JPH06195107A - Fuzzy temperature control system - Google Patents

Abstract

PURPOSE:To provide the fuzzy temperature control system which eliminates a large fluctuation of a control output and secures stability of control, and also, can shorten a response time. CONSTITUTION:The system for controlling a control object of a dead time plus first order lag or second order lag by applying fuzzy inference is constituted by providing a look-up table 4 for expanding a rule constituted of a deviation to a target temperature, and a difference (variation of deviation) of the present deviation to the previous deviation to a data table of a deviation-variation plane as numerical data, an integral processing part 5 for allowing the deviation to the target temperature to be subjected to integral processing, and outputting a value obtained by multiplying it by a prescribed coefficient, and an output processing part 7 for adding the value obtained by the integral processing part to a numerical value derived from the look-up table, based on the deviation to the target temperature and a variation of the deviation and obtaining a control output of the control object.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a temperature control system to which fuzzy reasoning is applied, and more particularly to a fuzzy temperature control system suitable for temperature control of a molding machine.

[0002]

2. Description of the Related Art In a conventional fuzzy PI temperature control system in which a fuzzy inference plus a first-order lag system (or a second-order lag system) is controlled by applying fuzzy inference, a fuzzy rule is used to provide versatility. Generally, the rules are shown in FIG.

That is, the abscissa represents the deviation (actual temperature measurement value-temperature target value), and the ordinate represents the deviation change (current deviation-previous deviation). Such a rule is a general rule in the past, but if a rule with a large dead time or a rule with a small time constant is controlled according to this rule,
The control becomes such that overshoot occurs or the limit is reached before the target value is reached.

Therefore, the present inventor considers the deviation-change plane constructed by such a rule as an XY plane as shown in FIG. 8, and the XY plane is constructed by the rule shown in FIG. It has been previously proposed that a control output value at each (X, Y) coordinate is calculated by fuzzy reasoning, and that the calculated numerical data is used as a table for control.

However, since the numerical table created in this way cannot be actually used as it is, the width of the deviation (X) and the width of the change (Y) are determined as follows. That is, as a method of determining each width, for example, the characteristic of the controlled object, the dead time (L), the time constant (T), and the process gain (Gp) are calculated by a general marginal sensitivity method, and the values are used to calculate the deviation. Width (E), width of change (ΔE)
Are determined by the following equations.

[0006]

[Equation 1] E = Gp × [1-exp (−mL / T)] ΔE = Gp × [1-exp (−τ / T)] where m is a coefficient and τ is a calculation time.

Since the dead time L is the time from when the control instruction is given until the response of the controlled object appears, the value of m basically needs to be larger than 1. on the other hand,
The width of change indicates how much the deviation changes in a certain time, and τ in the formula is the calculation time.

By determining the width of deviation (E) and the width of change (ΔE) in this way, temperature control can be performed using this table.

Actually, since this is a speed type control, if the control output value extracted from this numerical table is (ΔM _v ), the control output value M _{vn at} this time is

[0010]

## EQU00002 ## It can be obtained by the formula of M _vn = M _vn-1 + ΔM _vn (n = 0, 1, 2 ... t).

That is, the control output value M _vn = M _vn-1 + ΔM _vn
Is output, and time proportional on / off control is performed based on a preset output cycle.

[0012]

However, when the control is performed by such a method, the control output is a speed type output, so that the output fluctuation is very large, and as a result, the stability of the control is deteriorated. There was a problem left.

The present invention has been made in view of the above circumstances, and an object of the present invention is to provide a fuzzy temperature control system which eliminates large fluctuations in control output to ensure control stability and also shortens response time. To provide.

[0014]

In order to solve the above-mentioned problems, a fuzzy temperature control system of the present invention is a system for controlling a dead time plus a first-order lag system or a second-order lag system by applying fuzzy reasoning. , A look-up table in which a rule composed of the deviation with respect to the target temperature and the difference between this deviation with respect to the previous deviation (change of deviation) is expanded as numerical data in the deviation-change plane data table, and deviation with respect to the target temperature Is integrated and outputs a value obtained by multiplying this by a constant coefficient, and a value obtained from the look-up table based on the deviation with respect to the target temperature and the change in the deviation. An output processing unit that adds the obtained values to obtain the control output of the control target is provided.

[0015]

In the present invention, a look-up table in which a rule composed of the deviation with respect to the target temperature and the difference between the previous deviation and the present deviation (change in deviation) is developed as numerical data in the deviation-change plane data table To use. Then, a position type output method is adopted in which the output based on the numerical data (control output value) obtained from the look-up table is directly output to the control target. By adopting this position type output system, it becomes possible to reduce the fluctuation of the control output.

On the other hand, by adopting such a position type output method, since a steady deviation remains, an integration processing section is added to eliminate the steady deviation.

That is, the integration processing unit performs integration processing on the deviation with respect to the target temperature and outputs a value obtained by multiplying the deviation by a constant coefficient to the output processing unit.

