JPH0385602A - Fuzzy control device - Google Patents

Abstract

PURPOSE:To output a non-linear manipulated variable by summing the operated result of a linear control arithmetic part and correction quantity by a fuzzy inference arithmetic means, and outputting the sum as the manipulated variable to a controlled system. CONSTITUTION:A fuzzy controller 3 inputs deviation between a set value and an observed value and its differential value, and outputs the correction quantity to be added to the manipulated variable of a PID controller 1. Differentiation is executed by the daplacian operator 4. Accordingly, since basic configuration is designed by PID control, a control state is easy to grasp, and steady-state deviation near a neutral point is not caused as well. Further, if the control state changes greatly, the rule of the fuzzy controller 3 acts gradually. Thus, since the non-linear manipulated variable can be outputted by correcting the manipulated variable of linear control by using fuzzy inference, favorable control can be realized.

Description

[Detailed Description of the Invention] Industrial Application Field The present invention relates to control technology of a fuzzy controller.
The present invention relates to a fuzzy control device that realizes more fine-grained control in addition to the PIDID control linear control method.

The conventional technology fuzzy control will be explained. Fuzzy control uses human power information obtained from the control observed value input section 101 as shown in FIG. 8, such as control deviation e and its rate of change de,
When describing the relationship between the manipulated variable U output from the control manipulated variable output unit 103 as a 1f-then... rule,
A plurality of inference rules as shown below are prepared in the fuzzy inference rule storage unit 104.

R1:IF e is ZO and de is PM
THEN u is NM.

R2:IF e is zOand de is PS
THEN u is PS.

R3:IF e is NS and de
is NS THEN u is zO
.

R4: IF e is NS and cu
e is PS THEN u is P
M.

Rn:IF e is NB ancl d
e is NB THEN u is ZO
.

However, Rj (j-1, 2, 3, 4...n) is a fuzzy inference rule.

Here, the part if~ is called the antecedent part, and the part then... is called the consequent part, and S4 zO (Zero), PM (Posi
tive Medium) and NM (Negative Medium)
Medium) is a label representing a fuzzy number of human power or output used to describe a rule, and is a fuzzy variable. An example is shown in FIG.

Commonly used fuzzy variables are NB (negatively large), NM (negatively medium), and NS (negatively small).

20 (daitaise o), PS (exactly small rhinoceros), P
There are M (just medium), PB (just large), etc. 10th
The figure shows a rule table expressing inference rules.

Next, the functions of the fuzzy inference calculation section 102 will be described.

Now, a method for determining the degree of conformity ω of the antecedent part of the rule Rj to the human power information e and de will be explained using the first rule R1 as an example. Here, μzo(e), μpm(de
) represents the membership value of the fuzzy number Z○ of the antecedent proposition and the input information e and de for PM. Assuming that eo and deo are input from the control observation value input unit 101 in FIG. 8, (1) 1 = μzo(eo)A μpm(deo)
(However, LA is a min operation) And the fuzzy number μm of the conclusion of the consequent part of rule R1 is
The membership value μnm(u
) can be found as follows.

μl-ωlAμnm(u) This situation is shown in FIG. The real values eO and deo are observed, and the consequent fuzzy variable NM is cut (min) from the goodness of fit of μzo(eo) and μpm(deo).
It can be seen that μl is derived.

Since there are multiple inference rules Rj, the fuzzy number that combines the fuzzy numbers of all the conclusions is uT = μlVμ2vμ3v...Vpp (However, ■ is a max operation) As an example, the first rule and the second rule are shown in Figure 12. ma of
x performance uT-μIVμ2 is shown.

This fuzzy number uT is a conclusive fuzzy number indicating the amount of control operation. Since the actual amount of control operation uo is a real number, if we adopt the weighted center of gravity shown below, uo = (S
u・μT(u) du) / (Sμ,T(u)
du) and is output to the control operation amount output section t, Oa in FIG.

On the other hand, FIG. 13 is a general linear control configuration diagram. 1
05 is an object to be controlled; 106 is a filter for removing noise components of observed values; 107 is a controller that controls the object to be controlled based on the deviation between the target value and the observed value; 108 is a disturbance calculation unit that estimates and calculates disturbances. controller 1.0.7
i & For example, Kc (1+1/sTi+5Td) Kc gain Ti, Td: time constant S, like Laplace operator 4: UPID (proportional+integral-sufficient differential) controller. If integral action is included, the steady-state error rate will improve, but the reaction response and stability will deteriorate. In addition, by adding a differential element, the phase lag can be improved.
Often improves stability. In general, design methods for linear control methods such as PID control have been established, and the effects of L1 disturbances that are easily foreseeable can be eliminated if they can be determined by calculation. For example, if the inverse characteristic of the controlled object 105 is obtained, and the disturbance calculation unit 08 calculates the amount of operation necessary to cancel the disturbance, and inputs it to the controlled object 105, then theoretically there will be no influence of the disturbance on the control system. ,
As the filter 106, a digital filter such as a moving average method is often used.

