rate increases. Therefore, in order to increase the stability of the adaptive control system, the gain matrix Γ is set to decrease as the fuel correction coefficient KEGRN for the exhaust gas recirculation rate decreases.

The exhaust gas recirculation rate KEGRN is a coefficient for multiplying and correcting the fuel injection amount, and is determined to be, for example, 0.9. However, the gist of the present invention is not in the determination of the exhaust gas recirculation rate itself, and the determination of the exhaust gas recirculation rate is described in, for example, Japanese Patent Application No. 6-294,014 previously proposed by the present applicant. Is omitted.

Then, the program proceeds to S316, in which it is determined whether the canister / purge mechanism is operating. If the result is affirmative, the program proceeds to S318, in which the gain matrix Γ is corrected according to the purge mass. More specifically, a correction coefficient KΓPUG is obtained by searching a table showing the characteristics in FIG. 21 from the purge mass KPUG, and the obtained correction coefficient KΓPUG is multiplied by a gain matrix 補正 for correction. The correction coefficient KΓPUG is the purge mass as shown
Since the disturbance increases as KPUG increases, it is set to increase as the purge mass KPUG increases. The purge mass is also described in, for example, Japanese Patent Application Laid-Open No. Hei 6-101,522 previously proposed by the present applicant, and therefore the description is omitted.

Subsequently, the flow proceeds to S320, where the detected atmospheric pressure P
Modify the gain matrix Γ according to a. More specifically, from the detected atmospheric pressure Pa, a table showing the characteristics in FIG. 22 is searched to determine a correction coefficient KΓPa, and the obtained correction coefficient KΓ
Correct by multiplying the gain matrix 補正 by Pa. The reason for correcting the gain matrix Γ according to the detected atmospheric pressure Pa is that the detected atmospheric pressure Pa decreases, that is, the filling efficiency decreases as the altitude at which the engine is located increases, so that the setting is performed at normal pressure. In order to increase the stability of the adaptive control system, the detected atmospheric pressure Pa
Is set so as to decrease the gain matrix に つ れ て as decreases.

Then, the program proceeds to S322, in which the detected water temperature Tw is detected.
Modify the gain matrix Γ according to. More specifically, a correction coefficient KΓTw is obtained from the detected water temperature Tw by searching a table showing the characteristics in FIG. 23, and the obtained correction coefficient KΓTW is multiplied by the gain matrix 補正 to perform correction. The reason for correcting the gain matrix 修正 according to the detected water temperature Tw is that the correction coefficient KΓTW
As shown in the figure, when the detected water temperature Tw is at a low water temperature or a high water temperature, the combustion becomes unstable, so that a disturbance occurs to the data set at the normal temperature, so that the stability of the adaptive control system increases. As described above, when the water temperature is low or high, the gain matrix Γ is set to be small.

In the second embodiment, as described above, the gain matrix for determining the change (convergence) speed of the adaptive parameter θ hat is appropriately determined according to the operating state.
A stable adaptation parameter change rate can be obtained,
Controllability is improved.

Although the second embodiment determines the gain matrix Γ with a fixed gain, it is also possible to use a variable gain algorithm. In this case, the initial value of each element of the gain matrix Γ Is corrected in the driving state as described above,
The predetermined value may be set when the operating state changes.

Further, in the second embodiment, the fixed gain algorithm has been described. However, the calculation of the gain matrix Γ (k) is performed based on an operation rule other than the fixed gain algorithm such as the variable gain algorithm shown in Expression 9. If you do
By not performing the operation on the off-diagonal elements of the gain matrix Γ (k) and fixing it to 0, it is possible to reduce the amount of operation and facilitate the setting described in the second embodiment. It goes without saying that.

FIG. 24 is a flow chart showing a third embodiment of the apparatus according to this application.

In the first and second embodiments, the gain matrix Γ is calculated with a fixed gain. However, in the third embodiment, the calculation is performed using an algorithm other than the fixed gain. When the control result (plant output, more specifically, the detected air-fuel ratio KACT) using the adaptive parameter shows a good behavior, if the calculated value is stored according to the operating state of the engine, it can be used again in that region. It is not necessary to calculate the gain matrix Γ (k), and at the same time, the optimal gain matrix Γ (k) can always be used in that region.
Controllability is improved. Γ (k) stored at this time is 4TD
A processing value such as an average value between C may be used. When the gain matrix Γ is calculated from the fixed gain algorithm, it is determined that the behavior of the plant output is not good. The gain matrix Γ (k-1) at that time starts as an initial value stored for each operation area.

A description will be given with reference to FIG. 24 on the premise of the above.
This is an operation performed at the time of searching the map of the gain matrix Γ, such as S308, S310, or S304 in FIG. 18 in the flowchart of the third embodiment.

In the following, the engine speed N is determined at S400.
A map of the gain matrix Γ similar to that shown in the second embodiment is searched from e and the intake pressure Pb, and the process proceeds to S402 to determine whether or not the behavior of the detected air-fuel ratio KACT, which is the plant output, is good by an appropriate method. If the judgment is negative, the flow goes to S404 to calculate the gain matrix Γ (k), and the flow goes to S406 to store it in a predetermined area of the searched map. If the result in S402 is affirmative, the process immediately proceeds to S406. Whether the behavior of the detected air-fuel ratio KACT in S402 is good or bad is determined, for example, by 10 TDC.
If the detected air-fuel ratio KACT is within the target air-fuel ratio KCMD ± predetermined value, the determination is made as good.

Since the third embodiment is configured as described above, if the behavior of the detected air-fuel ratio KACT is good, the calculation of the gain matrix Γ (k) is performed without using the arithmetic expression shown in Expression 9. Since the search can be performed by simple map search, the amount of calculation can be reduced. Furthermore, when the behavior of the detected air-fuel ratio KACT is not good, the optimum gain matrix Γ (k) is recalculated and learned for each operation region of the internal combustion engine, thereby coping with deterioration over time of the internal combustion engine. Since the behavior of the detected equivalence ratio KACT (k) can always be improved, controllability can be improved.

FIG. 25 is a flowchart showing a fourth embodiment of the apparatus according to the present application.

In the fourth embodiment, a dead zone is provided in the characteristic of the detected air-fuel ratio KACT so that the adaptive control system does not become unstable. That is, the STR controller determines the detected air-fuel ratio.
Since KACT operates to match the target air-fuel ratio KCMD,
If the detected air-fuel ratio KACT input to the STR controller matches the target air-fuel ratio KCMD, the adaptation parameter hardly changes. Therefore, when the detected air-fuel ratio KACT fluctuates minutely from a minute disturbance such as sensor noise, the adaptive control system is not affected by the minute disturbance to perform unnecessary overcorrection as shown in FIG. As shown in the figure, a dead zone is provided near the target air-fuel ratio KCMD in the characteristics of the detected air-fuel ratio KACT. Specifically, the value of the detected air-fuel ratio KACT is set to be the same in the range of KCMD-β to KCMD + α.

Referring to the flow chart of FIG. 25, in S500, the detected air-fuel ratio KACT is set to a lower limit predetermined value KCMD−
is compared with β, and when it is determined that it is more than that, the routine proceeds to S502, where the detected air-fuel ratio KACT is compared with a predetermined upper limit value KCMD + α. If it is determined in S502 that the detected air-fuel ratio is equal to or smaller than the predetermined value KCMD + α, the process proceeds to S504, where the detected air-fuel ratio KACT is set to a predetermined value, for example, the target air-fuel ratio KCMD. When it is determined in S500 that the detected air-fuel ratio KACT is lower than the lower limit predetermined value KCMD-β, or when it is determined in S502 that the detected air-fuel ratio KACT is higher than the upper limit predetermined value KCMD + α, the program is immediately terminated. I do. Therefore, in that case, the detected value is directly used as the detected air-fuel ratio KACT. With the above processing, as shown in FIG. 26, a dead zone can be provided near the target air-fuel ratio KCMD in the characteristics of the detected air-fuel ratio KACT.