In the output processing unit, the value obtained from the integration processing unit is added to the numerical value obtained from the look-up table based on the deviation with respect to the target temperature and the change in the deviation, and this value is used as a control output to be controlled. It is designed to output.

By changing the timing of the integration processing in the integration processing section between the temperature raising side and the temperature lowering side, it is possible to control the temperature raising and lowering with different characteristics with the same parameter.

[0020]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the fuzzy temperature control system of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram showing the electrical construction of the fuzzy temperature control system of the present invention, which is a position type output system.

In the figure, the output of the target temperature setting section 1 for setting the target temperature is led to the plus input of the adder 2, and the minus input of the adder 2 controls, for example, an extruder. The detected temperature (actual temperature measurement value) detected by the target 8 is given. Further, the output of the adder 2 is led to a temperature change processing unit 3 that calculates a temperature change, a lookup table 4, and an integration processing unit 5, and the integration processing unit 5 stores various control parameters. Coefficient setting unit 6
The output of is derived.

The output of the temperature change processing unit 3 and the output of the coefficient setting unit 6 are led to the look-up table 4, and the output of the look-up table 4 is an output processing unit 7 which determines a control output. Have been led to.

The output of the integration processing unit 5 is guided to the output processing unit 7, and the output of the output processing unit 7 is guided to the controlled object 8 as a control output.

The adder 2 carries out an arithmetic process of adding the actual temperature value detected from the controlled object 8 to the temperature target value set by the target temperature setting unit 1 by subtracting it from the measured temperature value. The obtained deviation (actual temperature measurement value-temperature target value) is output to the temperature change processing unit 3, the lookup table 4, and the integration processing unit 5, respectively.

The temperature change processing unit 3 obtains the difference between the previous deviation and the deviation of this time (change of deviation), and provides the look-up table with data indicating the change of deviation.

The look-up table 4 has a rule composed of data showing a deviation with respect to the target temperature and data showing a change in the deviation given by the temperature change processing unit 3 as a numerical data in a deviation-change plane data table. The expanded numerical data is held. However,
The numerical data here is the numerical data indicating the position type output.

The integration processing unit 5 integrates the deviation with respect to the target temperature and multiplies the deviation by a constant coefficient given by the coefficient setting unit 6 to output a value to the output processing unit 7.

In the output processing section 7, the value obtained from the integration processing section 5 is added to the numerical value obtained from the lookup table 4 to obtain the control output of the controlled object 8.

Next, the operation of the fuzzy temperature control system having the above structure will be described.

The fuzzy rule in the fuzzy temperature control system of the present invention is similar to the rule shown in FIG.

That is, the horizontal axis represents the deviation, and the vertical axis represents the change in the deviation. Then, the deviation-change plane constituted by such a rule is regarded as an XY plane as shown in FIG. 8, and this XY plane is regarded as a plane constituted by the rule shown in FIG. , Y) coordinate control output values are calculated in advance by fuzzy inference, and the calculated numerical data are used as a table for control. For example, if the X-axis and the Y-axis are each 0.1 step, 4
Numerical data of 41 (21 × 21) will be calculated.

However, since the numerical table created in this way cannot be actually used as it is, the width of the deviation (X) and the width of the change (Y) can be expressed by the formula shown in [Equation 1] above. Determined by

In this embodiment, m shown in [Equation 1] is empirically set to 3. That is, the value that changes the temperature during the time three times the dead time L is used as the deviation width.

Since the dead time L is the time from when the control instruction is given until the response of the controlled object appears, the value of m basically needs to be larger than 1. on the other hand,
The width of change indicates how much the deviation changes in a certain time, and τ in the formula is the calculation time.

For example, in the case of an extrusion molding die, Gp = 6
00 (degrees), L = 30 (sec), T = 5000 (se
c), and m = 3, the width (E) of the deviation is

[0037]

## ^EQU00003 ## If E = ^600.times. (1- ^e.sup. - ^{90 / 5000} ) = 10.70 (degrees) and the calculation time .tau. = 4 (sec), the width of change (.DELTA.E) is

[0038]

## EQU4 ## ΔE = 600 × (1−e ^−4/5000 ) = 0.479 (degrees / sec).

By thus determining the width of deviation (E) and the width of change (ΔE), temperature control can be performed using this table as the lookup table 4.

In the present invention, since the control is of the position type, the control output value LM _v extracted from this numerical value table is directly output to the controlled object 8 via the output processing unit 7.

On the other hand, the integration processing unit 5 integrates the deviation with respect to the target temperature and outputs a value obtained by multiplying the deviation by a constant coefficient to the output processing unit 7.

However, if the integral calculation is always performed during the temperature rise, there arises a disadvantage that the integral calculation value in the vicinity of the set value varies depending on the temperature rise start temperature. Therefore, in order to keep the integral calculation in the vicinity of the set value always the same, the integral calculation is started when the deviation is within the range (E).

