JPH01267334A - Fuel injection quantity controller for internal combustion engine - Google Patents

Abstract

PURPOSE:To make improvements in secular accuracy of control and its stability by estimating a parameter expressing a fuel behavioral model by means of an output equation alone and then limiting this parameter to a value within the specified range, while renewing it after smoothing. CONSTITUTION:Engine speed of an internal combustion engine M1, fuel quantity being evaporated out of an intake pipe wall surface M2, fuel quantity flowing into a cylinder M4, and fuel quantity being sprayed out of a fuel injection valve 5 are detected each by a means M6. In addition, fuel behavior is predicted by a means M7 on the basis of each detected result. Furthermore, a parameter to express a fuel behavioral model is estimated on the basis of the predicted result. On the other hand, whether the estimated parameter is in the specified range or not is judged by a means M9, while the limited parameter is smoothened by a means M10. Then, on the basis of this smoothened parameter, the parameter of the fuel behavioral model is renewed by a means M11, and simultaneously a fuel injection quantity is controlled by a means M12 on the basis of this renewed parameter.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention controls the amount of fuel injected from a fuel injection valve in accordance with a fuel behavior model that describes the behavior of fuel flowing into the cylinder of an internal combustion engine. The present invention relates to a fuel injection amount control device for an internal combustion engine.

[Prior Art] Conventionally, fuel injection control devices for internal combustion engines based on linear control theory have been disclosed (JP-A-59-196930, JP-A-60-162027, etc.). In such technology, for example, a correction value for the basic fuel injection amount of an internal combustion engine is used as the control human power, an actual value of the air-fuel ratio detected using an air-fuel ratio sensor is used as the control output, and a linear approximation is established between the human output values. A model describing the dynamic behavior of the internal combustion engine is determined, and the fuel injection amount is controlled based on this model.

[Problems to be Solved by the Invention] However, in the conventional technology, the relationship between the amount of control human power and the amount of control output is inherently nonlinear, and the dynamic behavior of the actual internal combustion engine is not necessarily constant. As a result, the control of the fuel injection amount may become unstable or the control accuracy may deteriorate as described below.

(I) Conventional problems due to the nonlinear relationship between human power and output: In conventional technology, a model is created for each of multiple operating regions that can be considered to hold linear approximation, and based on this model, each operating region is A control law was set for each engine, and the law was switched depending on the operating state of the internal combustion engine.

Therefore, control accuracy becomes insufficient, and control becomes unstable at the boundary points of each operating region due to switching of the control law.

(II) Conventional problems caused by the dynamic behavior of internal combustion engines not being constant: The dynamic behavior of internal combustion engines can be solved by, for example, using a device that has a variable mechanism in the intake system and affects the flow inside the intake bow. If the engine is equipped with a fuel injection device that atomizes the injected fuel and is strongly affected by the flow, or if the internal combustion engine has changes in fuel behavior due to deposits on the intake boat, intake valve, etc. Change.

For this reason, automatic adjustment (self-tuning) of model parameters is also practiced (Japanese Unexamined Patent Application Publication No. 60/1998) mentioned above.
-162027, etc.).

This automatic adjustment involves modeling the controlled object as shown in equation (1) below, and calculating a1 from the measurement results of Y(k) and u(k).
~an, bl~bn are automatically adjusted.

Y(k)=alφV(k,-1)+a2◆V(k-2)
+-・-+an◆Y(k-n)+bl-u(k-1)+
b2◆u(k-2)+=+bn◆u(k-n)-(1)
u(k): Control human power Y(k): Control output al~an, bl~bn: Parameters However, in this case, since there are many estimated parameters, the delay in response to parameter changes tends to be large, and the injection amount Control i
There was a problem that the responsiveness of wholesalers decreased.

In order to solve these problems, a technique has been proposed in which a fuel behavior model is constructed from various state equations and output equations, and parameters are estimated using the output equations to control fuel injection. However, even when these models are used, noise from an air-fuel ratio sensor etc. may affect the performance.

The present invention reduces the influence of noise from air-fuel ratio sensors, etc.
The purpose of this invention is to set the detected value of the air-fuel ratio and various values calculated using the detected value as appropriate values, and to improve both the stability and accuracy of control of an internal combustion engine based on these values.

