JPH04160645A - Schedule formation system - Google Patents

Abstract

PURPOSE:To improve the modification efficiency of a schedule formation system by recording an inference process and analyzing the process after completing the inference. CONSTITUTION:The schedule formation system is provided with an inference processing part 2 having a task selecting means 3, a means 4 for selecting a resource to which a task is to be allocated, a means 5 for selecting a task and a resource allocating method, a means 6 for applying restriction to the set of selected resources, a means 7 for determining the candidate of an allocating resource, and a means 8 for controlling respective means. In addition, the system includes an inference process recording part 9 for recording the results of respective means in the processing part 2, an inference process output means 14 having a display means 15 for displaying an optional part of a storage, a means 16 for editing records and an analytical means 17 for offering a parameter or the like to be used for schedule formation based upon the records. In this case, the means 17 calculates the success rate of restriction application based upon a restriction application result in a recording area. While considering the success rates, the priority of restriction having a low success rate is improved. Consequently, the modification efficiency of the schedule formation system is improved.

Description

[Detailed Description of the Invention] [Field of Industrial Application] The present invention relates to a schedule creation method, in particular, to a method for creating a schedule, and in particular, to create a schedule by selecting combinations of individual tasks and resources from a set of tasks and resources that constitute a schedule. This relates to a schedule creation method that creates an entire schedule by repeating creation operations.

[Conventional technology]

Creating schedules, such as train operating schedules, work schedules, production planning schedules, etc., requires a lot of manpower and time, and is a task that occurs regularly. Such schedules are created by experienced personnel who take into account various conditions.

However, manual schedule creation is difficult to deal with sudden occurrences of conditions or urgent schedule changes. It is also difficult to create a qualitatively uniform schedule.

Therefore, a scheduler can create such a schedule.
There is also a desire to realize an information processing system that utilizes the know-how of the industry, a so-called expert system. An expert system is not something that can be completed all at once, but rather is built and refined in stages while verifying the system's results and its inference process.

The verification was performed by displaying the inference process such as tracing or breaking, or by temporarily suspending the inference to check the state at that time.

[Problems to be Solved by the Invention] In the conventional schedule creation method described above, inference has to be frequently interrupted and is inefficient. Also,
This was only a temporary presentation of information during the inference, and it was not possible to review the inference process after the inference was completed. Furthermore, it was difficult to modify the parameters that were obtained for the first time during the entire schedule creation process.

[Means to solve the problem]

In order to solve the above-mentioned problems, the method of the present invention creates an overall schedule by repeating the operation of creating individual combinations of tasks and resources from a set of tasks and a set of resources that constitute the schedule. A schedule creation method for creating a schedule includes a task selection means for selecting a task to be allocated, a resource selection means for selecting a resource to be allocated to a task, and a rule selection means for selecting a task and resource allocation method. , a constraint applying means for applying a constraint to the set of resources selected by the resource selecting means, a candidate determining means for determining one resource candidate to which a task is allocated based on the result of the constraint applying means, and each of the above-mentioned an inference processing section having an inference control means for controlling the means; an inference process recording section for recording the results of each means of the inference processing section; a display means for displaying any part of the record of the inference process recording section; The inference process output unit has an editing means for editing or simulating records, and an analysis means for presenting evaluation information such as parameters used in schedule creation based on the records, and the inference process output unit has It is characterized by making it possible to analyze

〔Example〕

Embodiments will be described below with reference to the drawings.

FIG. 1 shows the configuration of an embodiment of the present invention, and FIG. 2 shows a process explanatory diagram of the inference processing section shown in FIG. 1. FIG. 3 shows an example of the constraint application results recorded in the recording area of the inference process recording section shown in FIG. 1.

(Function) In Fig. 1, the knowledge management unit 1 provides information on the tasks and resources that constitute the schedule, the conditions for creating the schedule, and the creation method (know-how) obtained from experience.
Manage information about constraints and rules that represent for example,
Information regarding constraints includes a parameter that indicates how important a constraint is (constraint strength), and a parameter that does not indicate the order in which constraints are applied, such as priority of constraint application.

The inference processing unit 2 is a part that reads information from the knowledge management unit and actually proceeds with schedule creation, and includes a task selection unit 3, a resource selection unit 4, a rule selection unit 5, a constraint application unit 6, and a candidate determination unit. 7, including an inference control means 8.

The task selection means 3 selects one optimal task from the set of tasks managed by the knowledge management section 1. The resource selection means 4 selects a set of resources to which the task selected by the task selection means 3 is to be allocated.

The rule selection means 5 selects a rule set for allocation.

The constraint applying means 6 applies constraints to a set of resources and obtains the result. Applying a constraint is to obtain a numerical value of -6 = how much a certain resource satisfies the constraint (condition); if it is met, a higher value will be obtained, and if it is not, a lower value will be obtained. .

The candidate determining means 7 selects one optimal candidate for task allocation from the resource set based on the result of the constraint applying means 6.
Select one. For example, the one with the highest total score obtained by adding the product of each applied value obtained by the constraint applying means 6 and the constraint strength is selected as a candidate.

