JPH02277137A - Rule producing device - Google Patents

Abstract

PURPOSE:To simplify a task for production of a rule group by selecting plural basic rule groups from a storage means and synthesizing these rule groups based on a prescribed rule for production of a desired rule group. CONSTITUTION:A storage 1 stores previously plural types of basic rule groups, and a rule selector 2 selects 2 or more basic rule group to be synthesized out of the storage 1 and gives them to a rule processor 3. The processor 3 synthesizes the basic rule group selected by the selector 2 based on a prescribed rule to produce a desired rule group. The processor 3 owns a function to correct one of rule groups based on a prescribed rule for synthesization of these rule groups if the basic rule groups to be synthesized include the rules having the conflicting components. Thus a rule corrector 4 performs the correction or the addition of rules as necessary to the rule groups produced by the processor 3. A rule setter 5 sets the new rule groups produced through the processor 3 and the corrector 4 at a fuzzy inference part (not shown in the diagram).

Description

[Detailed Description of the Invention] <Industrial Application Field> The present invention relates to a rule generation device used for generating fuzzy rules. The present invention relates to a rule generation device for generating a target fuzzy rule.

<Prior Art> Generally, in fuzzy control, when an input signal is given to a fuzzy inference section, a predetermined set of rules is activated, and a control operation of a controlled object is determined by inference based on this set of rules.

The rule group includes many fuzzy rules that share roles in order to achieve a predetermined control objective, and these fuzzy rules are usually expressed in a format called an if-then rule.

The above-mentioned rule group is generated for each application, and conventionally, an operator creates fuzzy rules one by one through a man-machine interface and sets them in the fuzzy inference section. Each time there is a change in the application, a new set of rules must be created and modified manually.

<Problems to be solved by the invention> However, generating and setting a group of rules using this method is extremely troublesome for the operator, especially when creating and modifying rules using the same method for each application. The workload involved in doing so is extremely large. Furthermore, even if rule groups are generated and stored one after another, each rule group is only used for its own application and is not effectively used to generate new rule groups.

This invention was made by focusing on the above problem. A basic rule group is prepared in advance, and when generating a new rule group, the user selects and synthesizes the rule group. The purpose is to facilitate the creation and modification of rules, as well as to make effective use of the accumulated rules.

Means for Solving the Problems> A rule generation device according to the present invention includes a storage means for storing a plurality of types of basic rule groups in advance, a selection means for selecting a plurality of basic rule groups from the storage means, A rule generating means is provided for synthesizing the selected basic rule group based on a predetermined rule to generate a target rule group.

In addition, in the present invention, in order to prevent the basic properties of the rule group from being destroyed by synthesis, the rule generation means is configured to: A function is provided to modify one of the rule groups based on a predetermined rule.

Effect> The manufacturer side prepares a plurality of basic rule groups in advance, stores them in a storage means, and supplies them to the user side. On the user side, when generating a target rule group, the user selects a plurality of basic rule groups from the storage means and combines them based on a predetermined rule to generate the target rule group.

Therefore, the work of generating, setting, and modifying a rule group for each application is significantly simplified, and the work burden on the operator is reduced. Furthermore, if the generated rule group is stored in a storage means, it can be effectively used to synthesize a new rule group.

Furthermore, when composing rule groups, ``Even if the basic rule group to be synthesized contains a rule with competing rule components, one of the rule groups is modified based on a predetermined rule, so the basic rule group of the rule group can be modified by compositing. properties are prevented from being destroyed.

<Embodiment> FIG. 1 shows a schematic configuration of a rule generation device according to an embodiment of the present invention, in which a storage device 1, a rule i! [Use 2, rule processor 3. Rule corrector 4. It includes a rule setting device 5 and the like.

The storage device 1 is a part that stores in advance a plurality of basic rule groups. The rule selector 2 selects two or more basic rule groups to be synthesized from the storage device 1 and supplies them to the rule processor 3.

The rule processor 3 synthesizes the basic rule group selected by the rule selector 2 based on a predetermined rule and processes it into a target rule group. This rule processor 3 has a function to modify and synthesize one of the rule groups based on a predetermined rule when each basic rule group to be synthesized includes rules with competing rule components, as will be described in detail later. . The rule corrector 4 is a part for correcting or adding rules to the rule group generated by the rule processor 3 as necessary. The rule setter 5 sets the new rule group generated by the rule processor 3 and the rule corrector 4 in a fuzzy inference section (not shown).

