JPH0565812B2 - - Google Patents

Description

[Detailed Description of the Invention] <Technical Field> The present invention relates to a learning control method in engine testing, and relates to an air-fuel ratio (A/F), which is the mixing ratio of air A and fuel F, an exhaust gas recirculation rate (EGR rate), etc. The control variable can be set to the set value in a short time.

<Technical background and problems> When performing engine performance tests, for example, A/
F value, EGR rate, etc. may be used as parameters,
In this case, it is necessary to set the A/F value, EGR rate, etc. to desired values by controlling the pump motor, EGR valve, etc. that control the amount of fuel supplied.

In this case, the EGR rate is generally calculated from the components of the exhaust gas, and the A/F value is also calculated from the components of the exhaust gas in the case of a method called the output method. When calculating from exhaust gas components in this way, a time lag of 4 to 5 seconds occurs in response.
Therefore, especially when such a large time lag occurs, it is difficult to perform normal continuous feedback control and set the control variables such as A/F value and EGR rate to desired values. It is.

In such a case, the simplest method is to change the operation amount of the control target such as the pump motor or EGR valve that controls the amount of fuel supplied in steps at regular intervals, and reduce the time lag each time. There is a method in which the set value is compared with the detected value of a control variable such as A/F, EGR rate, etc., and the manipulated variable is sequentially changed in steps until the detected value reaches the set value. However, this method has the disadvantage that the time required for setting is extremely long because it is necessary to reduce the amount of change per step in order to increase the setting accuracy.

Therefore, the next possible method is to change the manipulated variable in steps as before, but change the amount of change in one step according to the deviation amount (difference between the set value and the detected value). In this method, the relationship between the amount of deviation and the amount of change is determined in advance and created in a table. However, the relationship between the throttle valve opening and the A/F value, the throttle valve opening and the fuel flow rate,
Especially in engine performance tests, such as the relationship between the EGR valve opening and the EGR rate, the relationship between the manipulated variable and the controlled variable is not uniquely determined depending on the engine condition, and the relationship also differs depending on the engine model. There is a problem in that it is necessary to prepare a huge number of tables, and furthermore, it is necessary to collect data for a long time to create the table.

For this reason, conventionally, automatic control of control variables that have a time lag in response, such as A/F value and EGR rate, has not been performed, and manual operation using charts and the like has been the only option.

<Object of the Invention> In view of the drawbacks of the prior art described above, it is an object of the present invention to provide a learning control method that allows a control amount, which is a parameter in an engine test, to be set to a set value in a short time.

<Summary of the Invention> The present invention, which achieves the above object, improves the step control method in engine tests to change the manipulated variable in steps with respect to a predetermined set value, but the next change in the manipulated variable is , is predicted based on the relationship between the previous operation amount and the detection result of the corresponding control amount, and this is successively brought closer to the set value.

<Examples> Examples of the present invention will be described in detail below based on the drawings.

FIG. 1 is a block diagram showing an apparatus for implementing a first embodiment of the present invention. In this embodiment, the A/F value is controlled to a set value Q _S , and in this case, the control amount Q is the A/F value, and the operation amount S is the rotation speed of the pump motor 21 that controls the amount of fuel supplied. . In Figure 1,
1 is a learning control system, 2 is a fuel supply device, 3 is an engine, and 4 is a setting device. The fuel supply device 2 of this embodiment uses a float internal pressure type carburetor, and includes a pump motor 21, a cylinder 22, a buffer 23, and a carburetor 24. That is, by giving a command to the pump motor 21 and operating it, the cylinder 22 generates pressure according to the command, and this pressure is applied to the carburetor float chamber 25 of the carburetor 24 via the buffer 23. The carburetor float chamber 25 is communicated with a vaporization pipe 26, and fuel 27 in an amount corresponding to pressurization is mixed with air 28 and supplied to the engine 3. Therefore, the pump motor 21
By operating , the A/F value changes.

The learning operation control system 1 includes an A/F value detector 11 using an output method and a manipulated variable prediction calculation device 12. A/
Exhaust gas from the engine 3 is introduced to the F value detector 11, and the A/F value is detected by gas analysis. The operation amount prediction calculation device 12 changes a signal representing the operation amount (operation signal) in a stepwise manner and supplies it to the pump motor 21, but at this time, it changes the signal representing the operation amount in a stepwise manner and supplies it to the pump motor 21. predict the amount of operation.

Next, using the above device, the setting value of the control amount (A/F value in the case of this embodiment) is set by the setting device 4.
A learning control method according to the method of this embodiment when Q _S is set will be explained.