Problems to be Solved by the Invention In such conventional PID control devices, when the controlled object is a highly nonlinear system, it is difficult to output an appropriate manipulated variable, and good control cannot be achieved (
1 Also, when controlling by predicting the disturbance and adding it to the manipulated variable, the force is good in a steady state, but there is a delay in response to the transient change in the true disturbance.As a result, the PID control does not work well. It disappears.

On the other hand, in conventional fuzzy control devices, there is a problem of how to determine fuzzy inference rules that are capable of outputting nonlinear manipulated variables. That is, it is difficult to predict where on the control operation amount U axis to set many consequent part fuzzy variables, and a certain degree of design guidelines are required.

The present invention has been made in view of these points, and the present invention has been made in view of the above points.
It is an object of the present invention to provide a fuzzy control device capable of correcting a control manipulated variable using fuzzy inference and outputting a nonlinear manipulated variable.

Another object of the present invention is to provide a fuzzy control device that monitors and corrects abnormal changes in the manipulated variable caused by delays in disturbance calculation, etc.

Means for Solving the Problems The invention according to claim 1 comprises linear control calculation means for constructing a control system using a linear model of a controlled object, and fuzzy inference calculation means for constructing an inference system using a logical control law.
The linear control calculation means (using the deviation between the target value and the observed value from the control object as human power, outputs a calculation result that can control the control object even by itself); It is characterized in that it is configured to output a correction amount of a linear control calculation result, and to add the calculation result of the linear control calculation means and the correction amount and output it as an operation amount to the controlled object.

It is preferable to use at least the second-order differential value of the observed value from the controlled object, or the second-order differential value of the deviation between the observed value from the controlled object and the target value, for the human power of the fuzzy inference calculation means. The described invention includes a control calculation means for controlling the controlled object by inputting a deviation between a target value and an observed value from the controlled object, and for calculating a disturbance applied to the controlled object and outputting a value that offsets the influence of the disturbance. The disturbance calculation means and the fuzzy inference calculation means that configures an inference system based on logical control laws (#e) add the above control calculation result and the disturbance calculation result to obtain the operation amount for the controlled object. The inference calculation means is characterized in that it is configured to manually calculate the manipulated variable and the observed value from the controlled object, and output a correction amount and add it to the manipulated variable only when it is inferred that there is an abnormality in the control state. The invention according to claim 4 includes a control calculation means for controlling the water level of the dam by using the water level as an observed value and the opening amount of the discharge gate as an output, and calculating the amount of inflow flowing into the dam and outputting a value that offsets the influence of the amount of inflow. and a fuzzy inference arithmetic means that configures an inference system based on a logical control law; the fuzzy inference arithmetic means adds the control arithmetic result and the disturbance arithmetic result to obtain the gate opening amount; The present invention is characterized in that the gate opening amount and the water level are input, and when it is inferred that there is an abnormality in the control state, the next correction amount is outputted and added to the gate opening amount.

Operation The invention according to claim 1 provides the fuzzy inference calculation means in parallel with the linear control calculation means,
Since the amount of correction by the fuzzy inference calculation means can be added to the control operation amount of the linear control calculation means, good control can be realized.

The invention according to claims 3 and 4 (1) Since changes in the manipulated variable with a delay due to disturbance etc. and changes in observed values from the controlled object can be monitored and corrected using fuzzy inference, fine-grained control is possible. It is possible to control the target.

Embodiment FIG. 1 is a block diagram of a fuzzy control device in a first embodiment of the present invention. In the figure, 1 is a PID controller that uses human power as the target value deviation and outputs the manipulated variable to the controlled object 2, 3 is a fuzzy controller, and 4 is a Laplace operator representing first-order differentiation.

It will be explained that a nonlinear manipulated variable can be efficiently output using the control device of the first embodiment configured as described above.

First, the controlled object 2 is linearized to configure a 1 PID controller 1 using a conventional method. Next, a fuzzy controller 3 is created using a fuzzy inference rule table. The fuzzy controller 3 manually calculates the deviation between the target value and the observed value and its differential value, and outputs a correction amount to be added to the operation amount of the PID controller 1. Differentiation is performed using Laplace operator 4.

