JPH0367334A - Limitation relaxing inference system for expert constructing tool - Google Patents

Abstract

PURPOSE:To carry out a limitation relaxing inference with use of a simple program by defining previously the limitation intensity for each limitation rule and omitting the application of the limitation rules in the order of lower limitation intensities. CONSTITUTION:A production planning expert system 1 consists of a limitation rule group 2 decided previously, a product order group 3, and an inference part 4 which assigns the production plans of products based on the group 2. The group 2 includes three limitation rules with the limitation intensity defined to each rule, e.g., the limitation intensities 1 - 3 are defined in the order of higher intensities. Then the rules of intensities 1 - 3 are applied and the production of a product Y is impossible under the conditions of these rules. In such a case, the part 4 omits the application of the rule 3. Then the part 4 omits the rules 2 and 1 in sequence until a solution is obtained for production of the product Y. In such a constitution, a limitation relaxing inference is attained with use of a simple program and with execution of the limitation relaxing inferences.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a constraint relaxation inference method for expert construction tools.

[Conventional technology]

An example of a conventional constraint relaxation inference method will be explained.

First, it is assumed that one solution is found from a set of candidate solutions by applying the following two rules.

Ruled: In the A1 condition, the one with element A is the solution.

Rule 2...In the B1 condition, a solution is one that has element B.

Here, when a solution cannot be found by applying these two rules, when using a relaxation method that removes the application of Rule 2, Rule 1: IF x≧3 AND Condition AITH
EN Solution=hold element A.

Rule 2: IF X≧2 AND condition A2TH
EN Solution = retain element B.

It is described as above. Then, first set X=2,
Make both rules applicable. If a solution cannot be found here, set X=3 and perform control such that only rule 1 is applied.

[Problem to be solved by the invention]

However, in such a conventional method, a condition for checking the constraint strength is necessarily required in the constraint rule, and a rule for controlling this is also required. Therefore,
Not only is the program complex, but it also requires a lot of man-hours to modify, and there are also many mistakes.

The present invention was made in view of these circumstances, and its purpose is to provide a constraint relaxation inference method for an expert system construction tool that allows constraints to be easily relaxed.

[Means to solve the problem]

In order to solve the above problems, the present invention defines a constraint strength for each predetermined constraint rule, and when a solution cannot be found by applying all the previous constraint rules, the previous constraint strength is increased until a solution is found. It is characterized by performing inference for relaxing constraints by removing the application of constraint rules in descending order of weakness.

[For production]

In the present invention, constraint strength is defined for each predetermined constraint rule, and each constraint rule is ranked. Then, if a solution cannot be found by applying all of the constraint rules, the ranked constraint rules are removed in descending order of constraint strength. Exemption from this constraint rule is
This is done sequentially until a solution is found.

〔Example〕

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. FIG. 1 is a block diagram showing an embodiment of the constraint relaxation inference method of the expert system construction tool of the present invention.

In FIG. 1, the production planning expert system 1 is
It was constructed using an expert system construction tool. This production planning expert system 1 is comprised of a predetermined constraint rule group 2, an order group 3 for allocating products, and an inference unit 4 that allocates a product production plan based on the constraint rule group 2.

In this embodiment, the constraint rule group 2 includes the following three constraint rules, and the constraint strength is defined for each constraint rule.

Constraint rule 1: Observe product delivery deadlines. (Constraint strength = 3) Constraint rule 2: Products with special parts A are manufactured by machine or mouth. (Constraint strength = 2) Constraint rule 3: Product Y will not be manufactured next to product X. (Constraint strength = 1) The constraint strength is defined from 1 to 3, and the greater the number, the stronger the constraint. Therefore, rule 3 has the weakest restriction, and rule 1 has the strongest restriction. In this embodiment, assigned order group 2 includes product Y, and the product conditions are that the delivery date is December 20th, special part A is used, and the manufacturing period is 4 days.

Here, the method for allocating the product Y will be explained. First, the inference unit 4 tries to allocate the production plan by applying the three constraint rules of the constraint rule group 2. In this case, as shown in the Gann chart of the production plan in FIG. 2, a schedule that satisfies all of the constraint rules cannot be established, and product Y cannot be produced under this condition.

Therefore, the inference unit 4 removes the application of constraint rules in order of decreasing constraint strength. First, the inference unit 4 removes the application of the constraint rule 3, which has the weakest constraint strength, and assigns the product Y to the production plan. Specifically, the application of constraint rule 3, "Product Y will not be manufactured after product X," is removed. By removing this application, as shown by diagonal lines in Figure 2, the distance between the product
Product Y can be allocated by day 0.

In this way, when the inference unit 4 cannot find a solution, it sequentially removes the application of constraint rules from the weakest constraint strength until the solution is found. Therefore, since constraint relaxation inference is performed by sequentially removing constraint rules from weakest to weakest, there is no need for a complicated program like in the past.
Constraint relaxation inference can be executed with a very simple program.

[Effects of the Invention] As explained above, according to the present invention, the constraint strength is defined for each constraint rule, and the application of the constraint rules is removed in descending order of constraint strength until a solution is found. This has the effect of allowing constraint relaxation inference to be executed with a much simpler program compared to .

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention, and FIG. 2 is a Gantt chart showing an example of a product production plan. 1...Production planning expert system, 2...Sho 1
Itotori rule group, 3... Assignment order group, 4... Guidance Ha
Department.

Claims (1)

[Claims] A constraint strength is defined for each predetermined constraint rule, and if a solution cannot be found by applying all the constraint rules, the application of the constraint rules is sequentially removed in descending order of constraint strength until a solution is found. This is a constraint relaxation inference method for expert systems that performs constraint relaxation inference by