Since the fourth embodiment is configured as described above, for example, even when the detected air-fuel ratio KACT fluctuates minutely, S
The TR controller can operate stably without being affected by the influence, so that a good control result can be obtained. Note that the target air-fuel ratio KCMD is set as the detected air-fuel ratio in S502, but may be set to an appropriate value in a range from KCMD-β to KCMD + α.

FIG. 27 is a flow chart showing a fifth embodiment of the apparatus according to the present application.

The fifth embodiment prevents instability of the adaptive control system as in the fourth embodiment, and provides upper and lower limiters for the identification error signal e asterisk to provide stable adaptive parameters. I got it.

That is, as is apparent from Equation 8, the rate of change of the adaptive parameter θ hat can be limited by limiting the value of the identification error signal e asterisk within a certain range. Thus, the adaptive parameter θ hat
This is because overshooting of the optimal value of (k) can be prevented, and as a result, the adaptive control system can be operated stably and a good control result can be obtained.

Referring to the flow chart of FIG. 27, first, the identification error signal e asterisk (k) calculated in S600 is compared with the upper limit a (shown in FIG. 28). Proceeds to S602 for a predetermined value,
For example, the upper limit a is set as the identification error signal e asterisk (k). On the other hand, in S600, the identification error signal e asterisk (k)
If it is determined that is less than or equal to the upper limit value a, the process proceeds to S604, where the calculated identification error signal e asterisk (k) is compared with the lower limit value b (shown in FIG. 28).
Proceeding to 606, a second predetermined value, for example, the lower limit value b is set as the identification error signal e asterisk (k). When the identification error signal e asterisk (k) is determined to be equal to or larger than the lower limit b in S604, the program is immediately terminated. Therefore, in that case, the identification error signal e asterisk (k) remains at the calculated value.

Since the fifth embodiment is configured as described above, by limiting the value of the identification error signal e asterisk (k) to a certain range or less, the adaptive parameter .theta.
The rate of change of (k) can be limited. As a result, overshoot of the adaptive parameter θ hat (k) with respect to the optimum value can be prevented, and the adaptive control system can be operated stably to obtain good control results.

In S602 to S606, the value of the identification error signal e asterisk (k) is set to the upper and lower limits.
It may be an appropriate value between the upper and lower limits or an appropriate value near the upper and lower limits.

FIG. 29 is a flow chart showing a sixth embodiment of the apparatus according to the present application.

In the sixth embodiment, in the STR controller shown in the first embodiment, the constant 1 used for the denominator of the equation (10) for the identification error signal e asterisk for determining the adaptive parameter θ hat is used. By making it variable, the rate of change is stabilized and controllability is improved.

The sixth embodiment is based on the premise that the adaptive control as shown in the drawing is realized by a low-level on-vehicle microcomputer by limiting the change range of an intermediate variable used for calculation in the parameter adjustment mechanism. This is described in Japanese Patent Application Laid-Open No. Hei 6-161511, which was previously proposed by the present applicant, and will not be described.

That is, in the theoretical formula, the identification error signal e asterisk (k) is calculated as shown in Expression 10. Now, ζ
(k) and y (k) are multiplied by 1/10 (hereinafter, referred to as j) and input to the parameter adjustment mechanism. ) Is constant for a fixed gain).

[0167]

[Equation 28]

Here, the right term is the square of the coefficient by which ζ (k) and y (k) are multiplied. When this coefficient is a small value of 1 or less (in the example, 1/10 ² = 1/100) , Becomes extremely smaller than the left term = 1. Therefore, no matter how the right term changes, the denominator of the identification error signal e asterisk (k) becomes a value close to 1, and the change rate of the identification error signal e asterisk (k) before multiplication by the coefficient changes. I will. In order to solve this problem, the left term may be set to a value other than 1. As a guideline, the coefficients of the When j, if put and j ^2, can be the same change rate as before multiplication coefficient j.

Conversely, the changing speed of the identification error signal e asterisk (k) is the change (convergence) of the adaptive parameter θ hat (k).
Since it is proportional to the speed, that is, θ (k) is calculated using Equation 8, the change speed of the adaptive parameter θ hat (k) can be changed by giving a value other than j ^2. . Accordingly, in the arithmetic expression of the denominator of the identification error signal e asterisk (k) shown in Expression 29, i in the expression takes a value other than 1, that is, i ≠ 1.

[0170]

(Equation 29)

Referring to the flow chart of FIG. 29, first, in S700, the identification error signal e asterisk (k)
It is determined whether or not to perform an operation of changing the change (convergence) speed of the adaptive parameter θ hat (k) according to the above. If the determination is affirmative, the process proceeds to S702, where i is detected as a value other than 1, more specifically, i. From the engine speed Ne and the intake pressure Pb, a map whose characteristics are shown in FIG. 30 is searched to obtain i. On the other hand, S7
The i proceeds to S704 if negative in 00 at a j ^2, the same change rate as before multiplication coefficient j. still,
Since j is a constant, in the map characteristics shown in FIG.
Is set in consideration of j ² , for example, i = j ² × 0.5 to i = j ² × 2.

Specifically, j is usually set to a value smaller than 1. For example, if j = 1/10, if the result in S700 is NO, i = j ² = 1/100. Therefore, even if the result in S700 is affirmative, the i-map value is set in FIG. 30 so that it is, for example, between 1/50 and 1/200 with i = 1/100 as the center. At this time, as i is smaller (for example, 1/200), the change (convergence) speed of the adaptive parameter θ hat (k) is larger, and as i is larger (for example, 1/50), the adaptive parameter θ hat (k) is larger.
Changes (convergence) speed becomes smaller. Therefore, in FIG. 30, the i-map value is set to be larger (for example, 1/50) in a high-speed and high-load state and to be small (for example, 1/200) in a low-speed and low-load state.

Since the sixth embodiment is configured as described above, the identification error signal e for determining the adaptive parameter .theta.
By making the asterisk constant variable, the rate of change of the adaptive parameter θ hat is stabilized in harmony with the coefficient for the input, and good controllability can be achieved.

In the sixth embodiment, the STR controller used in the first embodiment is taken as an example.
The adaptive controller is not limited to the one shown in the first embodiment, but all applicable, including an MRACS-type adaptive controller, are applicable as long as they operate based on Landau et al.'S adjustment rule.

FIG. 31 is a flow chart showing a seventh embodiment of the apparatus according to the present application.

In the seventh embodiment, the control cycle of the parameter adjusting mechanism and the STR controller shown in the first embodiment is made variable, and the operating state, specifically, the engine The control cycle is determined according to the rotation speed. That is, by making the control cycle of the parameter adjustment mechanism or the controller of the adaptive controller variable according to the operation state, the operation load can be reduced as much as possible, and the adaptive control can be performed even in the operation state where the operation time is short such as at a high rotation speed. And good controllability.