[Means for Solving the Problem] The structure of the present invention made to achieve the above object is as follows:
As illustrated in the figure, the behavior of the fuel flowing into the cylinder M4 of the internal combustion engine M1 is expressed by using the amount of fuel attached to the intake pipe wall M2 of the internal combustion engine M1 and the amount of evaporated fuel fv in the intake pipe M3 as state variables. A fuel injection amount control device for an internal combustion engine M1 that controls a fuel injection amount q from a fuel injection valve M5 in accordance with a fuel behavior model consisting of the described state equation and output equation, the rotational speed ω of the internal combustion engine M1 , the wall surface evaporation amount νf of fuel evaporated from the intake pipe wall surface M2, the inflow fuel amount mc◆λ corresponding to the amount of fuel flowing into the cylinder M4, and the operating state including the fuel injection amount q from the fuel injection valve M5. The operating state detecting means M6 detects the wall evaporation amount shif, the inflow fuel amount mc·λ, and the fuel injection amount q! of the rotational speed detected by the operating state detecting means M6. ,: a fuel behavior prediction means M7 for predicting the behavior of the fuel including the adhering fuel amount f− and the evaporated fuel amount fν according to the fuel behavior model.
Then, using the adhering fuel amount and the evaporated fuel amount fv predicted by the fuel behavior prediction means M7, the inflow fuel amount mc is calculated.
Parameter estimating means M8 that estimates parameters representing the fuel behavior model based only on the linear output equation that determines the relationship between λ and fuel injection amount q, and the parameters estimated by the parameter estimating means M8 are A parameter that determines whether or not it is within a range, and if it is within a predetermined range, the estimated parameter is adopted, and if the value is outside the predetermined range, a predetermined value is adopted as the parameter. limiting means M9; parameter smoothing means M10 for smoothing the parameters limited by the parameter limiting means M9; and parameter updating for updating the parameters of the fuel behavior model based on the parameters smoothed by the smoothing means MIO. means Mll; and injection amount control means M12 that controls the fuel injection amount q using the parameters updated by the parameter updating means Mll and according to the fuel injection amount calculation formula set based on the fuel behavior model. The gist is a fuel injection amount control device for internal combustion engine M1.

Here, the above-mentioned operating state detecting means M6 detects the rotational speed ω of the internal combustion engine M1, the wall surface evaporation amount Vf of the fuel evaporated from the intake pipe wall surface M2, and the amount of fuel flowing into the cylinder M4 by, for example, the method shown below. Corresponding inflow fuel amount mc・λ
, and the fuel injection amount q from the fuel injection valve M5.

That is, the rotational speed ω can be detected by today's rotational speed sensor.

Furthermore, the wall evaporation amount Vf, which is the amount of fuel evaporation from the intake pipe wall surface M2, is determined as a function of the saturated vapor pressure Ps of the fuel in the intake pipe M3 and the pressure inside the intake pipe M3 (intake pipe pressure) P. be able to. It is difficult to directly detect this saturated vapor pressure Ps with a sensor. However, the saturated vapor pressure Ps is a function of the temperature T of the fuel adhering to the intake pipe wall surface M2, and this adhering fuel temperature T can be represented by the overjacket water temperature of the internal combustion engine M1 or the cylinder head temperature near the intake boat. . Therefore, the saturated steam pressure Ps is determined by detecting the water jacket water temperature or the cylinder temperature using a temperature sensor, and using the detected temperature T as a parameter, for example, using the following equation (2). be able to.

Ps=β1◆T2-β2-T+β3 ・(2
) (However, β1.β2.β3: Constant) In other words, the wall evaporation amount Vf is determined by calculating the saturated vapor pressure Ps based on the detection signal from the temperature sensor that detects the water jacket water temperature or cylinder head temperature, and then calculating the saturated vapor pressure Ps based on the well-known suction The intake pipe pressure P may be detected using an atmospheric pressure sensor, and may be detected using a data meter or an arithmetic expression using these six values Ps and P as parameters. Also, this wall evaporation amount V
Since f changes greatly depending on the saturated vapor pressure Ps, it may be approximately determined using the following equation (3) using the saturated vapor pressure Ps as a parameter.

Further, the inflow fuel amount ITlc corresponding to the fuel amount that has flowed into the cylinder M4 is, for example, the air amount ITIc and the fuel-air ratio λ that satisfies the ratio of fuel to air of the fuel mixture that has flowed into the cylinder M4. It can be found as the product of The fuel-air ratio λ can be detected, for example, by an air-fuel ratio sensor, and the air amount mc can be detected by the following method.

(2) The air amount me is calculated using the following equation (4) using the intake pipe pressure P1, the intake air temperature T1, and the rotational speed ω of the internal combustion engine M1 as parameters.

mc=(βX(ta)Pi−βV(tw))/T i
...(4) However, βX(ω) and βY(ω) are ω
Function ■ Intake pipe pressure Pi and rotational speed ω of internal combustion engine interbrain 1
The basic air amount m is determined using a map using the parameter , and the calculated result is corrected using the intake air temperature Ti to determine the air amount mc.

■ Air amount m during the intake stroke from the detected value of the air flow meter
Estimate and find c.

Further, the above fuel injection IJ'J amount q is, for example, the fuel injection valve M
5 can be detected directly or from the control state of the injection amount control means M12, such as the pulse width of the valve opening signal.

Next, the fuel behavior prediction means M7 predicts the behavior of the fuel based on the fuel behavior model. Below, construction of this fuel behavior model and prediction of fuel behavior will be explained.