The inference control means 8 activates each means of the inference processing section 2 and repeats them. The inference process recording section 9 is a section that records the operation process of each means in the inference processing section for each loop.The inference process recording section 9 is a section that records the operation process of each means in the inference processing section for each loop. Including area 11.

The knowledge input unit 12 inputs and updates knowledge to the knowledge management unit. The result output unit 13 outputs the schedule result,
For example, output something like a schedule table.

Inference process output unit] 4 is a display means 15, editing means] 6,
It includes analysis means 17. The display means 15 reads an arbitrary loop from the recording area and outputs and displays it. Editing means 16
outputs and displays the inference process on the editing screen, and changes parameter values.

The analysis means 17 presents evaluation information throughout the recording for each loop of the recording area.

(Operation) Next, the operation of this embodiment centered on the inference processing unit 2 and inference process recording unit 9 will be described using FIGS. 2 and 3.

First, in FIG. 2, it is determined whether there is a task to be selected by the completion determination process 18. If there is a task, one task is selected by the task selection process 19, and the information is recorded by the recording process 24 in the inference process. It is recorded in the recording area shown in FIG. 3 of the recording section. If there are no tasks to select,
Finish creating the schedule.

Next, a resource set selection process 20 selects a resource set, and the recording process 25 records the information in the inference process recording section 9. Similarly, the rule set selection process 21 selects a rule set to be used for allocation, and the recording process 26 records it in the inference process recording unit 9. Furthermore, constraint application processing 2
The results obtained by step 2 are recorded in the inference process recording section 9 by recording processing 27. Finally, a candidate determination process 23 determines one resource to be assigned to a task from the result of the constraint application process 22, and the process 28 records this information in the inference process recording unit 9. When the inference is completed, the result output unit 13 outputs the results in the form of a schedule table.

Assume that task A and resource 1 are combined in this schedule table. Therefore, the reason why resource 1 was selected as a candidate for task A will be analyzed. First, the display means 15 of the inference process output unit 14 reads information on the constraint process result of task A from the recording area 11 and outputs it.

As shown in Figure 3, constraint 1, constraint 2, constraint 3, constraint 4
The overall evaluation, which is the product of the application result and the constraint strength, is 9700 for resource 1, 2600 for resource 3, and 2 for resource 4.
300, and resource 1 has the highest overall evaluation score, so it can be seen that resource 1 was selected.

Next, the editing unit 16 displays the constraint application results for task A output by the display unit 15 on the editing screen. On the screen, if the application result 10 of constraint 3 of resource 4 is 20,
The overall evaluation has been recalculated and the overall evaluation for resource 4 is now 3200.
Then, resource 3 is changed to rank 3, resource 4 is changed to rank 2, and the result is simulated on the screen.

Here, we will evaluate the priority of constraint application, which is one of the parameters used in schedule creation. The analysis means 17 calculates the success rate of constraint application based on the result of constraint application of the recording area. Taking this success rate into consideration, for example, constraints with a low success rate are given higher priority.

〔Effect of the invention〕

As described above, according to the present invention, by recording the inference process, the process can be analyzed after the inference is completed, and the efficiency of improving the schedule creation system can be improved. It is also possible to extract evaluation information about the inference itself from the records.

[Brief explanation of the drawing]

FIG. 1 is a configuration diagram of an embodiment of the present invention, FIG. 2 is an explanatory diagram of the processing of the inference processing section shown in FIG. 1, and FIG. 3 is a recording area of the inference processing process recording section shown in FIG. 1. An explanatory diagram is shown. DESCRIPTION OF SYMBOLS 1... Knowledge management unit, 2... Inference processing unit, 3... Task selection means, 4... Resource selection means, 5... Rule selection means, 6... Constraint application means, 7. ... Candidate determining means, 8... Inference control section, 9... Inference process recording section, 1
0... Recording management means, 11... Recording area, 12.
...Knowledge input section, 13...Result output section, 14...Inference process output section, 15...Display means, 16...Editing means, 17...Analysis means.

Claims (1)

[Claims] A schedule creation method that creates an entire schedule by repeating operations for creating individual combinations of tasks and resources from a set of tasks and a set of resources that constitute the schedule. , a task selection means for selecting a task to be allocated, a resource selection means for selecting a resource to be allocated to a task, a rule selection means for selecting a task and resource allocation method, and a resource selected by the resource selection means. a constraint applying means for applying a constraint to a set of , a candidate determining means for determining one resource candidate to which a task is allocated based on the result of the constraint applying means, and an inference control means for controlling each of the means. an inference processing unit having an inference processing unit; an inference process recording unit that records the results of each means of the inference processing unit; a display unit that displays any part of the record of the inference process record unit; and an editing unit that edits or simulates the record. and an analysis means for presenting evaluation information such as parameters used in schedule creation based on the records, making it possible to analyze the inference process after the inference is completed. Featured schedule creation method.