Referring to the flow chart of FIG. 31, first, the engine speed Ne detected in S800 is compared with a predetermined value Nep1, and the detected engine speed Ne is set to a predetermined value Nep1.
If it is determined that it is less than 1, the routine proceeds to S802, where the detected engine speed Ne is compared with another predetermined value Nec1. And S
When it is determined that the engine speed Ne detected in 802 is less than another predetermined value Nec1, the process proceeds to S804, and the parameter adjusting mechanism (abbreviated as P in FIG. 31) and the STR controller (FIG. 3)
A control cycle of 1 (abbreviated as C) is set for each TDC.

FIG. 32 is a diagram for explaining the operation of the flow chart of FIG. 31. As shown in the figure, when the predetermined values Nep1, Nec1 are in a relatively low rotation range, there is a margin in the calculation time, so that the control precision is given priority. Let the parameter adjustment mechanism and STR
As shown in FIG. 8 and FIG.
Operate every C.

In FIG. 31, when it is determined that the engine speed Ne detected in S802 exceeds the predetermined value Nec1, the program proceeds to S802.
Proceeding to 806, the detected engine speed Ne is set to a predetermined value Nec2.
If it is determined that it is less than the above, the process proceeds to S808, where the parameter adjustment mechanism operates every TDC and the STR controller operates every 2 TDC. On the other hand, when it is determined that the engine speed Ne detected in S806 is equal to or greater than the predetermined value Nec2, the process proceeds to S810, in which the parameter adjusting mechanism operates every TDC, and the STR controller operates every 4 TDC.

Also, the engine speed Ne detected in S800
When it is determined that is equal to or more than the predetermined value Nep1, the routine proceeds to S812, where the detected engine speed Ne is compared with the predetermined value Nep2. Compared with Nec3, if it is determined that the detected engine speed Ne is less than the predetermined value Nec3, the process proceeds to S816, where the parameter adjustment mechanism is operated every 2 TDC, and the STR controller is operated every TDC.

On the other hand, the engine speed Ne detected in S814
If is determined to be equal to or greater than the predetermined value Nec3, the flow proceeds to step S818, where the detected engine speed Ne is compared with the predetermined value Nec4. To work. If it is determined in S818 that the engine speed Ne is equal to or greater than the predetermined value Nec4, the process proceeds to S822, where the parameter adjustment mechanism is operated every 2 TDCs, and the STR controller is operated every 4 TDCs.

Further, the engine speed Ne detected in S812
If is determined to be equal to or greater than the predetermined value Nep2, the process proceeds to S824, where the detected engine speed Ne is compared with the predetermined value Nep3. Compared with Nec5, if it is determined that the detected engine speed Ne is less than the predetermined value Nec5, the process proceeds to S828, where the parameter adjustment mechanism is operated every 4 TDCs, and the controller is operated every TDC (shown in FIG. 16).

On the other hand, the engine speed Ne detected in S826
If it is determined that is equal to or greater than the predetermined value Nec5, the process proceeds to S830, where the detected engine speed Ne is compared with the predetermined value Nec6. Controller is 2T
When the engine speed Ne detected in step S830 is equal to or greater than the predetermined value Nec6, the operation proceeds to step S8.
Proceeding to 34, both the parameter adjustment mechanism and the STR controller are operated every 4 TDC (shown in FIG. 17). Note that S
If it is determined in 824 that the engine speed Ne is equal to or greater than the predetermined value Nep3, the flow proceeds to S836 to stop the adaptive controller STR.

In the seventh embodiment, as described above, the control cycle of the STR controller and the parameter adjustment mechanism of the adaptive controller are determined according to the engine speed.
The calculation load is reduced as much as possible, so that the adaptive control can be performed even in an operation state where the calculation time is short, such as at a high rotation speed, and good controllability can be realized.

Note that the adaptive controller ST shown in FIG.
It is not necessary to provide all of the operating states of R from 1 to 10 (indicated by circled numbers in the figure).
It may be appropriately selected according to the ability of U. For example, 1,
3,5,9,10 to 1,3,6,9,10 to 1,7,9,10 to 1,10 to 1,4
7, 10, etc. may be selected.

Further, although the engine speed is used as the operating state, the present invention is not limited to this, and may be determined in consideration of the engine load. In that case, for example, the change of the adaptive parameter θ hat is small in a high load state, so that the parameter adjustment mechanism may be processed every 4 TDC.

FIG. 33 shows an eighth embodiment of the apparatus according to the present invention.
4 is a subroutine flowchart showing a calculation operation of an average value such as KSTR.

In the case of the first embodiment, in order to avoid the influence of the exhaust air-fuel ratio of a specific cylinder, an average value is basically obtained for the element for determining the feedback correction coefficient KSTR, and a predetermined operation is performed. In the state, that is, in the idle state, the calculation of the average value is stopped.

In the eighth embodiment, in contrast to the first embodiment, the average value is not calculated in principle, and the average value is calculated only in a predetermined operating state, specifically, during execution of the exhaust gas recirculation (EGR). The value was calculated.

To explain this, when the exhaust gas is recirculated in the above-described exhaust gas recirculation mechanism, depending on the operating condition, the exhaust gas is not evenly introduced into the four cylinders, but, for example, a large amount is discharged into the cylinder close to the recirculation port 121b. Exhaust gas may be sucked in and a small amount may be sucked into a distant cylinder. Therefore, in such a case, TD
The air-fuel ratio KACT (k) detected for each C is greatly affected by the specific cylinder, and if the detected air-fuel ratio KACT (k) is used, it is attempted to adjust only the equivalence ratio of that cylinder to the target air-fuel ratio. The control values of all the cylinders are offset by the deviation of the cylinder, and the air-fuel ratios of the other cylinders are shifted. Therefore, in order to avoid this, it is desirable to obtain an average value as shown in the figure.

Referring to FIG. 33, at S900 E
It is determined whether GR (exhaust gas recirculation control) is being executed,
If the result is affirmative, the process proceeds to S902 and the subsequent steps, and an average value such as KACTAVE is obtained in the same manner as described in the first embodiment with reference to FIG. On the other hand, when the result in S900 is NO, S912
After that, the same processing as described in the first embodiment with reference to FIG. 11 is performed.

Since the eighth embodiment is configured as described above, even when the exhaust gas is recirculated, it is not greatly affected by the specific cylinder, and the controllability is improved.

FIG. 34 shows a ninth embodiment of the device according to the present application, and is a feedback correction coefficient similar to that of FIG.
4 is a subroutine flowchart showing a calculation operation of an average value such as KSTR.

As in the case of executing the exhaust gas recirculation, when the gas is supplied by performing the canister purging, the gas may not be uniformly introduced into the cylinder depending on the operation state, and as described in the eighth embodiment. A similar problem can occur. The ninth embodiment has dealt with it.

Referring to FIG. 34, in S1000, it is determined whether or not the canister purging is being executed. If the result is affirmative, the flow proceeds to S1002 and thereafter, similarly to the case of KACTAVE as described in the first embodiment with reference to FIG. Find the average value. On the other hand, if the result in S1000 is NO, S10
Proceeding to step 12 and thereafter, the same processing as described in the first embodiment with reference to FIG. 9 is performed.

Since the ninth embodiment is configured as described above, the controllability is improved without being greatly affected by the specific cylinder even when the canister purge is executed.

Although not shown, the atmospheric pressure P
When a is low, that is, when the vehicle is located at a high altitude, at a low water temperature, or during a lean burn operation, and the combustion is in an unstable state, it is desirable to similarly obtain an average value, thereby improving controllability. Can be improved.

FIGS. 35 and 36 are a flow chart and a block diagram showing a tenth embodiment of the apparatus according to the present application.