■ Building a fuel behavior model: Using the amount of attached fuel fw and the amount of evaporated fuel fv as state variables,
The behavior of the fuel flowing into the cylinder M4 of the internal combustion engine M1 is determined by the method described later, and is expressed as the following (5) in a discrete system using the intake cycle k of the internal combustion engine M1 as a sampling period.
It is described using the state equation shown in Equation (2) and the output equation shown in Equation (6) below. The following ■( indicates the intake cycle.

+[α1 α3]T-q (k) + [-a 2 a 2]T-Vf(k)/ω(k
)...(5)mc(k)λ(k)=[a9 a7ko・[fw(k) fv(k)]d+(1-α1-α
3)・q (k) ... (6) fw (k), fw
(k+1)-=adhered fuel amount fv(k), fv(k+1
)...Evaporative fuel amount αl, α2. α3. α7. α9...
Parameter q (k)...Fuel injection amount Vf(k)...Wall evaporation amount ω(k)...Rotational speed mc(k)・λ of internal combustion engine M1
(k)...Inflow fuel amount Next, the procedure for deriving the above state equation (5) and output equation (6) will be explained.

The amount of fuel fc flowing into the cylinder M4 of the internal combustion engine M1 can be described as shown in equation (7) below.

fc(k)=a4・q(k)+a9・fw(k
) + a 7◆fv(k)...(7) α4...Parameter (α4=1-α1-α3)α4・
q (k)...direct inflow amount α9 from fuel injection valve M4
・fw(k)...Indirect inflow amount α from the suction pipe wall surface M2
7・fv(k)...Inflow amount of evaporated fuel Also, the above (7
) The inflow fuel amount fc(k) can be expressed by the formula:
As mentioned above, using the air amount mc(k) and the fuel-air ratio λ(k), it can be described as shown in the following equation (8).

fC(k): Inc(k) entering (k) - (8) Furthermore, the amount of fuel fw(k) adhering to the intake pipe wall surface M2 is determined by the amount of fuel adhering to the intake pipe wall M2 due to the inflow of the fuel mixture into the cylinder M4 in each intake cycle. , decreases at a rate corresponding to parameter α9, increases at a rate corresponding to parameter α1 due to adhesion of a part of the fuel injected from fuel injection valve M5, and further increases α2 due to evaporation of fuel into intake pipe M3. Vf(k)
/ω(k).

Therefore, the adhering fuel amount fw(k) can be described as shown in equation (9) below.

fw(k+1)=(1-a 9)fw(k)+a
1φq (k)-α2・Vf(k)/ω(k)...
・(9) On the other hand, the amount of evaporated fuel fv(k
) decreases at a rate corresponding to parameter α7 due to the inflow into cylinder M4 in each intake cycle, increases at a rate corresponding to parameter α3 due to evaporation of a part of the fuel injection amount q(k), and further It increases by α2-Vf(k)/ω(k) due to the evaporation of fuel from the intake pipe wall surface M2.

'Therefore, the amount of evaporated fuel fv(k) can be written as shown in equation (10) below.

fv(k+1)=(1-a?)fv(k)+a
3・q(k)+α2・Vf(k)/ω(k)・
...(10) Therefore, with the intake cycle as the sampling period, the above equation (7) or 1 expressed in a discrete system
0) equation as adhering fuel amount fw (k) and evaporated fuel amount fv (k)
is the state variable, and the inflow fuel amount fc(k) (mc(k) ・
When λ(k)) is regarded as the output and summarized, the above-mentioned equations (5) and (6) are obtained, and the values of each parameter are determined using the system identification method.

■ Prediction of fuel behavior: Fuel injection amount q(k) determined by measurement or calculation.

Wall evaporation amount fν(k) 9Rotational speed (k), inflow fuel amount mc(k)・input(k), state variables fw(k), fν(k) using equations (5) and (6) Predict.

The fuel behavior prediction means M7 may be, for example, a minimum dimension observer (Minimal Order Observer).
server), same-dimensional observer (Identit
y 0bserver ), finitely settled observer (De
ad Beat 0bserver), Linear Function Observer (Linear Function 0bserver)
er) or adaptive observer (Adaptive 0
bserver), using the well-known design method detailed in ``Basic System Theory'' by Katsuhisa Furuta et al. (1978) Corona Publishing, or ``Mechanical System Control'' by Katsuhisa Furuta et al. (1988) Ohm Publishing, etc. Can be configured.

Further, the parameter estimating means M8 is a fuel behavior model in which the adhering fuel amount fw(k) and the evaporated fuel amount fν(k) are state variables, and the state equation and the output equation are special as described below. This method takes advantage of the fact that it has a unique structure.

That is, the control system of this embodiment is based on the above-mentioned equations (5) and (6).
It is designed based on the physical model shown in Eq. Since this physical model is nonlinear, the physical model is first approximated linearly.

