JPH11223145A - Air-fuel ratio control device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To perform an accurate air-fuel ratio control by estimating a transient state amount at the time of starting of an engine and calculating a fuel injection amount at the time of starting using the estimated value in the configuration of a neuro-engine model at a starting time such as a low temperature starting time in which input term of an air-fuel ratio sensor is absent. SOLUTION: The air-fuel ratio control device comprises a state detecting part 12 for detecting a parameter representing an operating state of an engine 11 which is an internal combustion engine, a counter detecting part 13 for counting the number of combustion in a cylinder after starting of the engine 11, and an air-fuel ratio estimating means 14 which estimates an air-fuel ratio immediately after the starting of the engine 11 from the parameter representing the operating state and the number of combustion in the cylinder. An intake pipe temperature rising change is handled associating with the number of combustion in the cylinder, and air-fuel ratio control is carried out by estimating a transient state amount at the time of starting of engine.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a gasoline engine of a fuel injection control type for an internal combustion engine, and more particularly to a neural network in which knowledge is accumulated in the network itself and the network itself operates according to the purpose and environment. To improve an air-fuel ratio control device for controlling an air-fuel ratio of an engine.

[0002]

2. Description of the Related Art Conventionally, in air-fuel ratio control, feedback control is generally performed using an air-fuel ratio sensor, such as a 02 sensor, or a linear air-fuel ratio sensor (LAF sensor). ing. However, in a transient state such as acceleration or deceleration, the air-fuel ratio cannot be accurately controlled to the target air-fuel ratio due to a response delay of the sensor output. To compensate for this, the fuel is increased or decreased in response to mechanical changes such as the throttle opening.However, not all of the injected fuel flows into the cylinder, but the wall of the intake pipe. It is difficult to control the air-fuel ratio in a transient state such as during acceleration / deceleration or engine start because some of the fuel adheres to the fuel tank and intake valves and evaporates from the attached fuel and enters the cylinder.

In particular, US ULEV (Ultra Low Emission)
In order to clear the regulations, the amount of HC emission at the time of starting the engine occupies about 80% of all the test modes. Therefore, a highly accurate air-fuel ratio control at the time of starting the engine is indispensable.

Therefore, a fuel adhesion model is constructed, and a fuel correction amount is obtained from an inverse system of the adhesion model. As shown in Japanese Patent Application Laid-Open No. 3-235723, a nonlinear element such as the fuel adhesion is used as a neural network. (Hereinafter abbreviated as NN) to improve the response performance in a transient state. In the NN, "units" for performing simple calculations are connected by weighted "directional links" to form a network, and each unit performs information processing by propagating an output via the link. ,
Since knowledge is accumulated in the network itself and the network itself operates according to the purpose and environment, it can be expected that the air-fuel ratio control during the transition can be performed with high accuracy.

[0005]

The conventional air-fuel ratio control device is constructed as described above. However, in an internal combustion engine, usually, even when fuel is injected from an injector, all the fuel is injected into a cylinder. Instead of flowing in, some adhere to the intake pipe wall. The amount of adhesion varies in a complicated manner depending on the operating state (rotation speed, load (intake air pressure), etc.) and the external environment (intake air temperature, cooling water temperature, atmospheric pressure, etc.), and evaporates from the attached fuel and flows into the cylinder. The amount of fuel also varies depending on the operating conditions and the external environment. Therefore, if the amount of fuel actually flowing into the cylinder can be known, especially during transition,
More accurate air-fuel ratio control can be performed. However, in the adhesion model method, such a complicated system cannot be expressed, and can be given only by approximation. Therefore, satisfactory air-fuel ratio control cannot be performed in a transient state.

Further, in the control system using the NN, complicated behavior can be learned, but in order to obtain an estimated value having generalization, it is necessary to provide an air-fuel ratio sensor output in the NN input term. Although it is necessary, when the air-fuel ratio sensor is inactive immediately after the engine is started at an extremely low temperature or when the engine is started, the correction control using the NN that calculates the output value of the air-fuel ratio sensor as input data cannot be performed, and there is generalization. It was very difficult to estimate the air-fuel ratio with high accuracy.