FIG. 2 shows the format of the basic rule group stored in advance in the storage device 1 in the form of a table. In the figure, two types of input θ1゜θ are shown using fuzzy labels at the left and top ends, and at the vertical and horizontal intersections of the respective fuzzy labels, there are labels to specify what kind of action to perform in response to the input. The output U of is set with a fuzzy label. Here, there are seven types of fuzzy labels, and PL (Positive
Large), Ps (Positive Mi
dium).

Ps (Positive Small), Z+e
(Zero), Ns (Negative Small)
l), Ns (Negative Medium)
.

It is expressed as NL (Negative Large).

FIGS. 3(1) to 3(8) show examples of shapes of a plurality of basic rule groups stored in advance in the storage device l.

For each basic rule group, the set pattern of the rule group, that is, the arrangement shape when the fuzzy label of the output U is written in the table of Fig. 2, forms a predetermined basic form, and Fig. 3 (1)
Figure 3 (2) is a diagonal crossover shape, Figure 3 (3) is a rectangular ring shape, Figure 3 (4) is a left-down shape, and Figure 3 (5) is a right-hand shape. The shape of the drop, Fig. 3 (6
) is the shape of one horizontal row, Figure 3 (7) is the shape of one vertical row,
Figure (8) has a rectangular block shape.

Thus, the storage device 1 stores the arrangement pattern (shape) and the contents (components) of the fuzzy labels arranged for these basic rule groups.

FIG. 4 shows an example of a basic rule group having the set pattern shown in FIG. 3 (6). The basic rules in the diagram are:
One input θ is fixed at P, and the rule group in the table is expressed as a 1f-then statement as follows.

f θI =P S r θz=Pt then
PLf θI=P! , θz=Ps then
P3f θl ”= P S r θt=Za
then Z.

f θ, −p , θt = Ns then N
3f θI ”' P S * θz = Nt t
hen NL In each of the above fuzzy rules, the fuzzy label of the consequent corresponds to the component of this basic rule group, and assuming that the condition of input θ1 is P, this is expressed as M
When written as (P,), this basic rule group can be expressed as M (Ps, Vl) when the rule component is represented by vl.

Returning to FIG. 1, the rule selector 2 of this embodiment selects two basic rule groups and provides them to the rule synthesizer 3, and the rule synthesizer 3 converts these two basic rule groups into predetermined rules. are synthesized based on the rules and processed into the desired set of rules.

5 and 6 show two basic rule groups selected by the rule selector 2 in table form, and the rule group in FIG. 5 is of the type shown in FIG. 3 (6). The rule group in FIG. 6 is the type of rule group shown in FIG. 3 (7).

Each basic rule group includes seven fuzzy rules, among which 'if θ,=21I.

0g = N, then...,...'' (indicated by diagonal lines in the figure), the fuzzy label of the consequent in the rule group of Figure 5 is Ns, and the fuzzy label of the consequent in the rule group of Figure 6 is Ns. The fuzzy label is ZII, and both rule components will conflict.

The rule processing device 3 creates a basic rule group M shown in FIG.
(Z * + V t ) and the basic rule group P (Ns, Vl) shown in FIG.
(Zl, v, ) +P (NS, Vl
), and in this case, the basic rule group shown in FIG. 5 is used as a rule with a high priority level, and the competing fuzzy rules are processed.

In other words, for competing fuzzy rules, after fixing the component of the fuzzy rule with a high priority level (Ns in this case), the component of the fuzzy rule with a low priority level is modified from Zl to N, and the amount of modification ( NsZ8)
is calculated by a rule addition/subtraction device. In this case, the amount of correction is N.

Z*=Ns, and the correction amount is added to the components of other fuzzy rules using an adder/subtractor to perform correction processing.

FIGS. 7(1) and 7(2) show addition and subtraction rules applied to addition and subtraction by the addition and subtraction device. Here, the addition/subtraction rules mean rules for executing addition or subtraction, and these rules are shown in the form of a table in FIGS. 7(1) and (2).