In the above equations (5) and (6), y (k) = input (k) mc (k) - (1-al-a3
) ・q(k)...(11) x(k)=[fw(k) f v(k)ko T
...(12) ■= [α9 α7
...(1-5), then equation (5), (6)
The formula is x(k+1)=@・x(k)+f−q(k)+n・Vf
(k)/ω(k) −(16)y(k)=■・x(k)
...(17) can be expressed.

As a result, the parameter (2) can be estimated and updated using only the linear equation (17). That is, the operation described later is performed on the estimated parameter (2) to create a new parameter (2), and the new parameter (2) is updated as the parameter (2) of equation (16), which does not require linearity.

An example of estimating the above parameter 0 will be explained using the above equations (5) and (6). Here, the parameter α1, which is greatly influenced by fuel behavior, is considered. Assume that α9 changes, and parameters α2. α3. It is assumed that α7 does not change because the influence of fuel behavior is extremely small. The parameter identification method used here is described in "Identification Method for Linear Discrete-Time Systems - ■" by Takayoshi Nakamizo, System and Control Vol. 1.2.
5 Nag p476-489 1981.

The output equation (6) can be written as shown in the following equation (18).

mc(k)*λ(k) −a?・fv(k)=a9
Order fw(k) + (1-αt-α3)q(k) −(1
B) The left side of equation (18) and each parameter etc. are expressed as follows (1
It is expressed as Equation 9) to Equation 23).

Y(k)=mc(k)λ(k)−α7◆fv(k)−
(19)XI(k) = fw(k)−(20)X2(
k) = q(k)...(21)al=α9...(
22) α2ko1−α1−α3 (23) By the above formula (19) or 23), the above (1B)
The formula can be written as shown in formula (24) below.

V(k)=al order X1(k)+a2・X2(k)−(
24) Therefore, the fuel amount mc(k)・λ(k) and the fuel amount q
(k) and using those measured values (5)
The fuel amount fw(k), fv(
By calculating k) and substituting it into equation (19) or 21), Y(k), Xl(k), X2 in equation (24)
(k) can be known.

Next, 't'(k), Xl(k) obtained as above
, X2(k), the coefficient 09. A method for determining α1 will be explained.

First, Y(k) and al・Xl(k)+a2・X2(k)
Describe the error e(k) as shown in equation (25) below,
This evaluation function Je is expressed as the following equation (26).

e(k)=Y(k)-al ◆X1(k)-a2 dispute X2
(k)・(25)Je: Σρ′・(e(k))
2 i=O :Σo k-' (V(k)-al-Xl(k)-a2
・X2(k))2...(26)i=0 ρ is set between 0 and ρ≦1, and the For is used to reduce the weight as it becomes a past value to reduce its contribution to the estimated value.
It is a getting Factor.

Therefore, al(k) that minimizes the above equation (26),
By determining a2(k), parameters α1. α9 can be estimated from the following equations (27) and (28).

a 9=al(k)...-(27) a 1=1-a2(k)-a 3-= (2B) Also,
The parameter limiting means M9 means the parameter α1. It is determined whether α9 has an inappropriate slope, such as a negative value, due to the influence of noise, etc., and then sets an appropriate value. As a specific example, α1 or α9 of the above parameters is 0.
If it is below, α1. Set α9 to 0, while α
1 or α9 is 1 or more, respectively α1. α9 is set to 1.

The parameter smoothing means MIO also includes parameters α1. α9 is smoothed to a more appropriate value. For example, each parameter α1. Smooth α2.

■ Smooth the capacity by arithmetic averaging the values of the past N intake cycles. That is, a 1(k) = 1/
N(a 1(k-1)+α1(k-2)・-a 1(k
-n))...(29) a 9(k) = 1/N・(a 9(k-1)+α9(
k-2)-a 9(k-n))...(30) ■ Perform first-order lag filter processing to smooth the capacitance. That is, α l(k+1) = A conflict α 1(k) + (1-A)
◆α1(k-1)...(31)α9(k+1) =A
・α9(k)+(1-A)・α9(k-1)...(3
2) A is a factor that sets the weight, and is set in the range O (8 < 1. This allows smoothing to be performed with emphasis on either the previously estimated parameter or the currently estimated parameter, and the next parameter is Can be set.

Further, the parameter updating means Mll refers to the above (29).
) and (30), or (31) and (32), the parameters α1(k+1) and α9 are obtained by smoothing.
(k+1) is used to update the parameters of the above equation (5), which is the state equation.

Further, the injection control means M12 calculates the fuel injection amount q (k) based on the fuel behavior model using the parameters updated by the parameter update means M12,
It is something to control.

Next, a method of determining the fuel injection amount q(k) of the fuel behavior model expressed by the above equations (5) and (6) will be explained.

First, the evaluation function J of equations (5) and (6) is expressed as (33) below.
It shall be a formula.