SUMMARY OF THE INVENTION The present invention has been made to solve the above problems, and has no input term of the air-fuel ratio sensor.
In the N configuration, the transient state quantity at the time of engine start is predicted and estimated,
It is an object of the present invention to provide an air-fuel ratio control device capable of calculating a fuel injection amount at the time of starting using this estimated value and performing highly accurate air-fuel ratio control.

Further, an NN engine model for starting which expresses dynamic characteristics at the time of engine starting is constructed, and an air-fuel ratio control device capable of performing fuel injection control based on this model so that the air-fuel ratio at engine starting becomes a target air-fuel ratio. The purpose is to provide.

[0009]

An air-fuel ratio control device according to a first aspect of the present invention is a device for controlling an air-fuel ratio at the time of fuel injection of an internal combustion engine, and detects a parameter representing an operation state of the internal combustion engine. A state detector that counts the number of explosions in a cylinder after the engine is started, and an air-fuel ratio estimator that estimates an air-fuel ratio immediately after the start of the internal combustion engine from the operating state parameter and the number of in-cylinder explosions. It is provided with.

The air-fuel ratio control device according to claim 2 of the present invention is the air-fuel ratio control device according to claim 1, wherein the air-fuel ratio estimating means is configured to determine the operating state parameter and the number of cylinder explosions. The air-fuel ratio immediately after the start of the internal combustion engine is estimated by a neural network in which the input parameters are output and the output is the air-fuel ratio at engine start.

An air-fuel ratio control device according to a third aspect of the present invention is a device for controlling an air-fuel ratio at the time of fuel injection of an internal combustion engine. And a correction parameter calculating section for calculating an engine start fuel correction parameter from the Toutnn, and a fuel injection at start using the engine start fuel correction parameter. And an engine start fuel calculation unit for calculating the amount.

According to a fourth aspect of the present invention, there is provided an air-fuel ratio control apparatus for controlling an air-fuel ratio at the time of fuel injection of an internal combustion engine. Is used to determine the fuel injection amount Toutnn at which the air-fuel ratio becomes the target air-fuel ratio.
An engine start adhesion parameter calculating unit for calculating an engine start adhesion parameter from an intake pipe adhesion model at engine start; and an engine start fuel from the engine start adhesion parameter and the intake pipe adhesion model at engine start. And an engine starting fuel calculating section for calculating an injection amount.

According to a fifth aspect of the present invention, there is provided an air-fuel ratio control apparatus according to the fourth aspect, wherein the engine-start-time adhesion parameter calculating section calculates the air-fuel ratio dynamic characteristic at the time of engine start. Using the neuro engine model to be expressed, the parameters of the cylinder adhesion model at the time of engine start are derived, and the engine start fuel calculation unit is:
An intake pipe and a cylinder adhesion parameter are used as the engine start adhesion parameter, and a fuel injection amount at the time of engine start is calculated from these parameters and the intake pipe adhesion model and cylinder adhesion model at the time of engine start.

According to a sixth aspect of the present invention, there is provided an air-fuel ratio control apparatus according to the fifth aspect, wherein the engine start-time adhesion parameter calculation unit calculates an air-fuel ratio dynamic characteristic at the time of engine start. The parameters of the intake pipe model and the cylinder adhesion model at the time of engine start taking into account the evaporation temperature of gasoline are derived using the expressed neuro engine model.

According to a seventh aspect of the present invention, in the air-fuel ratio control device according to the first aspect, the counter detection unit is provided with a top dead center from cranking from the start of engine start. The number of explosions in the cylinder is counted based on the position (TDC) count number.

The air-fuel ratio control device according to claim 8 of the present invention is the air-fuel ratio control device according to claim 1, wherein the counter detection unit is configured to determine the number of explosions in the cylinder from the start of engine start. The counting is performed based on the output of an internal pressure sensor that detects the pressure in the cylinder.