FIG. 2 shows an example of the rule table. The inputs are the target value deviation and its differential value. The target value deviation is the formula % formula % (). Since the rule value near the neutral point is zero (Z○), the PID controller l performs control independently. Furthermore, if the control state changes significantly, the value of the rule is added to the output of the PID controller 1 as a correction amount to control the controlled object 2.

As described above, according to the first embodiment, the basic configuration is PI
Since it is designed with D control, it is easy to understand the control situation, and there is no steady-state deviation near the neutral point.Furthermore, when the control state changes significantly, the rules of Fuzzyco 2-controller 3 come into play gradually, so the amount of operation is reduced. A good control system can be constructed without any sudden changes in the control state having an adverse effect on the control state. It is also much more efficient and predictable than designing fuzzy control alone.

FIG. 3 is a block diagram of a fuzzy control device in a second embodiment of the present invention. In the figure, 5 is a PID controller that receives the target value deviation as input and outputs the manipulated variable to the controlled object 6, 7 is a fuzzy controller, and 8 is a Laplace operator representing second-order differentiation.

It will be explained that by using the fuzzy control device of the second embodiment configured as described above, a control system that can quickly respond to changes in the target value can be obtained.

First, the controlled object 6 is linearized to configure the conventional PID controller 5. Next, fuzzy controller 7
is created using a fuzzy inference rule table.

The fuzzy controller 7 detects the deviation between the target value and the observed value,
Input the second derivative value and L PID controller 5
In addition to the manipulated variable, the 3-power correction amount is output. Differentiation is performed using a second-order differential Laplace operator 8.

FIG. 4 shows an example of the rule table. The deviation between the target value and the observed value (target value deviation) and its second-order differential value are used as input. Target value deviation is positive, that is, (target value deviation) - (
When target value)-(observed value)>0 and the second-order differential value of the target value deviation is negative, a positive correction amount is added to the PID output.

Also, when the target value deviation is negative, that is, (target value deviation) - (target value) - (observed value) <0, and the second derivative value of the target value deviation is positive, a negative correction amount is sent to the PID output. will be added. For other cases, the correction amount is set to zero.

Since the second-order differential value represents the amount of acceleration, by using this as the correction amount, it is possible to configure a control system that is more resistant to transient response.

Expressed as a formula, it is as follows.

G(t) = PID(t) 10 F(t
)G(t): Operation amount PID(t) to the controlled object at time t
: PID controller output at time t 4 F(t) Fuzzy controller output at time t (correction function) When the step response is measured as shown in Figure 5, the result is better when the correction amount is added than when using only PID control. Start-up is faster and settling time is also shortened.

As described above, according to the second embodiment, the basic configuration is PI
Since it is designed with D control, it is easy to understand the control situation, and there is no steady-state deviation near the neutral point.Furthermore, the second-order differential value of the observed value is used to determine the correction amount for fuzzy inference, making it possible to quickly respond. It is possible to construct a control system with Further, when the control state changes significantly, the rules of the C-fuzzy controller 7 are gradually applied, so that a good control system can be constructed without any sudden changes in the manipulated variable having an adverse effect on the control state.

FIG. 6 is a block diagram of a fuzzy control device in a third embodiment of the present invention. In the figure, 9 is a PID controller that controls the controlled object, 10 is the controlled object, 11 is a fuzzy controller, and 12 is an inflow amount calculation unit that calculates the amount of water flowing into the dam 10 from the upstream dam or river, that is, the disturbance. It is.

A procedure for monitoring and correcting the control state using the control device of the third embodiment configured as described above will be explained.

Consider the case in which the fuzzy controller 11 is not present in FIG. The PID controller 9 performs control to bring the dam water level to the target water level. Dam 1° has an inflow (disturbance) from upstream, and this fluctuation causes disturbances in the water level. Therefore, by using the inflow amount calculation unit 12, the inflow amount is calculated from the water level change and the amount released from the gate.△ If the result is applied to the gate opening degree, it should become more robust against disturbances. However, in reality, there is a delay in calculating the inflow, so deviations occur when there is an accelerated change in the true inflow.

Therefore, the fuzzy controller 11 is arranged as shown in FIG. The inputs of the fuzzy controller 11 are the deviation between the target water level and the current water level and the gate opening degree. The output is a correction amount added to the gate opening degree.The operation of the fuzzy controller 11 will be explained in more detail using FIGS. 7(a) to 7(d). FIG. 7(a) is a rule table that is applied when the current dam water level is higher than the target dam water level. In the ZO (zero) part in the rule table, the correction amount is zero and is controlled only by the PID controller 9. The control condition is abnormal when, as shown in Figure 7 (b), the gate opening is decreasing even though the water level in the dam is increasing, and the amount of water discharged should normally be increasing. The gate opening should be on an increasing trend.