Referring first to FIG. 36, the tenth
In the case of the embodiment, the PID is added to the configuration of the first embodiment.
Except for the feedback loop (correction coefficient KLAF) of the equivalent ratio of the exhaust system consisting of the control law, the feedback loop for each cylinder (correction coefficient #
nKLAF) was inserted.

That is, based on the output of a single air-fuel ratio sensor disposed in the exhaust system collecting section, the applicant of the present application described above in Japanese Patent Laid-Open No. 5-1 / 5-1.
The air-fuel ratio # nA / F (n: cylinder) of each cylinder is estimated using the observer proposed in Japanese Patent Application Publication No. 80040, and according to a deviation between the estimated value and a predetermined target value of the cylinder-specific air-fuel ratio F / B. A feedback correction coefficient #nKLAF for each cylinder is obtained using the PID control law, and the output fuel injection amount Tout is multiplied and corrected.

More specifically, the feedback correction coefficient #nKLAF for each cylinder is a value obtained by dividing the collective air-fuel ratio by the previously calculated value of the average value of the feedback correction coefficient #nKLAF for each cylinder (this is the above-mentioned value). Thus, the target air-fuel ratio F / B is referred to as a “target value of the cylinder-specific air-fuel ratio FCMD.” Therefore, the deviation between the target air-fuel ratio KCMD and the estimated observer air-fuel ratio # nA / F is eliminated. It is determined using the PID control rule. For details, refer to Japanese Patent Application No.
-251, 138, and the description is omitted. The illustration of the adhesion correction compensator is omitted.

Further, in the tenth embodiment, L
A sampling block (shown as Sel-VOBSV in the figure) for sampling the AF sensor output at an appropriate timing is provided, and a similar type of sampling block (shown as Sel-VSTR in the figure) is provided for the STR controller.

Here, the sampling block and the observer will be described. The sampling operation block is shown as "Sel-VOBSV" in FIG.

Since the exhaust gas is exhausted in the exhaust stroke in the internal combustion engine, the behavior of the air-fuel ratio in the exhaust system assembly of the multi-cylinder internal combustion engine is clearly synchronized with TDC. Therefore, when the above-described wide-range air-fuel ratio sensor is provided in the exhaust system of the internal combustion engine to sample the air-fuel ratio, the TDC
However, depending on the sample timing of the control unit (ECU) that processes the detection output, the behavior of the air-fuel ratio may not be accurately grasped. That is,
For example, the air-fuel ratio of the exhaust system collecting part with respect to TDC is
In this case, the air-fuel ratio recognized by the control unit has a completely different value depending on the sample timing as shown in FIG. In this case, it is desirable to perform sampling at a position where the actual change in the output of the air-fuel ratio sensor can be grasped as accurately as possible.

Further, the change in the air-fuel ratio also differs depending on the arrival time of the exhaust gas to the sensor and the reaction time of the sensor.
The arrival time at the sensor changes depending on the exhaust gas pressure, the exhaust gas volume, and the like. Further, since sampling is performed in synchronization with TDC based on the crank angle, it is inevitably affected by the engine speed. As described above, the detection of the air-fuel ratio largely depends on the operating state of the engine. For this purpose, for example, in the technology described in JP-A-1-313,644, it is determined whether the detection is appropriate or not at every predetermined crank angle. In addition to the possibility that the air-fuel ratio sensor may not be able to complete the detection, the inflection point of the output of the air-fuel ratio sensor may be missed when the detection is determined.

FIG. 39 is a flow chart showing the sampling operation of the LAF sensor. Since the detection accuracy of the air-fuel ratio is closely related to the estimation accuracy of the observer in particular, the description will be made before the description of FIG. Here, the air-fuel ratio estimation by the observer will be briefly described.

First, in order to accurately separate and extract the air-fuel ratio of each cylinder from the output of one LAF sensor, it is necessary to accurately clarify the detection response delay of the LAF sensor. Therefore, this delay is tentatively modeled as a first-order delay system, and a model as shown in FIG. 40 is created. Here LA
When F: LAF sensor output and A / F: input A / F, the state equation can be expressed by the following equation (30).

[0208]

[Equation 30]

If this is discretized by the period ΔT, it becomes as shown in Expression 31. FIG. 41 is a block diagram of Equation 31.

[0210]

(Equation 31)

Accordingly, the true air-fuel ratio can be obtained from the sensor output by using the equation (31). That is, when the expression 31 is transformed, the expression 31 becomes as shown in the expression 32.
The value at the time k-1 can be calculated backward from the value at the time as shown in Expression 33.

[0212]

(Equation 32)

[0213]

[Equation 33]

More specifically, if Equation 31 is expressed by a transfer function using Z-transformation, Equation 34 is obtained, and the inverse input transfer function is multiplied by the current LAF sensor output LAF to determine the previous input air-fuel ratio in real time. Can be estimated. FIG. 42 shows a block diagram of the real-time A / F estimator.

[0215]

(Equation 34)

Next, the method of separating and extracting the air-fuel ratio of each cylinder based on the true air-fuel ratio obtained as described above will be described. Considering the weighted average in consideration of the temporal contribution of the air-fuel ratio of the cylinder, the value at the time k is represented as in Expression 35. Note that "F / A" is used here because F (fuel amount) is a control amount, but "Air / fuel ratio" will be used in the following description for convenience of understanding unless there is a problem. Note that the air-fuel ratio (or the fuel-air ratio) means a true value obtained by correcting the response delay previously obtained by Expression 34.

[0219]

(Equation 35)

That is, the air-fuel ratio of the collecting portion is determined by adding the weight Cn to the past combustion history of each cylinder (for example, the most recently burned cylinder is 4%).
0%, before that 30%. . . , Etc.). This model is represented by a block diagram as shown in FIG.

The state equation is as shown in Equation 36.

[0220]

[Equation 36]

When the air-fuel ratio of the collecting portion is defined as y (k),
The output equation can be expressed as in Equation 37.

[0222]

(37)

In the above, since u (k) cannot be observed, even if an observer is designed from this equation of state, x
(K) cannot be observed. Therefore, if it is assumed that x (k + 1) = x (k−3) assuming that the air-fuel ratio before 4TDC (that is, the same cylinder) is in a steady operation state in which the air-fuel ratio does not suddenly change, Equation 38 is obtained.

[0224]

(38)

Here, simulation results are shown for the model obtained as described above. FIG. 44 shows a case where the air-fuel ratio of three cylinders is set to 14.7: 1 and the fuel is supplied to one cylinder at 12.0: 1 in the four-cylinder internal combustion engine. FIG.
Numeral 5 indicates the air-fuel ratio of the collecting portion at that time obtained by the above model. Although a step-like output is obtained in the same figure, if the response delay of the LAF sensor is further taken into consideration, the sensor output has a waveform that is smoothed as shown in FIG. 46 as “model output value”. In the figure, “actual measurement value” is an actual measurement value of the LAF sensor output in the same case, and it is verified by comparison with this that the above model models the exhaust system of the multi-cylinder internal combustion engine well.

Therefore, the problem is reduced to a normal Kalman filter problem of observing x (k) in the state equation and the output equation shown in Expression 39. When the Riccati equation is solved with the weight matrices Q and R as shown in Equation 40, the gain matrix K becomes as shown in Equation 41.

[0227]

[Equation 39]

[0228]

(Equation 40)

[0229]

[Equation 41]

From this, the A-KC is obtained as shown in Equation 42.