[0017]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An air-fuel ratio control device according to the present invention will be described below. Embodiment 1 FIG. FIG. 1 shows a functional block diagram of an air-fuel ratio control device according to Embodiment 1 of the present invention. Embodiment 1
Is that the air-fuel ratio (A / F) value immediately after the start of the engine greatly affects the sudden change in the cylinder temperature. Therefore, it is necessary to add an input relating to this temperature change. Focusing on the fact that it depends on the number of explosions in the cylinder, the number of crankings from the start of the engine, that is, Tcount, is used as one of the parameters indicating the number of in-cylinder explosions.

In FIG. 1, reference numeral 11 denotes an engine such as gasoline as an internal combustion engine, 12 denotes a state detecting section for detecting the operating state of the engine 11, and 13 denotes a counter detecting section for detecting the number of cranking of the engine. A TDC (compression top dead center) of a piston is detected by using a crank angle sensor provided in a fuel injection type internal combustion engine. 14 is an air-fuel ratio estimating means for estimating the air-fuel ratio at the time of starting the engine based on various parameters, and 15 is a compensator for calculating the fuel injection amount so as to reach the target air-fuel ratio.

The operation will be described below. Engine 1
In order to estimate the air-fuel ratio at the time of the start-up of
Thus, the fuel injection amount Gf, the intake air pressure Pb, the rotation speed Ne, and the cooling water temperature TW, which are parameters representing the operation state of the internal combustion engine, are detected. The count Tcount is detected.

Using these five parameters, the air-fuel ratio estimating means 14 estimates the air-fuel ratio at the time of starting the engine,
The compensator 15 calculates the fuel injection amount at which the estimated value of the air-fuel ratio becomes the target air-fuel ratio, and performs the air-fuel ratio control at the time of starting the engine.

As shown in FIG. 2, the air-fuel ratio estimating means 14 controls the various operating state parameters (Gf, P
b, Ne, TW) and the TDC count number Tcount as their input parameters, and the air-fuel ratio (A / Fnn) may be estimated by a neural network (NN) whose output is the air-fuel ratio at engine start. In this case, if the time series data of the operating state parameter is used as the input data of the NN, the accuracy of estimating the air-fuel ratio at the time of transition can be improved.

In the first embodiment, the case where TDC is detected by using a crank angle sensor as the counter detector 13 has been described. However, an internal pressure sensor used to measure the pressure in the cylinder is used. You may. T
While DC detection counts misfires, etc., internal pressure sensors can be used to reliably detect explosions in cylinders and provide more accurate and reliable air-fuel ratio control. Can be.

As described above, according to the first embodiment, the state detecting unit 12 for detecting the parameter representing the operating state of the engine 11 which is an internal combustion engine, and for counting the number of in-cylinder explosions after the engine 11 starts. Counter detector 1
3 and air-fuel ratio estimating means 14 for estimating the air-fuel ratio immediately after the start of the engine 11 from the parameters representing the operating state and the number of in-cylinder explosions, so that the intake pipe temperature rise change is handled in association with the number of in-cylinder explosions. Therefore, even in the configuration of the start-up neuro engine model having no input term of the air-fuel ratio sensor, it is possible to predict and estimate the transient state amount at the time of engine start, and calculate the fuel injection amount at start-up using this estimated value. Highly accurate air-fuel ratio control can be performed.

Embodiment 2 FIG. Next, Embodiment 2 of the present invention
Will be described. In a control system using an NN that outputs an air-fuel ratio, feedback control is performed so that a deviation from a target air-fuel ratio is zero. At this time, there is no theoretical method for setting a stable control gain. A problem arises in that the control gain must be adjusted by repeating the experiment.

Therefore, in the second embodiment, the air-fuel ratio dynamic characteristics at the time of engine start are learned by neuro, and the fuel injection at the start air-fuel ratio becomes the target air-fuel ratio using the NN engine model obtained in this way. It is designed to calculate the quantity.

Hereinafter, a method of calculating the fuel injection amount will be described. FIG. 3 shows a flow chart for calculating a fuel correction parameter Kgf at the time of engine start in the air-fuel ratio control device according to the third embodiment.

