DE4026995A1 - Control circuit of fuzzy logic type - has measuring sensor for actual value coupled to differentiator and subtraction input of first summator - Google Patents

Abstract

In the circuit a control signal (U) is generated from the values of a table by an allocation function. A measuring sensor (D) measures the actual value and is coupled to the input of a differentiator (DF) and to the subtraction input of a first summator (Sl). The rated value (SW) is applied to its adding input. The control difference (E) between the actual and rated values at the first summator output is fed to the control's (RG) first input. The differentiator output is coupled to the control's second input. By the allocation function the control generates a digital control signal from the differentiator output signal and the control difference. The digital control signal is integrated and supplied to the circuit adjuster (PWM). ADVANTAGE - Optimised circuit.

Description

The invention relates to a control loop according to the fuzzy fen controls logic by using the values from a value table with the control signal is generated by means of a membership function.

In "der elektroniker", No. 3, 1990 is on page 39-43 Article by Günter Trautzl with the title "Unclear logic: Fuzzy Logic "appeared.

For the term "fuzzy logic" are next to the label Fuzzy logic still called the rule-base system, ambivalent logic and multi-valued logic in use.

The fuzzy logic was developed as early as 1965 not being able to process inaccurate and incomplete data records nen. For this purpose, with the help of a membership function on, often referred to as the membership function  a table of values e.g. B. the control signal for a control loop won.

In the cited reference, this standard procedure is on Example of a carriage explained on which a stick with a mas se is tiltably attached in the direction of movement of the carriage. The The control loop moves the carriage in a horizontal direction that the stick does not fall over with the mass, but balanced becomes. In other words, that means the car imitates the art of balancing an acrobat who is trying to balancing a stick in his cupped hand.

The location of the rod is determined by its angular velocity and describes the deflection angle, which the input variables of the Represent control loop. The output variable is the speed speed of the car.

In the literature cited, this control cycle is described on page 41 in the following words:
"The approach using fuzzy logic is as follows: The system has the two input variables angle R and angular velocity ω and an output variable velocity v. First, you collect all input and output variables in terms of value and also specify the membership function for this. For example, divides the angle is large-positive, with tel-positive, small-positive, zero, small-negative, medium-negative, large-negative, and large means all angles from 20 to 90 degrees (theoretical limit) be, with medium all angles between 5 and 25 degrees, with small all angles between 0 and 10 degrees, as zero allow all values between -0.5 degrees and +0.5 degrees. The membership function for large for example in such a way that all values above 45 degrees have a membership factor of 1, but the membership then drops to 20 degrees at 0.005. In a similar way, the value range ch is defined by w and by v. In this way e you get an assignment of the system variables to the membership function.

The second part of the approach is that now a system of rules is drawn up in the following form:

• 1st rule: If R is zero and ω is zero, then v is equal zero.
• 2nd rule: If R is zero and ω is positive-small, then v is equally positive-small.
• 3rd rule: If R is positive-small and ω is zero, then v equally positive-small.
• 4th rule: If R is zero and ω is negative-small, then v is equal negative-small

and so on. Through such a field of rough, approximate The entire system can describe rules (mostly empirical values) be practiced. "

It is an object of the invention to optimize a control loop ren, who works on the principle of fuzzy logic.

The invention solves this problem in that a sensor, which measures the actual value, with the input of a differentiator and ver with the subtraction input of a first summation point is bound to the addition input of the setpoint that the control difference E - the difference between the actual value and the setpoint - at the output of the first summation point to which it Most input of a controller is fed that the output of the Differentiator connected to the second input of the controller is that the controller uses the membership function from the Output signal ΔE of the differentiator and the control difference E a digital control signal that generates the digital control signal  is integrated in an integrator at the output of the controller and that the integrated control signal the actuator of the rule circle is fed.

Show it

Figure 1 shows an embodiment of the invention.

Fig. 2 blen an embodiment of a matrix with input variables;

Fig. 3 shows an example of a membership function.

The invention will now be described and explained with reference to the embodiment shown in FIG. 1.

The speed of the electric motor M is detected by means of a sensor D which controls a counter Z. The counter reading of the counter Z is therefore always proportional to the instantaneous speed of the electric motor M. The output of the counter Z is connected to the subtraction input of a summation point S 1 and the addition input of a summation point S 2 . In addition, the output of the counter Z is connected to the subtraction input of the summation point S 2 via a delay element VZ1. The setpoint SW lies at the addition input of the summation point S 1 . The control difference E - the difference between the actual value IS and the setpoint SW - can therefore be removed at the output of the summation point S 1 . The control difference E at the output of the summation point S 1 and the output signal ΔE of the summation point S 2 are fed to a controller RG, whose digital output signal ΔU is fed to the first input of a summation point S 3 . The output of the summation point S 3 is connected to the input of a pulse width modulator PWM, which controls the electric motor M, and is connected via a delay element VZ 2 to the second input of the summation point S 3 .