J = e◆((mc(k)λr-mc(k)λ(k+1
))2=e・[mc(k) entered r-a 9(1-a 9)
fw (k) − α7 meeting (1 − α7) fv (k) − (α9◆α l+α7◆α3) conflict q(k) − (1 − α
1-α3)◆q(k+1) −(a 7- a 9) a 2◆Vf(k)/ (,
J(k)]? -(33)e...Target fuel-air ratio here q(k)''; By setting q(k-1), (3
Equation 3) becomes Equation (34) below.

J = e([mc(k)λr-a 9(1-a 9)f
w(k)-a?・(1-a?)*fv(k)-(α
9・α1+α7◆α3+(1-α1-α3))・q(k
)-(a7-a9) a2・Vf(k)/ω
(k)]2) - (34) Next, use the following equation (35) to determine the fuel injection amount q that minimizes the evaluation function J of the above equation (34).
(k) is obtained from the following (36).

θJ/θq(k)=-2[mc(k) input r-a 9(1
-a 9)fw(k) -α7◆(1-α7)◆fν(k) -(α9◆α1+α7◆α3+1-01-α3)・q(
k)-(a 7-a 9)a 2・Vf(k)/ω(k
) co・(α9◆α1+α7◆α3+1-01-α3)=
0...(35) (1(k)=(mc(k) entered r-a 9・(1-a 9
)fw(k)-α7・(1-α7)◆fv(k) −(a 7-a9) α2・Vf(k)/ω(k)
)/(α9◆α1+α7◆α3+1−α1−α3)・・
・(36) Therefore, by this equation (36), the fuel injection amount q
(k) is found.

[Function] The fuel injection amount control 'In device for the internal combustion engine M1 of the present invention detects the rotational speed ω of the internal combustion engine M1 and the wall surface evaporation amount Vf of fuel evaporated from the intake pipe wall surface M2 by the operating state detection means M6.
, detects the operating state including the inflow fuel amount me・in corresponding to the amount of fuel that has flowed into the cylinder M4, and the fuel injection amount q from the fuel injection valve M5, and based on a fuel behavior model using the detected values. Then, the behavior of the fuel including the amount of adhered fuel and the amount of evaporated fuel fv is predicted by the fuel behavior prediction means M7.

Next, the parameter estimating means M8 predicts the parameters of the fuel behavior model only from the linear output equation. Then, the parameter limiting means M9 limits inappropriate parameters among the predicted parameters, and the parameter smoothing means MIO smoothes the limited parameters.

Next, the parameter updating means Mll updates the parameters representing the fuel behavior model using the smoothed parameters.

Then, the injection amount control means M12 controls each fuel amount m according to the fuel behavior model using the updated parameters.
The fuel injection amount q from the fuel injection valve M5 is controlled based on c.on, q, fW, fv, etc.

That is, the parameter limiting means M9 and the parameter smoothing means MIO reduce the influence of noise, etc. on the parameters, so that the fuel injection control executed based on the fuel behavior model using the parameters is appropriately performed. .

[Example] An embodiment of the present invention will be described below based on the drawings.

FIG. 2 shows a schematic configuration of a system to which the present invention is applied, centering on an internal combustion engine (engine).

The engine 10 is controlled by an engine controller 12, and an intake air temperature sensor 16 is provided near the air cleaner 14 to detect an intake air temperature Ti and output an intake air temperature signal.

A throttle valve 20 is disposed downstream of this intake temperature sensor 16, and this throttle valve 20 has an R ON state (LL R ON) when the throttle valve is fully closed.
An idle switch 22 and a throttle sensor 24 that detects the opening degree of the throttle valve 20 are attached.

A surge tank 2 is provided downstream of the throttle valve 20.
6 is formed, and an intake pressure sensor 27 that detects the intake pipe pressure Pi and outputs an intake pressure signal is provided.

An intake manifold 28 and an intake boat 30 are provided downstream of the surge tank 26 in which the intake pressure sensor 27 is provided.

The suction boat 30 includes an engine controller 1.
A fuel injection valve 32 that opens in response to a valve opening signal from 2 is attached.

Furthermore, an exhaust manifold 3 is located downstream of the combustion chamber 34 in which the fuel injected from the fuel injection valve 32 is combusted.
6 is provided.

Furthermore, an oxygen sensor 38 is attached to the exhaust manifold 36 for detecting sinking into the combustion chamber from the residual oxygen concentration of the exhaust gas and outputting a combustion chamber ratio signal.

The engine block 40 forming the combustion chamber 34 includes:
An engine coolant temperature sensor 42 is attached to detect the coolant temperature in the water jacket and output a coolant temperature signal.

A high voltage from an igniter 46 whose ignition time is controlled according to the output from the engine controller 12 is supplied to the spark plug 44 attached to the combustion chamber 34 via a distributor 48 .