First, at step 301, a neuro engine model of the output air-fuel ratio A / F expressing the air-fuel ratio dynamic characteristics at the time of engine start is learned. Then, at step 302, the neuro engine model and the evaluation actual vehicle data are used. The target fuel injection amount G at which the air-fuel ratio becomes the target air-fuel ratio
fnn is calculated. Next, at step 303, a fuel correction parameter Kgf at the time of starting the engine is calculated from the target fuel injection amount Gfnn and the fuel injection amount Gf of the evaluation actual vehicle data.

At this time, as shown in FIGS. 4A and 4B, the neural engine model NN1 whose output is the air-fuel ratio (FIG. 4A)
Alternatively, the neural network NN2 whose output is the fuel injection amount Gf (FIG. 4B) may be used.

FIG. 5 shows a method of deriving the target fuel injection amount Gfnn at which the air-fuel ratio becomes the target air-fuel ratio when each neuro engine model is used. FIG. 5A shows a neural network NN whose output is learned as an air-fuel ratio.
FIG. 4 is a diagram illustrating a method of calculating a target fuel injection amount Gfnn using No. 1, and an output A / Fnn (k) of a neural network NN1 using an input data sequence at time k becomes a target air-fuel ratio A / Ft. As described above, the fuel injection amount Gf, which is an input of the neural network NN1, is corrected, and the target fuel injection amount Gfnn (k) is calculated.

In FIG. 5B, since the output is the fuel injection amount Gf, the air-fuel ratio A / F of the input data is changed to the target air-fuel ratio A / F.
The target fuel injection amount Gfnn is calculated as the output of the neural network NN2 by changing to t and using the input data sequence of the neural network NN2.

When the target fuel injection amount Gfnn is calculated by any of the above methods, a correction parameter Kgf for correcting the original fuel injection amount Gf is obtained by the following equation, for example.

[0032]

(Equation 1)

A method of directly calculating the correction parameter Kgf as an output of the neural network may be used. FIG. 6 shows a method of learning the neural network NN3 that outputs the correction parameter Kgf. FIG. 7 shows a control system configuration using the neural network NN3.

An operating state parameter is detected from the engine 71, and a neural operation of the neural network NN3 is performed by a correction parameter calculating section 72 to calculate a correction parameter Kgf. Using the correction parameter Kgf, the target fuel injection amount Gfnn at which the air-fuel ratio at the time of starting the engine becomes the target air-fuel ratio is calculated by the starting fuel calculating unit 73.

When the correction parameter Kgf is given as a map value such as time-series data, the correction parameter calculator 72 performs a map search. With the above configuration, it is possible to uniquely calculate the fuel injection amount at which the air-fuel ratio at the time of starting the engine becomes the target air-fuel ratio.

Embodiment 3 Next, Embodiment 3 of the present invention
Will be described. As described above, not all the injected fuel flows into the cylinder and contributes to combustion, and a part of the fuel adheres to the intake pipe wall surface. This amount of adhesion is not particularly problematic because it is almost constant in a steady state after the engine is warmed up, but it becomes a problem at the time of transition, such as immediately after the start of the engine. In particular, when the engine is started at a low temperature, the fuel injection amount is generally injected more than usual in order to avoid a misfire or the like, and the operating state changes rapidly, such as an increase in the engine (intake pipe) temperature. Therefore, the fuel adhesion dynamics also change greatly. For this reason, it is necessary to perform fuel injection amount control at the time of starting the engine so that the air-fuel ratio becomes the target air-fuel ratio.

FIG. 8 shows a conceptual diagram of fuel adhesion.
8, reference numeral 80 denotes a fuel injection nozzle (injector) for injecting fuel into an intake pipe 81; 83, a cylinder block accommodating a piston 84 therein; 86, an exhaust pipe connected to the cylinder block 83; 84
A combustion chamber 87 is formed in a space inside a cylinder block 83 surrounded by the intake valve 82 and the exhaust valve 85. The fuel injection amount is Gf, the direct ratio representing the ratio of the fuel injection amount directly flowing into the cylinder is X, the fuel amount attached to the intake pipe is Mf, and the dynamic characteristics of evaporating from the fuel attached to the intake pipe and flowing into the cylinder are expressed. Assuming that the evaporation time constant is τ and the amount of fuel flowing into the cylinder is Gfe, the fuel adhesion model can be expressed by the following equation.