(1) Calculation of the first manipulated variable S ₁ (in the case of this example, the rotation speed of the pump motor 21): The detected value of the controlled variable Q, which is the initial setting and is the basis for predicting the manipulated variable S. Therefore, by giving the set value Q _S to the manipulated variable prediction calculation device 12, the manipulated variable S 1 is calculated based on an arbitrary relational expression Q=f(s) that indicates the relationship between the controlled variable Q and the manipulated variable _S. The controlled object (in this example, the pump motor 21)
control. At this time, the relational expression Q=f(s) is not particularly limited and may be approximated by an approximate expression expected to express the relationship between the two. For example, quadratic function approximation as shown in FIG. 2 may be used, linear function approximation, and approximation using other curves or polygonal lines may also be used.

(2) Calculation of the second operation amount _S2 : The actual control amount (in this embodiment, the A/F value) due to the first operation is detected and determined by the A/F value detector 11. Next, the difference ΔQ ₁ between the set value Q _S and the control amount Q ₁ detected by the first operation is determined using the following equation (1).

ΔQ ₁ = Q _S −Q ₁ ...(1) Next, the corrected manipulated variable ΔS ₁ corresponding to the difference ΔQ ₁ is determined using the relational expression used to calculate the first manipulated variable S ₁ , and the following formula ( 2) The second operation amount is
Calculate S ₂ and control the controlled object.

S ₂ = S ₁ + ΔS ₁ ...(2) At this time, in the case shown in Figure 2, the differential coefficient m of the relational expression Q = f(s) with respect to the manipulated variable S ₁ is used as ΔS ₁ = ΔQ/m. However, the first operation amount
There are no particular limitations on how to use the relational expression used in determining S ₁ . For example, in the case shown in FIG. 2, the differential coefficient at the point of the manipulated variable S=0 of the curve represented by the relational expression Q=f(s) may be used. In addition, an appropriate control constant k ₂ for the correction operation amount ΔS ₁
You can also do this by multiplying by

(3) Calculate the manipulated variable at the n-th step from the 3rd time onward: Since there are two or more detected values of the controlled variable Q, which is the basis for predicting the manipulated variable S,
The second manipulated variable S _o-2 and the corresponding actually detected controlled variable Q _o-2 , and the previous one, that is, the n-th
Using the first manipulated variable S _o-1 and the corresponding control variable Q _o-1 that was actually detected, the n-th manipulated variable is calculated using the following formula (3), and the manipulated variable Sn is calculated and controlled. Control the subject.

S _o = S _o-1 + k _o ×S _o-1 −S _o-2 /Q _o-1 −Q _o-2・ΔQ _o-1 …(3) However, k _o = control constant ΔQ _o-1 =Q _S −Q _o-1 (4) As the step change in the manipulated variable S described above progresses and the difference between the set value and the actually detected controlled variable Q falls within the allowable range, the manipulated variable S is kept constant at that value. Keep it.

In the above description, the fuel supply device 2 uses a float internal pressure type carburetor, and the pump motor 21 is controlled. When an electronic fuel supply system is used, the correction resistance value that is output to the electronic controller is generally manipulated. Note that the corrected resistance value is a function of the fuel increase ratio, and when conducting normal tests, the corrected resistance value is adjusted so that A/F = 14.7 (theoretical value) at a fuel increase ratio of around 1.5.

In addition, in the above explanation, from the third operation onwards, the operation amount was predicted by a linear function approximation based on the detected values at each of two past points, but it is predicted by linear function approximation based on the detected values at three or more points. It may be predicted by approximation. Furthermore, as shown in equation (3), the control constant may be changed for each prediction.

FIG. 3 is a block diagram showing an apparatus for implementing a second embodiment of the invention. In this embodiment, the EGR rate is controlled to a set value Q _S , and in this case, the control amount Q is the EGR rate, and the manipulated variable S is the operating pressure of the EGR pulp. In FIG. 3, 1 is a learning control system, 2 is a fuel supply system, 3 is an engine, 4 is a setting device, and 5 is an EGR device. Among these, the learning control system 1 has an EGR rate detector 13 instead of the A/F value detector 11. The other fuel supply device 2, engine 3, and setting device 4 are the same as those shown in FIG. 1, so redundant explanation will be omitted. The EGR device 5 of this example includes a differential amplifier 51, an amplifier 52, pulse oscillators 53a, 5
3b, inverter 54, solenoid valves 55a, 55b,
Vacuum pump 56, buffer 57, pressure sensor 58
and EGR pulp 59. The EGR valve 59 takes in exhaust gas from its inlet 59a and discharges it from its outlet 59b, thereby introducing the exhaust gas into the air-fuel mixture supplied from the fuel supply device 2. This controls the amount of exhaust gas introduced. this
The opening degree of the valve body 59c of the EGR valve 59 is determined by the pulse signal P ₁ output from the pulse oscillators 53a and 53b.
Change the duty of P ₂ and use solenoid valves 55a and 55b.
is controlled by duty control.
To be more specific, the pulse oscillator 53a receives an electrical signal representing the manipulated variable S directly, and the pulse oscillator 53b
are supplied with the electric signals via the inverter 54, so that the pulse signal P ₂ is the same as the pulse signal P ₁
It is the inverted version of . Therefore, the solenoid valve 55
The open and closed states of the solenoid valves 55a and 55b are inversely related to each other, so that they are never open at the same time, and moreover, the solenoid valve 55 is controlled by the pulse signal _P1.
When a opens, the vacuum acts on the EGR valve 59 via the buffer 57, and when the solenoid valve 55b controlled by the pulse signal _P2 opens, atmospheric pressure acts on the EGR valve 59 via the buffer 57.
In the former case, the opening degree of the valve body 59c increases, and in the latter case, the opening degree of the valve body 59c decreases. Thus, EGR
The operating pressure of the valve 59 is changed by changing the duty of the pulse signals P ₁ and P ₂ . Therefore, in this example, the controlled objects controlled by the signal representing the manipulated variable S, which is the output signal of the learning control system 1, are the pulse oscillators 53a and 53b.