[0231]

(Equation 42)

The general observer configuration is as shown in FIG. 47, but since there is no input u (k) in this model, there is a configuration in which only y (k) is input as shown in FIG. , And this is represented by Equation 43.

[0233]

[Equation 43]

Here, an observer that receives y (k) as an input,
That is, the system matrix of the Kalman filter is represented by Equation 44.

[0235]

[Equation 44]

In the present model, when the elements of the weight distribution R of the Riccati equation: the elements of Q = 1: 1, the system matrix S of the Kalman filter is given by Equation 45.

[0237]

[Equation 45]

FIG. 49 shows a combination of the above-described model and observer. The simulation result is omitted since it is shown in the earlier application, but the air-fuel ratio of each cylinder can be accurately extracted from the air-fuel ratio of the collecting portion.

Since the air-fuel ratio of each cylinder can be estimated from the air-fuel ratio of the collecting section by the observer, the air-fuel ratio can be controlled for each cylinder using a control law such as PID. More specifically, as shown in FIG. 50 in which only the feedback portion of the observer in FIG. 36 is extracted, the collective feedback correction coefficient KLAF is obtained from the sensor output (collective air-fuel ratio) and the target air-fuel ratio using the PID control law. Then, a feedback correction coefficient #nKLAF (n: cylinder) for each cylinder is obtained from the observer estimated value # nA / F.

Feedback correction coefficient #nKLAF for each cylinder
More specifically, the target value and the observer estimated value #nA obtained by dividing the air-fuel ratio of the collecting section by the previously calculated value of the average value of the feedback correction coefficient #nKLAF for each cylinder for all cylinders are calculated.
It is determined using the PID rule so as to eliminate the deviation from / F.

As a result, the air-fuel ratio of each cylinder converges to the air-fuel ratio of the collecting portion, and the air-fuel ratio of the collecting portion converges to the target air-fuel ratio. As a result, the air-fuel ratio of all the cylinders becomes the target air-fuel ratio. Converge. Here, the fuel injection amount #nTout of each cylinder (specified by the valve opening time of the injector) is obtained by: # nTout = Tcyl × # nKLAF × KLAF.

Here, returning to the flow chart of FIG. 39, sampling of the LAF sensor output will be described. This program is started at the TDC position.

Description will be given below with reference to the flow chart of FIG. First, in S1200, the engine speed Ne, the intake pressure Pb, and the valve timing V / T are read out, and in S120
In step S1206, a timing map (described later) for HiV / T or LoV / T is retrieved, and in step S1208, the sensor output used for the observer calculation for Hi or Lo valve timing is sampled. In particular,
A timing map is searched from the engine speed Ne and the intake pressure Pb, and one of the twelve buffers is assigned to its No. And select the sampling value stored therein.

FIG. 51 is an explanatory diagram showing the characteristics of the timing map. As shown in the drawing, the characteristics are such that a value sampled at a faster crank angle is selected as the engine speed Ne is lower or the intake pressure (load) Pb is higher. Is set to Here, “early” means a value sampled at a position closer to the previous TDC position (in other words, an old value). Conversely, the higher the engine speed Ne or the lower the intake pressure Pb, the slower the crank angle, that is, the later TDC
A setting is made so that a value sampled at a crank angle close to the position (in other words, a new value) is selected.

That is, as shown in FIG. 38, it is best to sample the output of the LAF sensor at a position as close as possible to the inflection point of the actual air-fuel ratio. Assuming that the reaction time of the sensor is constant, the value occurs at a faster crank angle as the engine speed decreases, as shown in FIG. Also, it is expected that as the load increases, the exhaust gas pressure and the exhaust gas volume increase, and accordingly, the flow rate of the exhaust gas increases and the arrival time at the sensor is shortened. In that sense, the sample timing is shown in FIG.
The settings were as shown in FIG.

Further, regarding the valve timing, an arbitrary value Ne1 of the engine speed is set to Ne1-Lo, Hi for the Lo side.
Side is Ne-Hi, and any value of the intake pressure is Pb1-LO for the Lo side and Pb1-Hi for the Hi side.
Then, the map characteristics are Pb1-Lo> Pb1-Hi Ne1-Lo> Ne1-Hi. That is, in the case of HiV / T, the opening time of the exhaust valve is earlier than that of LoV / T, so that if the values of the engine speed or the intake pressure are the same, an earlier sampling value is selected. Is set.

Then, the flow advances to S1210 to perform the operation of the observer matrix on HiV / T, and then to S1212, the same operation is performed on LoV / T. Then S12
Proceeding to S14, the valve timing is judged again, and according to the judgment result, the process proceeds to S1216, S1218 to select the calculation result and end.

That is, since the behavior of the air-fuel ratio collecting portion changes with the switching of the valve timing, it is necessary to change the observer matrix. However, the estimation of the air-fuel ratio of each cylinder is not instantaneous, and the calculation of the air-fuel ratio of each cylinder requires several operations before the convergence is completed. The calculation using the changed observer matrix is performed in an overlapped manner, and even if the valve timing is changed, it can be selected according to the changed valve timing in S1214. After each cylinder is estimated, as described above, a feedback correction coefficient is obtained so as to eliminate the deviation from the target value, and the injection amount is determined.

With this configuration, the detection accuracy of the air-fuel ratio can be improved. That is, as shown in FIG. 53, since sampling is performed at relatively short intervals, the sampled value almost exactly reflects the sensor output, and the values sampled at relatively short intervals are sequentially stored in the buffer group. Further, the inflection point of the sensor output is predicted in accordance with the engine speed and the intake pressure (load), and a value corresponding thereto is selected from a group of buffers at a predetermined crank angle. Thereafter, an observer calculation is performed to estimate the air-fuel ratio of each cylinder, and as described with reference to FIG. 50, feedback control of the air-fuel ratio for each cylinder is also possible.

Therefore, as shown in the lower part of FIG.
The core 70 can accurately recognize the maximum value and the minimum value of the sensor output. Therefore, even when estimating the air-fuel ratio of each cylinder using the above-described observer with this configuration, a value approximating the actual behavior of the air-fuel ratio can be used, and the estimation accuracy of the observer is improved. The accuracy in performing the cylinder-by-cylinder air-fuel ratio feedback control described with respect to 50 is also improved. The details are described in detail in Japanese Patent Application No. 6-243277 previously proposed by the present applicant, and further description will be omitted.

The above is the sampling operation (shown as Sel-VOBSV in FIG. 36) performed by the observer on the LAF sensor output. The STR controller also performs the same sampling operation (shown as Sel-VSTR in FIG. 36).

That is, this Sel-VSTR is also obtained according to the same procedure as that performed by Sel-VOBSV, that is, the procedure shown in the flowchart similar to FIG. The Sel-VOBSV detects the air-fuel ratio at the most convenient timing for the cylinder-by-cylinder air-fuel ratio estimation by the observer (for example, the timing at which the observer weight coefficient C is optimal for the model). VSTR is set to Sel- so that it is the most convenient timing to operate the STR (for example, the detection timing of the air-fuel ratio most affected by the cylinder in the latest exhaust stroke).
The air-fuel ratio is detected using a map similar to that shown in FIG. 51 indicated by VOBSV.

The tenth embodiment will be described with reference to the flow chart of FIG. 35 on the premise of the above. Steps S1100 to S1110 similar to those of the first embodiment will be described.
To S1112, where LA by Sel-VSTR
The output of the F sensor is sampled, that is, the air-fuel ratio KACT (k) is detected. Next, the process proceeds to S1114, where the feedback correction coefficient KSTR is calculated as in the first embodiment. More specifically, the flow chart of FIG. 11 used in the first embodiment is used.
This is performed using a chart.