[0038]

(Equation 2)

Then, the above equation is discretized, and the intake pipe adhesion amount M
When f is eliminated, the following fuel adhesion model is obtained.

[0040]

(Equation 3)

Here, a = X and b = 1 / τ.

FIG. 9 shows an example of a method for deriving parameters of an intake pipe adhesion model at the time of engine start using the neuro engine model at the time of engine start of the present invention. In the figure, N
N4 is a neural network whose output is the adhesion parameters a and b. The fuel injection amount Gf is calculated from the output ann, bnn, the target cylinder fuel inflow amount Gfet, and the adhesion model H (z), and the calculated fuel injection amount Gf is set to the target air-fuel ratio calculated by the above-described start-time neuro engine model. Compared with the hourly fuel injection amount Gfnn, the coupling coefficient of the neural network NN4 is corrected so that this error becomes zero,
Do the learning.

FIG. 1 is a block diagram of a fuel injection control system using the neural network NN4 thus obtained.
0 is shown. The engine 101 detects an operating state parameter that is input data of the neural network NN4, and the arithmetic unit 102 of the neural network NN4 calculates estimated adhesion parameters ann and bnn, and calculates the estimated adhesion parameter and the target cylinder fuel inflow Gfet. , The fuel calculation unit 103 can calculate the fuel injection amount Gf at which the air-fuel ratio becomes the target air-fuel ratio.

The calculated adhesion parameter is:
Since simply connecting the obtained points results in a saw-tooth shape, the points are smoothed by applying a process such as passing through a filter such as a low-pass filter, and by having a smooth curve shape, cylinder variations and noise are reduced. Can be reduced.

As described above, according to the third embodiment, using the neuro engine model expressing the air-fuel ratio dynamic characteristics at the time of engine start, the intake pipe adhesion parameter at the time of engine start and the intake pipe adhesion model at the time of engine start are determined. Is used to calculate the fuel injection amount when the engine is started, so that the fuel entering the cylinder is the target cylinder fuel amount (combustion result A / F = target A).
/ F, the fuel injection amount) can be uniquely determined.

Embodiment 4 FIG. Next, an air-fuel ratio control device according to Embodiment 4 of the present invention will be described. When the engine is started at a low temperature, some of the fuel flowing into the cylinder is not burned, but remains in the cylinder as fuel. For this reason, in order to perform more accurate air-fuel ratio control, it is necessary to construct an adhesion model in the cylinder.

FIGS. 11 (a) and 11 (b) show conceptual diagrams of fuel adhesion in the cylinder. In FIG. 11A, 110
Is a fuel injection nozzle (injector) for injecting fuel into the intake pipe 111, 113 is a cylinder block that houses a piston 114 therein, and 116 is a cylinder block 1
A combustion chamber 117 is formed in a space inside a cylinder block 113 surrounded by a piston 114, an intake valve 112, and an exhaust valve 115. All of the fuel in the cylinder inflow fuel amount Gfe does not contribute to the combustion, and a part of the fuel adheres to the cylinder head or the like to form a fuel liquid film. Therefore, by considering the cylinder attachment model in the same manner as the intake pipe attachment model,
It can be considered as shown in FIG. 11B and can be expressed by the following equation.

[0048]

(Equation 4)

Here, c is a combustion contribution ratio representing the ratio of the amount of fuel actually contributing to the combustion in the fuel flowing into the cylinder, and d is the fuel evaporated from the liquid film adhering to the cylinder head or the like. Is the reciprocal of the evaporation time constant τ, and Gf
e ^* is the amount of fuel actually burned in the cylinder. Therefore,
The following Expression 5 is derived from Expressions 3 and 4.