Furthermore, an engine rotation speed sensor 50 that detects the engine rotation speed ω and outputs an engine rotation speed signal, and a cylinder discrimination sensor 52 that outputs a cylinder discrimination signal are attached to the distributor 4.

The engine controller 12 includes a human output interface 64, a storage section 66, and a central processing section,
Perform the processing shown below.

(1) Processing of inputting signals etc. from sensors of various parts of the engine 10 via the input/output interface 64.

(2) Based on the above-mentioned various manually inputted signals, the storage unit 66
The fuel injection control routine shown in FIG. 3, which is stored in
A process in which various drive signals are calculated by the central processing unit 68 according to programs and data of various control routines (not shown).

(3) Based on the calculation results of the central processing unit 68, drive signals etc. for each part of the engine 10 are sent to the human output interface 64.
Processing to output from.

Next, the fuel injection control routine of this embodiment, which is executed when the engine 10 is operating, will be explained with reference to the flowchart shown in FIG.

When the routine shown in FIG. 3 is activated, initialization processing is first performed (steps 100 to 130).

That is, the following processes are executed in order.

Set the initial value fw(0) to the adhering fuel amount fw, and set the evaporated fuel amount fv
An initial value fv(0) is set to (step 100).

Then, the value a1. which will be described later. Variable Y used to calculate a2
Set the initial value v(0) to the variable XHz, set the initial value XI(0) to
Set the initial value X2 (0) to the variable X2 (step 10
5).

Next, the parameter α9 is set to an initial value α9(0), which is determined based on the state in which the amount of adhering fuel at the time of starting is "zero," and the parameter α1 is set to the initial value α1, which is determined based on the water temperature, etc. at the time of starting. (0) (step 110).

Next, set the parameter α9 to the value a1 (directly set “1-α” to a2
1-α3” (step 120).

Next, the determinant P is set (step 130). Note that p is a sufficiently large positive number.

After the above initialization is completed, as shown below,
The rotational speed of the engine 10, the air amount mc, the fuel chamber ratio λ are measured, and the fuel wall evaporation amount Vf is calculated (Step 1
40).

The rotational speed is determined by manually inputting the detection peak of the engine rotational speed sensor 50 via the human output interface 64. The air amount mc is calculated using the rotational speed ω and the intake pipe pressure Pi detected by the intake pressure sensor 27.
The basic air amount m is obtained from the map shown in the figure, and is corrected and calculated using the intake air temperature T1 detected by the intake air temperature sensor 16. The combustion chamber settling is determined by manually inputting the detected value of the oxygen sensor 38 via the human output interface 64.

The wall evaporation amount Vf of fuel is calculated based on equations (2) and (3) based on the detected value (cooling water temperature) T of the engine water temperature sensor 42 manually inputted via the human output interface 64.

After detecting the six values indicating the actual operating conditions, the target fuel chamber ratio input r is calculated using the preset map shown in FIG. 5 (step 150). Based on the rotational speed ω of the engine 10 and the intake pipe pressure P
According to i, a value that optimizes the operating condition of the engine is determined.

After detecting and calculating the above six values, proceed to fuel injection! 1-I
Control is performed to calculate the amount q of fuel and actually inject the amount q of fuel (step) 60.

That is, first, based on equation (36), the detected, calculated, or set air amount mc, target fuel chamber ratio λr, parameter α9. α1. Adhering fuel amount fw, evaporated fuel amount fv, wall evaporation amount Vf, rotational speed ω, and preset parameter α7. α3. Fuel injection collapse q is calculated according to α2. Next, when the crank position of the engine 1o reaches a predetermined position, the fuel injection valve 32 is opened for a time corresponding to the fuel injection amount q. That is, fuel injection is performed.

After executing the fuel injection described above, the amount of adhered fuel fw and the amount of evaporated fuel fv are predicted (step 170).

In this prediction, the amount of adhering fuel f- and the amount of evaporated fuel fv are predicted by the observer from the state equation (5) and the output equation (6) that represent the fuel behavior model of this embodiment.

By calculating the predicted fuel amounts fw and fv, the values Y and Xi are determined by the above-mentioned equations (19) to 24).

x2 is calculated (step 180).

Subsequently, values a1 and a2 that minimize the evaluation function Je of the error e(k) shown in equation (25) described above are calculated from equation (26) (steps 190 to 210). That is, here (
26) Minimize the equation (ial, a2) using the following algorithm.

■ Calculate the matrix [KI K2]' using the following equation (37) (step 190).

[K1(k) K2(k) koT=P(k-1)・[X
l(k) K2(k)]'/(ρ+[Xl(k) K
2(k)]・P(k-1)・[Xl(k) K2(k
)koT) ...(37) Forgetting
Factor D: 0 (D≦1■) A matrix P(k) is calculated using the following (3 days) formula (step 200).

P(k) = 1/ρ ・(E-[1(k) K2(k
)]1-Xl(k) K2(k)ko)◆P(k-1
)...(3 days) ■ Calculate the matrix [al a2]T using the following equation (39) (step 210).