In the figure, the label column a at the left end and the label column S at the top end are fuzzy labels that are to be added or subtracted, and the addition result is displayed at the intersection of the vertical and horizontal labels in Figure 7 (1). However, in Fig. 7 (2), the subtraction results are indicated by fuzzy labels. The fuzzy label is Pt.

Figure 8 shows the result of modifying the basic rule group in Figure 6 by this addition/subtraction process, and by superimposing the basic rule group related to this modification on the basic rule group in Figure 5, the final result is A rule group as shown in FIG. 9 will be generated. Then, after the rules are corrected or added to this rule group by the rule corrector 4 as necessary,
It is sent to the rule setter 5 and set to the fuzzy inference section.

FIG. 10 shows the procedure for generating a new rule group by this rule generation device.

First, in step 1 of the figure (indicated by rsTIJ in the figure), a plurality of basic rule groups to be synthesized are selected from the storage device 1 by the rule selector 2, and priority levels of the basic rule groups to be synthesized are set. be done.

In the next step 2, the rule processor 3 reads the shape and each component of the two basic rule groups to be synthesized, and then in step 3 it determines whether or not there are competing fuzzy rules. . If there is no competing fuzzy rule, proceed to step 4, each fuzzy rule of each basic rule group is expanded, and if, t
It is reduced to a fuzzy language equivalent to the hen rule.

If there is a competing fuzzy rule,
The process proceeds from step 3 to step 5, and after detecting competing fuzzy rules for the rule group with a low priority level, the above-described modification process is performed, and the process proceeds to step 4.

In step 4, a group of basic rules with a high priority level is first developed and translated into a fuzzy language, and then a group of modified basic rules with a low priority level is developed.

In the next step 6, it is determined whether there are further basic rule groups to be synthesized, and if the determination is “YF
! S'', the process returns to step 2 and the same procedure as above is executed.

If the determination in step 6 is "NO", then in step 7 it is determined whether or not there is a rule modification for the combined rule group. If the determination is "NO", the rule setting device 5 sets the rules to the fuzzy inference section, but if the determination in step 7 is "YES", the rule modification device 4 sets the rules in step 8. After corrections and additions are made, a set of rules related to the corrections will be set in the fuzzy inference section.

<Effects of the Invention> As described above, the present invention stores a plurality of types of basic rule groups in advance in a storage means, selects a plurality of basic rule groups from the storage means when generating rules, and uses the selected basic rule groups. Since the target rule group is generated by combining rule groups based on predetermined rules, the work of generating, setting, and even modifying rule groups for each application is significantly simplified and faster. This reduces the workload of the person. Another advantage is that by storing the generated rule group in the recording means, it can be effectively used to synthesize a new rule group.

Furthermore, when composing a group of rules, even if the basic rule group to be synthesized contains rules whose rule components conflict with each other, one of the rule groups is modified based on a predetermined rule, so the basic properties of the group of rules resulting from the synthesis This invention achieves the purpose of the invention and achieves remarkable effects, such as preventing damage to the device.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing a schematic configuration of a rule generation device according to an embodiment of the present invention, FIG. 2 is an explanatory diagram showing the format of a rule group in the form of a table, and FIG. 3 is a shape of a basic rule group. An explanatory diagram showing an example, Fig. 4 is an explanatory diagram showing an example of a basic rule group in the form of a table, and Figs. 5 and 6 are explanatory diagrams showing an example of the basic rule group to be synthesized in the form of a table. , Figure 7 is an explanatory diagram showing the addition and subtraction rules in the form of a table, Figure 8 is an explanatory diagram showing the modified basic rule group in the form of a table, and Figure 9 is an explanatory diagram showing the final form of the combined rule group. FIG. 10 is a flowchart showing the procedure for generating rules. 1...Storage device 2...Rule selector 3
・・・・Rule processor

Claims (2)

[Claims] (1) A rule generation device for generating fuzzy rules, comprising: a storage means for storing in advance a plurality of basic rule groups; a selection means for selecting a plurality of basic rule groups from the storage means; A rule generation device that generates a target rule group by synthesizing basic rule groups based on predetermined rules. (2) According to claim (1), the rule generation means has a function of modifying one of the rule groups based on a predetermined rule when the basic rule group to be synthesized includes rules with competing rule components. The rule generation device described.