Subsequently, the flow advances to S1116, S1118 to calculate the required fuel injection amount Tcyl (k) and the output fuel injection amount Tout (k), and advances to S1120 to sample the LAF sensor output by Sel-VOBSV, ie, the equivalent ratio Detect KACT (k). Next, proceeding to S1122, the air-fuel ratio # nA / F of each cylinder is estimated via the observer described above, and proceeding to S1124, a feedback correction coefficient #nKLAF for each cylinder is calculated,
Proceeding to S1126, the learning value #nKLAFsty is obtained from the weighted average value with the previous value, etc., and proceeding to S1128, the output fuel injection amount Tout is determined by the feedback correction coefficient #nKLA for each cylinder.
The multiplication correction is performed by F to obtain the output injection amount #nTout of the cylinder, and the process proceeds to S1130 to perform the suction pipe wall adhesion correction.
Proceed to 1132 and output.

When the result in S1108 to S1110 is NO, the program proceeds to S1134, where the required fuel injection amount Tcyl (k) is obtained as shown in the figure. S1
Proceeding to 138, the learning value is set as the correction coefficient #nKLAF. If it is determined in S1104 that the fuel cut is performed, the process proceeds to S1146 via S1144 to stop the matrix calculation, and proceeds to S1148 to set the feedback correction coefficient for each cylinder to the previous value. The remaining steps are not different from the first embodiment.

Since the tenth embodiment is configured as described above, the input to the parameter adjustment mechanism is synchronized with the combustion cycle while calculating the adaptive parameters, as in the first embodiment. The computational load of the parameter adjustment mechanism is greatly reduced, and the adaptive controller can be used in an actual machine while ensuring controllability. At the same time, it is possible to reduce the variation between cylinders.

In the same manner as in the first embodiment, the average value of the air-fuel ratio KACT or the average value of the adaptive parameters for one cylinder during one combustion cycle is obtained and input to the parameter adjusting mechanism, and the output of the STR controller is obtained. Is also obtained, so that there is no great influence of the combustion state of the specific cylinder.

Incidentally, in the tenth embodiment, the adaptive parameter θ hat or KS is used in the same manner as in the second embodiment.
It goes without saying that the average value of TR may be obtained, or the average value of the air-fuel ratio KACT and the average value of the adaptive parameter θ hat may be obtained together. Further, the target air-fuel ratio KCMD (k) may be the same value for all cylinders.

Further, in the tenth embodiment, the second
Embodiment, the third embodiment, the fourth embodiment,
The descriptions of the fifth embodiment, the sixth embodiment, the seventh embodiment, the eighth embodiment, and the ninth embodiment are all valid.

FIGS. 54 and 55 are a flow chart and a block diagram showing an eleventh embodiment of the device according to the present application.

In the case of the eleventh embodiment, as shown in FIG. 55, an STR controller and a parameter adjusting mechanism are inserted in series in a fuel injection amount calculation system. That is, similarly to the first embodiment, the basic fuel injection amount Tim is multiplied by the target air-fuel ratio correction coefficient KCMDM (k) and various correction coefficients KTOTAL to obtain the required fuel injection amount Tcyl (k), and then the correction is performed. Required fuel injection amount T
cyl (k) is input to the STR controller.

On the other hand, the average value KACTAVE or θ hat AVE is obtained from the detected exhaust system air-fuel ratio in the same manner as in the first embodiment, and the STR is calculated with respect to the required fuel injection amount Tcyl (k).
Dynamic correction is performed by the controller, and the corrected fuel injection amount Gfuel
-Calculate str (k).

At the same time, the feedback correction coefficient of the collecting part is calculated from the detected air-fuel ratio of the collecting part using the PID control law.
KLAF is obtained and multiplied by the required fuel injection amount Tcyl (k) to calculate a corrected fuel injection amount Gfuel-KLAF (k).

In FIG. 55, the STR controller determines the actual intake fuel amount (more precisely, the estimated intake fuel amount) Gfuel (k)
Is adaptively calculated so that the output fuel injection amount Gfuel-str (k) matches the target fuel amount Tcyl (k), and the output fuel injection amount Tou is calculated.
It is supplied to the internal combustion engine as t (k). The correction of wall adhesion in a virtual plant is described in Japanese Patent Application No. 4-2 proposed by the present applicant.
This is described in detail in JP-A-00331 (Japanese Patent Application Laid-Open No. 6-17681), and since the gist of the present invention is not there, the explanation is omitted.

Here, the actual intake fuel amount Gfuel (k) can be obtained by dividing the detected air amount by the detected air-fuel ratio. However, in the case of the embodiment, an air amount detector (air flow meter) is used. ), The target intake fuel amount (required injection amount) Tcyl (k) is multiplied by the detected air-fuel ratio.
As a result, the actual intake fuel amount can be obtained in a manner equivalent to detecting and obtaining the air amount. As mentioned earlier,
In this control, the target air-fuel ratio and the detected air-fuel ratio are actually expressed as equivalent ratios.

When the target air-fuel ratio is not the stoichiometric air-fuel ratio, the calculated value is further divided by the target air-fuel ratio to obtain the actual intake fuel amount. That is, when the target air-fuel ratio is the stoichiometric air-fuel ratio, the actual intake fuel amount is calculated as: actual intake fuel amount = required injection amount (target intake fuel amount) × detected air-fuel ratio (equivalent ratio). When the fuel ratio is other than the fuel ratio, the actual intake fuel amount = (required injection amount (target intake fuel amount) × detected air-fuel ratio (equivalent ratio)) / target air-fuel ratio (equivalent ratio).

The above will be described with reference to the flowchart of FIG. 54.
The process proceeds to S1318 through S1316 through S1316, and calculates the average value KACTAVE of the air-fuel ratio and the average value θhat-AVE of the adaptive parameter θhat.

Subsequently, the flow advances to S1324 via S1320 to S1322 to determine the instability of the adaptive control system (STR controller) as in the first embodiment.

FIG. 56 is a subroutine flowchart showing the operation.

The explanation will be given with reference to FIG.
Then, the stability of the STR control system is determined using each element of the adaptive parameter θ hat.

More specifically, the fuel injection amount Gfuel-STR (k) calculated by the STR controller is calculated as shown in Equation 46.

[0272]

[Equation 46]

Here, assuming that the adhesion correction is correct,
The transfer function of the virtual plant is as shown in Equation 47.

[0274]

[Equation 47]

From Equations 46 and 47, the transfer function from Tcyl (k) to the injection amount Gfuel-STR (k) is as shown in Equation 48.

[0276]

[Equation 48]

Here, since b0 is a scalar quantity for determining the gain, it cannot be 0 or negative. Therefore, the denominator function f (z) of the transfer function of Equation 48 = b0z ³ + r1z ² + r
2z + r3 + s0 is one of the functions shown in FIG. Therefore, if it is determined whether or not the real root is within the unit circle, that is, as shown in FIG. 15, it is determined whether or not f (−1) <0 to f (1)> 0, When the result is affirmative, the real root is within the unit circle, so that it can be easily determined whether the system is stable or not.