[al a21 blocks = [al(k-1) a2(k-
1)]”+[kl(k) 82(k)] T−(Y(
k)-[Xl(k) K2(k)]・[al(k-1)
a2(k-1)koT)...(39) By the above matrix [al a2]T calculation algorithm, the value a1. By finding a2, the already mentioned (2
7) and (2B), the parameter α1. α9 is calculated, and the parameters of equations (5), (6), and (36) described above are estimated (step 220).

Next, the estimated parameter α1. In order to prevent α9 from becoming an inappropriate value such as a minus value due to the influence of noise, a limiter is provided and set to an appropriate value. That is, the parameter α1 has a preset upper limit value α1max (
For example, when 1) is exceeded (step 230), the parameter αl is set to 1, which is the upper limit value α1max (step 240). Further, when the parameter α1 is less than the preset lower limit value α1m1n (for example, 0) (step 250), the parameter α1 is set to 0, which is the lower limit value α1max (step 260). On the other hand, when the parameter α9 exceeds the preset upper limit α9max (for example, 1) (step 270), the parameter α9 is changed to the upper limit α9max (for example, 1).
9max is set to 1 (step 280). Further, when the parameter α9 is less than the preset lower limit value α9m1n (for example, 0) (step 290), the parameter α9 is set to 0, which is the lower limit value α9max (step 30).
0).

Next, the parameter α1 set in this way. In order to more appropriately remove the influence of noise etc. from the value of α9, the parameter αl,
α9 is smoothed, and the parameter α1. Calculate a new value of α9.

The value of the new parameter obtained by this calculation is adopted, and the parameter α1 of the above-mentioned equations (5), (6), and (36) is calculated. α9 is updated (step 310).

In this way, by limiting and smoothing the estimated parameters, it is possible to obtain appropriate parameters that are less affected by noise, and by using these parameters to update the parameters of the fuel behavior model, for example, the air-fuel ratio A hill is created which reduces the influence of noise on the fuel chamber ratio λ detected by sensor 25. That is, step 230 above
By using the parameters corrected by 11 in step 310, the fuel injection amount q calculated by returning to step 160 becomes appropriate, and the value of aL a2 is calculated in step 210 using the fuel injection amount q. Ru. Furthermore, based on the values of al and a2, step 220
Since the parameters are calculated again, the parameters are more appropriate than the parameters calculated last time.

This parameter is then limited and smoothed again and used to calculate the next fuel injection amount q. Therefore, by repeating such processing, the fuel injection amount sJ&Ij can be suitably performed, so the stability and accuracy of the control of the fuel injection amount q of the engine 1 can be improved, and the engine 1 can always be controlled. Able to follow changes appropriately.

Next, parameter α1. The simulation results of the above example showing the effect of limiting the upper and lower limits of α9 and smoothing are shown in the sixth example.
The conventional parameters α1. Contrast this with self-tuning, which does not change α9.

Here, as the calculation conditions, as shown in Figure 6, for sampling numbers = 0 to 500, α1 = 0.4° α
9=0.1. fw(0)=0. fV(0): 0, target fuel amount mce included: 50mg l a2 * νf(k)
/ω(k)=10mg, α3=0°α7=0.09. For sampling numbers = 500 to 1000, α1 = 0.3. The controlled object was changed so that α9=0.02.

The simulation results under the above calculation conditions are as shown in FIG. 7, with parameters α1. The estimated value of α9 closely follows the calculation conditions αl and α9. That is, the parameter α1 of the fuel behavior model. α9 always follows changes in the engine and accurately monitors the engine's fuel behavior.

Actual fuel amount mc (k
)・λ(k) change rate (mcoλr −mc(k) ◆
λ(k))/mc・λ" is, as shown in Figure 8, in the conventional self-tuning with parameters α1 and α9-constant,
Although it increases until approximately 150 times have elapsed since the occurrence of the disturbance = 500, according to this embodiment, it quickly converges to "0" after the disturbance. That is, according to this embodiment, the fuel injection 'ff1q(k) is appropriately controlled, and the fuel amount mc
(k)·λ(k) is controlled to closely follow the target fuel amount me◆λ.

As explained above, this embodiment has the following extremely excellent effects.

(2) Among the parameters representing the fuel behavior model, the parameter α1, which is greatly affected by changes in the engine 10 over time such as deposits, can be automatically adjusted, so that the accuracy of control over time can be improved.

(2) Since the parameter α9, which is greatly affected by variations between models, such as variations in the intake system, can be automatically adjusted, the adjustment process during manufacturing can be significantly shortened, and control accuracy can also be improved.

(2) Since the equation of state (5) does not need to be linear, there is no need to perform linear approximation, and moreover, highly accurate control can be performed based on a model that is more suitable for the actual engine 10. Furthermore, control stability is also improved.