Learning control when using such a device and setting the set value Q _S with the setting device 4 is similar to the first embodiment, and from the third time onwards, the next manipulated variable S is determined based on the detection results of the previous and two previous times. This is performed by predicting and changing the manipulated variable S in a stepwise manner, thereby changing the controlled variable Q in a stepwise manner. However, in this example, the operating pressure P of the EGR valve 59
Considering that a relationship that can be well approximated by a linear equation (operating pressure = EGR rate x k) holds between and the EGR rate,
The operating pressure acting on the EGR valve 59 is detected by the pressure sensor 58, and this detected pressure is compared with the operating amount S by the differential amplifier 51. That is, the operating amount S at this time is given by pressure, for example, the pressure of 0 to 200 mmHg corresponds to the voltage of 0 to 10 V, and the differential amplifier 51 determines that the operating pressure of the EGR valve 59 is the operating amount. Pulse oscillator 53 to match S
a, 53b.

Thus, in this case, the first and second learning control
Since the relational expression representing the relationship between the controlled amount (EGR rate) and the manipulated variable (the operating pressure of the EGR valve 59) is well approximated by a linear equation, the EGR rate can be quickly set to its set value Q. .

Of course, the manipulated variable S at this time is the pulse signal P ₁ , which is the output signal of the pulse oscillators 53a and 53b.
P ₂ duty may be selected, but since the relationship between the EGR rate, which is the control amount Q at this time, and the duty is not uniquely determined, the first and second operation amount S in learning control There is a possibility that the initial value setting and the next setting of the manipulated variable S will increase the difference between the set value Q _S and the actually detected controlled variables Q ₁ and Q ₂ , and in this case, It takes a longer time for the control amount Q to converge to the set value Q _S than when the manipulated variable S is used as the operating pressure as described above. However, in this case as well, an appropriate function Q=
By selecting f(s), the same effect as the first embodiment can be obtained.

<Effects of the Invention> As specifically explained above with the examples,
According to the present invention, when setting the control amount to the set value,
In addition to changing the manipulated variable of the controlled object in steps, when setting the manipulated variable from the third time onwards,
In addition to predicting and determining the manipulated variable from the relationship between the previous multiple manipulated variables and the control amount corresponding to each manipulated variable,
Since the first and second manipulated variables are determined from an arbitrarily determined relational expression between the controlled variables and the manipulated variables, the controlled variables can be quickly set to the set values.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing an apparatus for realizing the first embodiment of the present invention, FIG. 2 is a graph showing the relationship between the manipulated variable and the controlled variable according to the embodiment of the present invention, and FIG. FIG. 3 is a block diagram showing an apparatus for realizing a second embodiment. In the drawing, 1 is a learning control system, 2 is a fuel supply device,
3 is an engine, 4 is a setting device, and 5 is an EGR device.

Claims (1)

[Claims] 1 In a control method for engine testing in which the manipulated variable is changed stepwise to change the controlled variable in a stepwise manner, and this controlled variable approaches the set value, the first manipulated variable relative to the set value of the controlled variable is the controlled variable. The manipulated variable is determined by calculation from an arbitrarily determined relational expression with the manipulated variable, the controlled object is controlled with this manipulated variable, and then the second manipulated variable is calculated from the controlled variable and set value for the first control. The corrected operation amount corresponding to the difference between Divide the difference between the two manipulated variables by the difference between the two immediately preceding controlled variables,
The calculated value is predicted and determined based on the value multiplied by the difference between the set value and the immediately preceding controlled variable, and the manipulated variable is set in a stepwise manner based on the above prediction until the detected controlled variable approaches the set value within the allowable range. A learning control method in an engine test characterized by changing.