Then, in S1402, the STR
It is determined whether or not the controller system is unstable. If the determination is affirmative, the process proceeds to S1404 to return the adaptive parameter θ hat to the initial value. Thereby, the stability of the system can be restored. Then, the process proceeds to S1406, where the gain matrix Γ is corrected. Since the gain matrix す る determines the convergence speed, this correction is performed so as to reduce the convergence speed, whereby the stability of the system can be similarly restored. Then, the process proceeds to S1408, where the correction coefficient KL based on the PID control law is used as a feedback correction coefficient as shown in the figure.
The output fuel injection amount Tout (k) is determined by using AF (k), using the corrected fuel injection amount Gfuel-KLAF, and adding the addition term TTOTAL thereto.

If it is determined in step S1402 that the STR controller system is not unstable, the flow advances to step S1410 to correct fuel injection using the adaptive control law correction coefficient KSTR (k) as a feedback correction coefficient as shown. Quantity G
The output fuel injection amount Tout (k) is determined by using fuel-str (k) and adding an addition term TTOTAL thereto.

Returning to the flow chart of FIG. 54, the program proceeds to 1326, where the output fuel injection amount is output, and the flow ends. In the case of the eleventh embodiment, the calculation of the average value such as the air-fuel ratio is not limited to the predetermined crank angle of the specific cylinder, and may be performed at the predetermined crank angle of each cylinder, unlike the previous embodiments. . The remaining configuration is not different from the previous embodiment.

In the eleventh embodiment, the configuration is as described above, and the input to the parameter adjustment mechanism may be synchronized with the combustion cycle while calculating the adaptive parameters, as in the first embodiment. In this method, the operation load of the parameter adjustment mechanism is greatly reduced, and the adaptive controller can be used for the actual machine while maintaining controllability. Further, the controllability can be improved by shortening the dead time.

Also in the eleventh embodiment, since the average value of the control amounts of all the cylinders is obtained and input to the parameter adjustment mechanism, the influence of the combustion state of the specific cylinder is not greatly affected.

As described above, in this embodiment,
Fuel injection amount control means for controlling the fuel injection amount of the multi-cylinder internal combustion engine, an adaptive controller for adaptively matching the fuel injection amount to a target value as an operation amount, and calculating an adaptive parameter used in the adaptive controller And a gain matrix 決定 for determining a convergence speed of the adaptive parameter θ hat according to an operation state of the multi-cylinder internal combustion engine. Configured.

In the first to eleventh embodiments, the simple average is shown as the average. However, the present invention is not limited to this, and a weighted average, a moving average, a weighted moving average, or the like may be used. . Further, the average value during one combustion cycle in which the input to the parameter adjustment mechanism is performed in synchronization is obtained, but the average value before two combustion cycles may be obtained.
Alternatively, an average value of less than one combustion cycle, for example, between 2 and 3 TDC may be obtained.

As described above, it is natural that the Sel-VOBSV and the Sel-VSTR are separately provided, and the optimum air-fuel ratio may be detected for each. However, depending on the characteristics of the engine and the layout of the exhaust system, the Sel-VOBSV may be used. Since Sel-VSTR and Sel-VSTR show almost the same detected air-fuel ratio in most operating regions, in such a case, these sampling functions are unified to detect the air-fuel ratio, and the output is output to both observer and STR. May be used as input. For example, only Sel-VOBSV in FIG. 36 may be used, and the output may be used for the observer and the STR.

Although the equivalent ratio is actually used as the air-fuel ratio in the first embodiment and the like, it goes without saying that the air-fuel ratio and the equivalent ratio may be determined separately. Further, the feedback correction coefficient KSTR, #nKLAF, KLAF is obtained as a multiplication term, but may be obtained as an added value.

Further, in the above description, ST is used as the adaptive controller.
Although R has been described as an example, MRACS (model reference adaptive control) may be used.

In the above description, the output of a single air-fuel ratio sensor provided in the exhaust system collecting section is used. However, the present invention is not limited to this. The air-fuel ratio is detected by providing an air-fuel ratio sensor for each cylinder. The air-fuel ratio feedback control may be performed every time.

[0289]

According to the first aspect, the convergence speed of the adaptive parameter can be optimized according to the operating state,
Thus, controllability can be improved.

According to the present invention, the convergence speed of the adaptive parameters can be optimized even when the engine load changes, for example, when the engine load suddenly changes, for example, especially when the engine is decelerated. Can be improved.

In the present invention, the convergence speed of the adaptive parameter can be optimized particularly in the idling operation state, so that the controllability can be improved.

In the present invention, the convergence speed of the adaptive parameter can be optimized, especially in the operating state in which the exhaust gas recirculation is performed, so that the controllability can be improved.

In the fifth aspect, in particular, the canister
In the operating state in which the purge is performed, the convergence speed of the adaptive parameter can be optimized, so that the controllability can be improved.

According to the sixth aspect, especially in an operation state in which the atmospheric pressure is largely different from the normal pressure, the convergence speed of the adaptive parameter can be optimized, so that the controllability can be improved.

In the present invention, the convergence speed of the adaptive parameters can be optimized especially in an operation state in which the water temperature is significantly different from the normal temperature, so that the controllability can be improved.

According to the present invention, the convergence speed of the adaptive parameters can be optimized, and the controllability can be improved, especially when the valve timing is switched.

[Brief description of the drawings]

FIG. 1 is a schematic diagram showing a fuel injection control device for an internal combustion engine according to the present application as a whole.

FIG. 2 is an explanatory diagram showing details of an exhaust gas recirculation mechanism in FIG. 1;

FIG. 3 is an explanatory diagram showing details of a canister / purge mechanism in FIG. 1;

FIG. 4 is an explanatory diagram showing valve timing characteristics of the variable valve timing mechanism in FIG.

FIG. 5 is a block diagram showing details of a control unit in FIG. 1;

FIG. 6 is a main flow chart showing the operation of the fuel injection control device for an internal combustion engine according to the present application.

FIG. 7 is a block diagram functionally showing the operation of the flow chart of FIG. 6;

FIG. 8 is a timing chart showing an example of the operation of the adaptive controller used in the fuel injection control device for an internal combustion engine according to the present application.

FIG. 9 is a timing chart showing another example of the operation of the adaptive controller used in the fuel injection control device for an internal combustion engine according to the present application.

10 is a block diagram in which the configuration of the block diagram in FIG. 6 is rewritten focusing on the STR controller and the adaptive parameter adjustment mechanism.

FIG. 11 is a subroutine flowchart showing an operation of calculating an average value such as a feedback correction coefficient according to the adaptive control law of the flowchart of FIG. 6;

FIG. 12 is a timing chart for explaining a calculation operation of the flow chart of FIG. 11;

FIG. 13 is a subroutine flowchart illustrating the determination of instability of the adaptive control system in the flowchart of FIG. 6;

FIG. 14 is an explanatory diagram illustrating the determination of instability in the flow chart of FIG. 13;

FIG. 15 is an explanatory diagram similar to FIG. 14, illustrating an instability determining operation in the flow chart of FIG. 13;

FIG. 16 is a timing chart showing another example of the operation of the adaptive controller similar to that of FIG. 8;

FIG. 17 is a timing chart showing another example of the operation of the adaptive controller similar to that of FIG. 8;

FIG. 18 is a flowchart showing a second embodiment of the apparatus according to the present application.

FIG. 19 is an explanatory diagram showing characteristics of a map used in the flow chart of FIG. 18;

FIG. 20 is an explanatory diagram showing characteristics of a table used in the flowchart of FIG. 18;

FIG. 21 is an explanatory diagram showing characteristics of a table similar to FIG. 20 used in the flow chart of FIG. 18;

FIG. 22 is an explanatory diagram showing characteristics of a table similar to FIG. 20 used in the flowchart of FIG. 18;

FIG. 23 is an explanatory diagram showing characteristics of a table similar to FIG. 20 used in the flowchart of FIG. 18;

FIG. 24 is a flowchart showing a third embodiment of the apparatus according to the present application.