(2) Compared to conventional self-tuning, only two parameters (α1, α9) are required to estimate, so parameters can be estimated and updated quickly. This improves control followability and accuracy.

■ When a parameter becomes an inappropriate value outside the specified range due to the influence of noise contained in the measured value, it is set to a value within the specified range, so the influence of noise etc. can be reduced and the control Accuracy can be improved.

■ Since the parameters are smoothed, the effects of noise etc. on the parameters can be reduced in the same way as above.
Control accuracy can be improved.

Note that the present invention is not limited to the above embodiments, and can be implemented in various ways.

[Effects of the Invention] The fuel injection amount control device for an internal combustion engine of the present invention estimates parameters representing a fuel behavior model using only an output equation, limits the values to values within a predetermined range, and then smooths and updates them. , the following extremely excellent effects are achieved.

■ Since the parameters representing the fuel behavior model can be automatically adjusted according to changes in the engine over time, the accuracy of control over time can be improved.

■ Since variations between models can be automatically adjusted,
The adjustment process during manufacturing can be shortened to a width of 2, and control accuracy can also be improved.

■ Since the equation of state does not need to be linear,
A more accurate control model can be used, and control stability is improved.

■ Compared to conventional self-tuning, there are fewer parameters to estimate, so parameters can be updated with high followability. This improves control followability and accuracy.

(2) When a parameter becomes an inappropriate value, it is set to an appropriate value within a predetermined range, so the influence of noise etc. on the parameter can be reduced and control accuracy can be improved.

■ Since the parameters are smoothed, the effects of noise etc. on the parameters can be reduced in the same way as above.
Control accuracy can be improved.

[Brief explanation of the drawing]

Fig. 1 is a block diagram illustrating the basic structure of the present invention, Fig. 2 is a block diagram of an embodiment, Fig. 3 is a flowchart of fuel injection control in the embodiment, and Fig. 4 is a fuel quantity mc map. An explanatory diagram, FIG. 5 is an explanatory diagram of the target combustion chamber subsidence r map, FIG. 6 is a graph showing the change characteristics of the nuclear fuel amount in the example, and FIG.
The figure is a graph showing how the estimated values of parameters change according to the embodiment, and FIG. 8 is a graph showing a comparison of the control characteristics of this embodiment and the conventional example. Ml...Internal combustion engine M2...Intake pipe wall surface M3...Intake pipe Ml...Cylinder M5...Fuel injection valve M6...Operating state detection means Ml...Fuel behavior prediction means M8... Parameter estimating means M9...Parameter limiting means MIO...Parameter smoothing means Mll...Parameter updating means Ml2...Injection amount control means 10...Engine 12...Engine controller 16...Suction Air temperature sensor 27...Intake pressure sensor 30...Intake boat 32...Fuel injection valve 34...Combustion chamber 38...Oxygen sensor 50...Engine speed sensor Representative Patent attorney Tsutomu Sadadate (et al.) 2 people) Figure 4 Main engine rotation speed ω Figure 5 Engine rotation speed ω 67 ￥7g Sampling number (k no-- Figure 8 ^ Sampling number (k) →

Claims (1)

[Scope of claims] Consisting of a state equation and an output equation that describe the behavior of fuel flowing into the cylinders of an internal combustion engine, with the amount of fuel adhering to the wall surface of the intake pipe between the internal combustion engines and the amount of evaporated fuel in the intake pipe as state variables. A fuel injection amount control device for an internal combustion engine that controls the amount of fuel injected from a fuel injection valve in accordance with a fuel behavior model, the device comprising: a rotational speed of the internal combustion engine; an amount of fuel evaporated from a wall surface of an intake pipe; an operating state detecting means for detecting an operating state including an amount of inflowing fuel corresponding to the amount of fuel flowing into the fuel injection valve and an amount of fuel injected from the fuel injection valve; and the rotational speed and wall surface evaporation amount detected by the operating state detecting means. , a fuel behavior prediction means for predicting fuel behavior including an adhering fuel amount and an evaporative fuel amount according to the fuel behavior model based on the inflow fuel amount and the fuel injection amount; and an adhering fuel amount predicted by the fuel behavior predicting means. Based only on the linear output equation that defines the relationship between the inflow fuel amount and the fuel injection amount, using
A parameter estimating means for estimating a parameter representing the fuel behavior model, and determining whether the parameter estimated by the parameter estimating means is within a predetermined range, and if it is within a predetermined range, the estimated parameter is parameter limiting means for adopting a predetermined value as a parameter in the case of a value outside a predetermined range; parameter smoothing means for smoothing the parameter limited by the parameter limiting means; parameter updating means for updating the parameters of the fuel behavior model based on the parameters smoothed by the smoothing means; and a fuel injection amount set based on the fuel behavior model using the parameters updated by the parameter updating means. A fuel injection amount control device for an internal combustion engine, comprising: injection amount control means for controlling the fuel injection amount according to a calculation formula;