The summation point S 2 and the delay element VZ 1 form a differentiating element DF, which differentiates the output signal of the counter Z. The integrator I, which is composed of the summation point S 3 and the delay element VZ 2 , integrates the digital control signal ΔU of the controller RG.

The digital control signal E and the differentiated digital output signal ΔE of the counter Z are fed to the controller RG. In the controller RG are in a matrix, for. B. the 7 × 7 matrix shown in FIG. 2 stores values for the digital signals E and ΔE ge. Using the membership function, the controller RG looks for the digital control signal ΔU from the matrix. The found control signal ΔU is integrated by means of the summation point S 3 and the delay element VZ 2 ; The integrated control signal U is fed to the input of the pulse width modulator PWM, which controls the electric motor M.

It is particularly advantageous to store the values symmetrically to the diagonal in the matrix. In the exemplary embodiment of a 7 × 7 matrix shown in FIG. 2, all column positions are each assigned a value in the first row, only a fourth column position is assigned a value in the second row, and those in the third row are Column positions 5 to 7 are assigned a value, in the fourth line all column positions are each assigned a value, in the fifth line the first four column positions are each assigned a value, in the sixth line only the fourth column position is assigned a value and finally, in the seventh row, all column places are each assigned a value.

The symbols used in the matrix have the following meaning:
NG: negative large,
NM: negative medium,
NK: negative-small,
ZO: zero,
PK: positive-small,
PM: positive medium,
PG: positive big.

If e.g. B. E and ΔE at the inputs of the controller RG the value ZO have, the digital control signal ΔU is the corresponding function also ZO. In contrast, E has the value PK and ΔE the value PG, the controller selects the value PG for ΔU.

An example of a membership function is shown in FIG. 3. Eight bits are used to determine the values of the output variables ΔU of zero, small, medium and large. Eight bits are also provided to determine the values of the input variables E and ΔE.

The values for the output variable ΔU and the input variables E and ΔE are empirical values. The for the respective rule best values for the output variable and the input variables are determined in experiments.

The invention is generally suitable for control loops according to using the fuzzy logic method.

Claims (6)

1. Control loop which controls according to the fuzzy logic method by generating the control signal (U) from the values of a value table by means of a membership function, characterized in that a sensor (D) which measures the actual value (IW) with which Input of a differentiator (DF) and connected to the subtraction input of a first summation point (S 1 ), at whose addition input the setpoint (SW) is present,
that the control difference E - the difference between the actual value (IW) and the setpoint (SW) - at the output of the first summation point (S 1 ) is fed to the first input of a controller (RG),
that the output of the differentiator (DF) is connected to the second input of the controller (RG),
that the controller (RG) generates a digital control signal (ΔU) by means of the membership function from the output signal ΔE of the differentiator (DF) and the control difference E,
that the digital control signal (ΔU) at the output of the controller (RG) is integrated in an integrator (I) and
that the integrated control signal (U) leads to the actuator (PWM) of the control circuit. 2. Control loop according to claim 1, characterized in that
that the output of the sensor (D), which measures the speed of an electric motor (M), is connected to the input of a counter (Z), the counter reading is proportional to the speed of the electric motor (M), that the output of the counter ( Z) is connected to the subtraction input of the most summation point (S 1 ) and to the input of the differentiator (DF),
that the integrated Regelsi signal (U) at the output of the integrator (I) is fed to a pulse width modulator (PWM) which controls the electric motor (M). 3. Control circuit according to claim 2, characterized in that the output of the counter (Z) with the addition input of a second summation point (S 2 ) is connected, the output of which is connected to the second input of the controller (RG), and that the output of the counter (Z) is connected via a first delay element (VZ1) to the sub traction input of the second summation point (S 2 ). 4. Control circuit according to claim 1, 2 or 3, characterized in that the output of the controller (RG) is connected to the first addition input of a third summation point (S 3 ), the output of which is connected to the input of the pulse width modulator (PWM) and via a second delay element (VZ 2 ) is connected to the second addition input of the third summation point (S 3 ). 5. Control loop according to claim 1, 2, 3 or 4, characterized characterized in that several in an nxn matrix re digital values for the signals E and ΔE are stored are and that the column spaces of the matrix are symmetrical too of a diagonal are assigned values. 6. Control loop according to claim 5, characterized in that a 7 × 7 matrix is provided,
that in the first line of the matrix all column spaces are filled with values,
that only the fourth column position is assigned a value in the second row,
that in the third line the fourth, fifth, sixth and seventh column positions are each assigned a value, that in the fourth line all column positions are each assigned a value,
that in the fifth row the first, second, third and fourth column spaces are each assigned a value,
that in the sixth row only the fourth column position is assigned a value and
that in the seventh row all column positions are each assigned a value.