This abnormal state is shown in a rule table as shown in Figure 7(a).
This is the part other than Z○. By adding these to the PID controller 9, the final gate opening degree is determined. The formula is as follows.

G(t) = PID(t) + F(t)C(t):
Gate opening degree PID(t) at time t: PID controller output F(t) at time t
): fuzzy controller output cuff at time t Since the gate opening is opened by the correction amount F(t), an abnormal state can be avoided.

FIG. 7(c) is a rule table that is applied when the current dam water level is lower than the target dam water level.

The basic operation is the same. In the rule table, select ZO (
In the part (0), the correction amount is zero and only the PID controller 9 is controlled. The control state is abnormal as shown in Figure 7 (d
), even though the water level in the dam is decreasing, the gate opening is increasing, and the gate opening, which reduces the discharge amount, should be decreasing. This abnormal state is shown in a rule table as shown in Figure 7 (
This is the part other than ZO in C). By adding these to the PID controller 9, the final gate opening degree is determined.

As described above, according to the third embodiment, the relationship between the manipulated variable and the observed value from the controlled object can be monitored in detail using fuzzy inference. Similarly, since the correction amount can be output smoothly, it is possible to quickly recover from an abnormal state.

8- As described in detail of the invention, according to the fuzzy control device of the present invention, the manipulated variable of linear control can be corrected using fuzzy inference and a non-linear manipulated variable can be output, so At the same time, since it is not necessary to design fuzzy inference independently, it is possible to significantly save time and effort.

Similarly, good control can be achieved by monitoring and correcting abnormal changes in the manipulated variable caused by delays in disturbance calculations, etc.

[Brief explanation of drawings]

FIG. 1 shows fuzzy response inhibition in the first embodiment of the present invention.
FIG. 6 shows the configuration of the fuzzy control device in the third embodiment of the present invention. The explanation shown in Fig. 8 is the structure of the fuzzy control device in the conventional example.
Figure 2 is for explaining the fuzzy inference calculation @ Figure 13 is a block diagram of a linear control system in a conventional example. 1, 5, 9...PID controller, 2, 6, 10
... Controlled object, 3, 7, 11 ... Fuzzy controller, 12 ... Inflow amount estimation calculation accent

Claims (4)

[Claims] (1) A linear control calculation means that configures a control system using a linear model of a controlled object, and a fuzzy inference calculation unit that configures an inference system based on a logical control law, and the linear control calculation unit is configured to calculate the target value and the control The fuzzy inference calculation means receives the deviation from the observed value from the object as input and outputs a calculation result that can control the control object even if it is used alone, and the fuzzy inference calculation means takes the deviation as input and outputs a correction amount of the linear control calculation result. ,
A fuzzy control device characterized in that the calculation result of the linear control calculation means and the correction amount are added together and outputted as a manipulated variable to the controlled object. (2) A claim characterized in that at least a second-order differential value of an observed value from a controlled object, or a second-order differential value of a deviation between an observed value from a controlled object and a target value is used as an input to the fuzzy inference calculation means. The fuzzy control device according to item 1. (3) A control calculation means that controls the controlled object by inputting the deviation between the target value and the observed value from the controlled object, and a disturbance that calculates a disturbance applied to the controlled object and outputs a value that offsets the influence of this disturbance. and a fuzzy inference arithmetic means that configures an inference system based on a logical control law, the control arithmetic result and the disturbance arithmetic result are added together as a manipulated variable for the controlled object, and the fuzzy inference arithmetic means A fuzzy control device configured to receive the manipulated variable and the observed value from the controlled object as input, and output a correction amount and add it to the manipulated variable only when it is inferred that there is an abnormality in the control state. . (4) A control calculation means that controls the water level of the dam by using the water level as an observed value and the opening amount of the discharge gate as an output, and a disturbance that calculates the amount of inflow flowing into the dam and outputs a value that offsets the influence of this amount of inflow. and a fuzzy inference arithmetic means that configures an inference system based on a logical control law, the control arithmetic result and the disturbance arithmetic result are added together to obtain the gate opening amount, and the fuzzy inference arithmetic means A fuzzy control device configured to receive an opening amount and the water level as input, and output a correction amount and add it to the gate opening amount only when it is inferred that there is an abnormality in the control state.