[0050]

(Equation 5)

FIG. 12 shows a method for deriving the parameters of the cylinder adhesion model at engine start using the neuro engine model at start which constitutes the air-fuel ratio control device according to the fourth embodiment.

The NN5 calculation unit 124 is a neural network whose output is the cylinder adhesion parameters c and d. The outputs cnn and dnn, the target cylinder fuel combustion amount Gfet ^*, and the cylinder adhesion model F (z) 123
Then, the cylinder inflow fuel amount Gfe is calculated, and the outputs ann and bnn of the intake pipe adhesion parameter generator 121, the cylinder inflow fuel amount Gfe, and the intake pipe adhesion model H (z) are calculated.
122, the fuel injection amount Gf is calculated, and the fuel injection amount Gf is compared with the starting fuel injection amount Gfnn which is the target air-fuel ratio calculated from the starting neuroengine model, and this error becomes zero. As described above, the coupling coefficient of the NN5 is corrected and learned.

FIG. 13 shows a configuration diagram of the fuel injection control system using the NN5 obtained as described above. Engine 1
31, various operating state parameters are detected, and these are detected by the intake pipe adhesion parameter generator 132 and the NN5 arithmetic unit 1.
33, the intake pipe adhesion parameters ann and bnn and the cylinder adhesion parameters cnn and cnn, respectively.
The estimated adhesion parameter, the target cylinder fuel combustion amount Gfet ^*, and the intake pipe adhesion model H (z) 13 are calculated.
4, based on the cylinder attachment model F (z) 135,
The fuel injection amount Gf at which the air-fuel ratio becomes the target air-fuel ratio can be calculated.

By applying a filter to the calculated adhesion parameter in the same manner as in the third embodiment, errors due to cylinder variations and noise can be reduced. Further, the intake pipe adhesion parameter generator 132 may be constructed by a neural network.

As described above, according to the fourth embodiment, since the intake pipe attachment model is introduced and the cylinder attachment model is introduced, complicated engine behavior at the start can be represented more accurately. Thus, there is an effect that the air-fuel ratio control performance can be further improved.

Further, the cylinder attachment model may be configured by considering the following. That is, the parameters of the intake pipe attachment model and the cylinder attachment model at the time of engine start taking into account the gasoline evaporation temperature are derived using the start-time neuro engine model. Hereinafter, a specific method will be described.

The gasoline evaporation rate increases with the cylinder wall temperature. Since this property is known in advance, it is used as a priori knowledge. The wall surface temperature of the cylinder is determined by the number of TDCs since the start of the engine, that is, Tcou.
nt has characteristics as shown in FIG.
Medium, the number of explosions in the cylinder may be used instead of the TDC number). Further, as shown in FIG. 15, the gasoline evaporation rate P (P (t)) has such a characteristic that the higher the temperature is with respect to the cylinder wall surface temperature Tp, the closer the evaporation rate P becomes to 1 (100% all contribute to combustion). have.

Therefore, the cylinder attachment model taking into account the dynamic characteristics of the gasoline evaporation rate P has a configuration as shown in FIG. That is, as shown in FIG. 16B, similarly to the method of FIG. 12, the evaporation rate P, the estimated adhesion parameters ann, bnn, and the target cylinder fuel combustion amount Gfet, as shown in FIG.
^* , And the combustion contribution rate e may be derived using a neural network.

As described above, the intake pipe adhesion model and the cylinder adhesion model are determined by the gasoline evaporation temperature (a priori knowledge).
Is taken into account, the model becomes a model more suited to physical characteristics, and an effect of further improving control accuracy can be obtained. It is also effective in reducing the number of derived parameters of the cylinder attachment model.

[0060]

As described above, according to the air-fuel ratio control apparatus according to the first aspect of the present invention, the air-fuel ratio at the time of starting the engine is estimated using the number of explosions in the cylinder since the start of the engine. Therefore, the in-cylinder temperature change can be taken into account, so that there is an effect that the air-fuel ratio can be accurately estimated at the time of starting which has a very strong nonlinearity.