FIGS. 2 and 3 show the state of the integral calculation at the time of temperature rise, and FIG. 2 shows that the current temperature target value (set2) is the previous temperature target value (set) when, for example, the setting is changed.
1) The case where it becomes higher than (1) (| set-now | ≧ E) is shown. In other words, the integral calculation is not performed when the measured temperature value is outside the range of the deviation (E), and the integral calculation is performed when the measured temperature value falls within the range (E) of the deviation.

FIG. 3 shows the temperature target value (set) of this time.
2) is changed within the deviation width (E) from the previous temperature target value (set1) (| set-now | <E)
The example of the integral calculation is shown. In this case, the integral calculation is continuously performed.

FIG. 4 and FIG. 5 show the state of the integral calculation at the time of temperature decrease. FIG. 4 shows that the current temperature target value (set2) is the previous temperature target value (set) due to, for example, a setting change.
The case (1) is lower than (1) (| set-now | ≧ E). In other words, the integral calculation is not performed when the measured temperature value is outside the range of the deviation (E), and the integral calculation is performed when the measured temperature value falls within the range (E) of the deviation.

Further, FIG. 5 shows the temperature target value (set) of this time.
2) is changed within the deviation width (E) from the previous temperature target value (set1) (| set-now | <E)
The example of the integral calculation is shown. In this case, the integral calculation is continuously performed.

The output processing unit 7 adds the value obtained from the integration processing unit 5 to the numerical value obtained from the look-up table 4 and outputs it to the controlled object 8 as a control output.

That is, since the present invention uses position type control, if the value given from the integration processing unit 5 is (ln), the current control output value M _vn is

[0049]

## _EQU5 ## It can be obtained by the formula of M _vn = LM _v + ln.

That is, this control output value M _vn (= LM _v +
In) is output, and time proportional on / off control is performed based on a preset output cycle.

Here, the time proportional ON / OFF control system is a system in which the control output calculated by calculation is converted into the ON / OFF output of the relay. This is to turn on the relay for a time proportional to the control output with respect to the preset output cycle, but it may be difficult to deal with the fluctuation of the output rate because of the time cycle. is there. That is, as shown in FIG. 6, even if the calculation result is 0%, if the output cycle for the previous calculation has not ended, there is a situation in which the output is not updated until the cycle ends. Therefore, in order to realize more precise control (in particular, to eliminate overshoot and minimize settling time), it is necessary to eliminate the control deviation due to the output cycle.

Therefore, in the present invention, as shown in FIG.
The output rate of the previous time (time t1) was 0% or more, and this time (time t1).
When the output rate in 2) is 0%, the output is cleared, and when the output changes from 0% to 0% or more (time t3), the output cycle is initialized and the control is executed again. ing.

[0053]

According to the fuzzy temperature control system of the present invention,
A look-up table in which a rule composed of deviations with respect to the target temperature and changes in the deviations is developed as numerical data in the data table of the deviation-change plane, and the deviations are integrated and a value obtained by multiplying them by a constant coefficient is output. An integration processing unit and an output processing unit that obtains a control output of a control target by adding a value obtained from the integration processing unit to a numerical value obtained from a look-up table based on a deviation with respect to a target temperature and a change in the deviation. Since the configuration is provided, it is possible to realize a more stable and high-performance temperature control by eliminating a large fluctuation in the control output. In addition, the non-linear control by fuzzy control, which has been considered difficult in the past, can be realized more easily and with high accuracy.

[Brief description of drawings]

FIG. 1 is a block diagram showing an electrical configuration of a fuzzy temperature control system of the present invention.

FIG. 2 is a graph showing a state of integration calculation during temperature rise.

FIG. 3 is a graph showing a state of integration calculation during temperature rise.

FIG. 4 is a graph showing a state of integration calculation during temperature reduction.

FIG. 5 is a graph showing a state of integration calculation during temperature reduction.

FIG. 6 is a graph showing an example of time proportional ON / OFF control in the output processing unit.

FIG. 7 is a table showing general fuzzy rules.

FIG. 8 is a diagram in which the fuzzy rules shown in FIG. 7 are normalized.

[Explanation of symbols]

 4 Look-up table 5 Integration processing unit 7 Output processing unit 8 Control target

Claims (1)

[Claims] 1. A system for controlling a dead time plus a first-order lag system or a second-order lag system by applying fuzzy reasoning, the difference between a deviation with respect to a target temperature and a deviation this time with respect to a previous deviation (change in deviation). ) And a look-up table in which a deviation-change plane data table is expanded as numerical data, and an integration processing unit that integrates deviations with respect to the target temperature and outputs a value obtained by multiplying this by a constant coefficient. And an output processing unit that obtains the control output of the control target by adding the value obtained from the integration processing unit to the numerical value obtained from the lookup table based on the deviation with respect to the target temperature and the change in the deviation. A fuzzy temperature control system characterized by being equipped.