FIG. 25 is a flow chart showing a fourth embodiment of the apparatus according to the present application.

FIG. 26 is an explanatory diagram showing characteristics of a dead zone used in the flow chart of FIG. 25;

FIG. 27 is a flow chart showing a fifth embodiment of the apparatus according to the present application.

FIG. 28 is an explanatory diagram showing characteristics of a limiter used in the flowchart of FIG. 27;

FIG. 29 is a flowchart showing a sixth embodiment of the apparatus according to the present application.

FIG. 30 is an explanatory diagram showing characteristics of a map used in the flow chart of FIG. 29;

FIG. 31 is a flowchart showing a seventh embodiment of the apparatus according to the present application.

FIG. 32 is an explanatory diagram for explaining the operation of the flowchart shown in FIG. 31;

FIG. 33 is a flow chart showing an eighth embodiment of the apparatus according to the present application.

FIG. 34 is a flowchart showing a ninth embodiment of the device according to the present application.

FIG. 35 is a flowchart showing a tenth embodiment of the device according to the present application.

FIG. 36 is a block diagram for explaining the operation of the flowchart in FIG. 35;

FIG. 37 is an explanatory diagram showing a relationship between TDC of a multi-cylinder internal combustion engine and an air-fuel ratio of an exhaust system assembly.

FIG. 38 is an explanatory diagram showing quality of sample timing with respect to an actual air-fuel ratio.

FIG. 39 is a flowchart showing the sampling operation of the air-fuel ratio in the Sel-V block in the block diagram of FIG. 36.

FIG. 40 is an explanatory view 1 of the observer in the block diagram of FIG. 36;
FIG. 4 is a block diagram showing an example in which the detection operation of the air-fuel ratio sensor described in the earlier application is modeled.

FIG. 41 shows a model obtained by discretizing the model shown in FIG. 40 with a period ΔT.

FIG. 42 is a block diagram showing a true air-fuel ratio estimator that models the detection behavior of an air-fuel ratio sensor.

FIG. 43 is a block diagram showing a model showing the behavior of the exhaust system of the internal combustion engine.

FIG. 44 shows a case in which fuel is supplied to the four-cylinder internal combustion engine by setting the air-fuel ratio of three cylinders to 14.7: 1 and the air-fuel ratio of one cylinder to 12.0: 1 using the model shown in FIG. 43. It is a data diagram.

FIG. 45 is a data diagram showing the air-fuel ratio of the aggregate of the model of FIG. 43 when the input shown in FIG. 44 is given.

46 compares the air-fuel ratio of the collective part of the model of FIG. 43 when the input shown in FIG. 44 is given in consideration of the response delay of the LAF sensor, and the measured value of the LAF sensor output in the same case. FIG.

FIG. 47 is a block diagram showing a configuration of a general observer.

FIG. 48 is a block diagram showing a configuration of the observer used in the earlier application in the observer shown in the block diagram of FIG. 36;

FIG. 49 is an explanatory block diagram showing a configuration obtained by combining the model shown in FIG. 43 and the observer shown in FIG. 48;

FIG. 50 is a block diagram showing feedback control of the air-fuel ratio in the block diagram of FIG. 36;

FIG. 51 is an explanatory diagram showing characteristics of a timing map used in the flow chart of FIG. 39;

FIG. 52 is an explanatory diagram illustrating sensor output characteristics with respect to the engine speed and the engine load, for explaining the characteristics of FIG. 51.

FIG. 53 is a timing chart illustrating a sampling operation in the flow chart of FIG. 39;

FIG. 54 is a flowchart showing an eleventh embodiment of the apparatus according to the present application.

FIG. 55 is a block diagram illustrating the operation of the flow chart of FIG. 54.

FIG. 56 is a subroutine flowchart showing the operation of determining the instability of the adaptive control system in the flowchart of FIG. 54;

FIG. 57 is a timing chart illustrating dead time in calculating the fuel injection amount of the internal combustion engine.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 10 Internal combustion engine 12 Intake pipe 20 Intake manifold 22 Injector 24 Exhaust manifold 26 Exhaust pipe 28 Catalyst device 34 Control unit 54 Wide area air-fuel ratio sensor (LAF sensor) 100 Exhaust recirculation mechanism 200 Canister purge mechanism 300 Variable valve timing mechanism 400 Switching mechanism

──────────────────────────────────────────────────続 き Continuation of the front page (51) Int.Cl. ⁷ Identification symbol FI G05B 11/32 G05B 11/32 F 13/02 13/02 A (72) Inventor: Koichi Nishimura 1-4-4 Chuo, Wako-shi, Saitama No. 1 Inside Honda R & D Co., Ltd. (56) References JP-A-6-42385 (JP, A) JP-A-59-196944 (JP, A) JP-A-4-365947 (JP, A) JP-A-1-155046 (JP, A) JP-A-4-342844 (JP, A) JP-A-59-54752 (JP, A) JP-A-4-209940 (JP, A) A) (58) Field surveyed (Int. Cl. ⁷ , DB name) F02D 41/00-41/40 F02D 43/00-45/00

Claims (8)

(57) [Claims] 1. A method according to claim 1, Fuel injection amount control means for controlling the fuel injection amount of the multi-cylinder internal combustion engine; b. Using the adaptive parameter with the fuel injection amount as the manipulated variable
At least one of the detected air-fuel ratio and the actual intake fuel amount.
Less becomes the control amount and the adaptive controller for one Itasa the target value, and c. At least the adaptive parameters used in the adaptive controller
An adaptive parameter adjustment mechanism that also calculates based on the control amount, a fuel injection control device for a multi-cylinder internal combustion engine comprising:
A fuel injection control device for an internal combustion engine, wherein a gain matrix for determining a convergence speed of an adaptive parameter calculated by the adaptive parameter adjusting mechanism is determined according to an operation state of the multi-cylinder internal combustion engine. 2. A fuel injection control device for an internal combustion engine according to claim 1, wherein a gain matrix for determining a convergence speed of said adaptive parameter is determined according to a load state of said multi-cylinder internal combustion engine. 3. The multi-cylinder internal combustion engine according to claim 2, wherein a gain matrix for determining a convergence speed of the adaptive parameter is different between when the multi-cylinder internal combustion engine is in an idle operation state. Fuel injection control device. 4. The multi-cylinder internal combustion engine includes an exhaust gas recirculation mechanism, and a gain matrix for determining a convergence speed of the adaptive parameter is made different between when and when exhaust gas recirculation is performed. 2. The fuel injection control device for an internal combustion engine according to claim 1. 5. The multi-cylinder internal combustion engine includes a canister purge mechanism, and a gain matrix for determining a convergence speed of the adaptive parameter is made different between when and when canister purging is performed. The fuel injection control device for an internal combustion engine according to claim 1. 6. The fuel injection control device for an internal combustion engine according to claim 1, wherein the operating state is an atmospheric pressure. 7. The fuel injection control device for an internal combustion engine according to claim 1, wherein the operation state is an engine cooling water temperature. 8. The fuel injection control of an internal combustion engine according to claim 1, wherein the multi-cylinder internal combustion engine includes a variable valve timing mechanism, and the operating state is a valve timing of the variable valve timing mechanism. apparatus.