Further, using a neuro engine model,
From this, the starting fuel correction parameter is derived and the fuel injection amount at the time of starting is calculated, so that there is an effect that the air-fuel ratio control at which the air-fuel ratio at the time of starting becomes the target air-fuel ratio can be performed.

Further, since the intake pipe adhesion parameter is calculated from the neural network engine model, the fuel entering the cylinder is equal to the target cylinder fuel amount (combustion result A /
There is an effect that it is possible to uniquely determine a fuel injection amount that satisfies F = in-cylinder fuel amount that becomes a target A / F).

In addition, since a cylinder attachment model is introduced in addition to the intake pipe attachment model, complicated engine behavior at the time of starting can be more accurately expressed, and the air-fuel ratio control performance is further improved. There is an effect that can be.

Further, the gasoline evaporation temperature (a priori knowledge)
The cylinder adhesion model that takes into account the above is constructed, so that it becomes a model that is more suited to the physical characteristics, control accuracy can be improved, and the number of derived parameters of the cylinder adhesion model can be reduced. The effect is that it can be simplified.

[Brief description of the drawings]

FIG. 1 is a functional block diagram of an air-fuel ratio control device according to a first embodiment of the present invention.

FIG. 2 is a configuration diagram when an air-fuel ratio setting unit of the air-fuel ratio control device according to the first embodiment is constructed using a neural network.

FIG. 3 is a view showing a flowchart when calculating a correction parameter at the time of engine start in the air-fuel ratio control device according to the second embodiment.

FIG. 4 is a diagram for explaining a configuration of a neuro engine model of the air-fuel ratio control device according to the second embodiment.

FIG. 5 is a diagram for explaining a method of deriving a target fuel injection amount of the air-fuel ratio control device according to the second embodiment.

FIG. 6 is a diagram for explaining a learning method of a neural network that outputs a correction parameter of the air-fuel ratio control device according to the second embodiment.

FIG. 7 is a functional block diagram for explaining a control system configuration using a neural network that outputs correction parameters in the air-fuel ratio control device according to the second embodiment.

FIG. 8 is a view for explaining an intake pipe fuel adhesion model used in the air-fuel ratio control device according to the third embodiment.

FIG. 9 is a diagram for explaining a method of deriving an intake pipe adhesion parameter by the air-fuel ratio control device according to the third embodiment.

FIG. 10 is a diagram showing a configuration of a fuel injection control system using a neural network which is the air-fuel ratio control device according to the third embodiment.

FIG. 11 is a diagram for explaining a cylinder adhesion model used in the air-fuel ratio control device according to the fourth embodiment.

FIG. 12 is a diagram for explaining a method for deriving an engine start-time cylinder attachment parameter using a start-up neuro engine model by the air-fuel ratio control device according to the fourth embodiment.

FIG. 13 is a diagram showing a configuration of a fuel injection control system using a neural network which is the air-fuel ratio control device according to the fourth embodiment.

FIG. 14 is a characteristic diagram showing a relationship between the number of TDCs and the cylinder wall surface temperature from the start of the engine in the air-fuel ratio control device according to the fourth embodiment.

FIG. 15 is a characteristic diagram showing a relationship between a gasoline evaporation rate and a cylinder wall surface temperature in the air-fuel ratio control device according to the fourth embodiment.

FIG. 16 is a diagram showing a cylinder attachment model that takes into account dynamic characteristics of gasoline evaporation rate, which is an air-fuel ratio control device according to Embodiment 4.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 11 Engine 12 State detection part 13 Counter detection part 14 Air-fuel ratio estimation means 15 Compensator 72 Correction parameter calculation part 80, 110 Fuel injection nozzle 81, 111 Intake pipe 82, 112 Intake valve 83, 113 Cylinder block 84, 114 Piston 85, 115 Exhaust valve 86, 116 Exhaust pipe 87, 117 Combustion chamber 73 Starting fuel calculation unit 102 Neuro calculation unit 103 Fuel calculation unit 121 Intake pipe adhesion parameter generator 122 Intake pipe adhesion model calculation unit 123 Cylinder adhesion model calculation unit 173 Adhesion parameter table 182 Transient adhesion parameter estimation unit 183 Steady adhesion parameter map 184 Adhesion parameter correction unit 192 Correction coefficient calculation unit

 ────────────────────────────────────────────────── ─── Continuing on the front page (72) Inventor Toru Kitamura 1-4-1 Chuo, Wako-shi, Saitama Prefecture Inside Honda R & D Co., Ltd. (72) Inventor Naohiro Kurokawa 1-4-1 Chuo, Wako-shi, Saitama Inside Honda R & D Co., Ltd. (72) Inventor Akira Kato 1-4-1 Chuo, Wako-shi, Saitama

Claims (8)

[Claims] 1. An apparatus for controlling an air-fuel ratio at the time of fuel injection of an internal combustion engine, comprising: a state detection unit for detecting a parameter representing an operation state of the internal combustion engine; and counting the number of cylinder explosions since the start of the engine. And an air-fuel ratio estimating means for estimating an air-fuel ratio immediately after the start of the internal combustion engine from the operating state parameter and the number of explosions in a cylinder. 2. The neural network according to claim 1, wherein the air-fuel ratio estimating means uses the operating state parameter and the number of cylinder explosions as its input parameters, and its output as an air-fuel ratio at engine start. An air-fuel ratio immediately after the start of the internal combustion engine. 3. An apparatus for controlling an air-fuel ratio at the time of fuel injection of an internal combustion engine, wherein the fuel injection amount is such that the air-fuel ratio becomes a target air-fuel ratio using a neuro-engine model expressing an air-fuel ratio dynamic characteristic at the time of engine start. A correction parameter calculation unit for calculating Toutnn and calculating an engine start fuel correction parameter from the Toutnn; and an engine start fuel calculation unit for calculating a fuel injection amount at start using the engine start fuel correction parameter. An air-fuel ratio control device characterized by the following. 4. An apparatus for controlling an air-fuel ratio at the time of fuel injection of an internal combustion engine, wherein the fuel injection amount is such that the air-fuel ratio becomes a target air-fuel ratio using a neuro-engine model that expresses an air-fuel ratio dynamic characteristic at the time of engine start. Toutnn, an engine start adhesion parameter calculation unit for calculating an engine start adhesion parameter from the Toutnn and an intake pipe adhesion model at engine start, an engine start adhesion parameter, and an intake pipe adhesion model at engine start An air-fuel ratio control device, further comprising: an engine start fuel calculation unit that calculates a fuel injection amount at the time of engine start. 5. The air-fuel ratio control device according to claim 4, wherein the engine-start-time adhesion parameter calculation unit uses a neuro-engine model representing an air-fuel ratio dynamic characteristic at the time of engine start, and a cylinder adhesion model at the time of engine start. The engine start fuel calculation unit uses the intake pipe and cylinder adhesion parameters as the engine start adhesion parameters, and calculates the engine start fuel based on these parameters and the intake pipe adhesion model and cylinder adhesion model at engine start. An air-fuel ratio control device, which calculates a fuel injection amount of the air-fuel ratio. 6. The air-fuel ratio control device according to claim 5, wherein the engine start-time adhesion parameter calculation unit uses a neuro-engine model expressing an air-fuel ratio dynamic characteristic at the time of starting the engine, and takes into consideration the gasoline evaporation temperature. An air-fuel ratio control device for deriving parameters of an intake pipe model and a cylinder adhesion model at the time of starting the inserted engine. 7. The air-fuel ratio control device according to claim 1, wherein the counter detector detects the number of explosions in the cylinder based on a top dead center (TDC) count from cranking since the start of the engine. An air-fuel ratio control device for counting. 8. The air-fuel ratio control device according to claim 1, wherein the counter detector counts the number of explosions in the cylinder from the start of the engine based on an output of an internal pressure sensor that detects a pressure in the cylinder. An air-fuel ratio control device characterized in that: