METHOD FOR GENERATING FUZZY KNOWLEDGE IN EXPERT SYSTEM

摘要

PURPOSE:To efficiently generate a fuzzy knowledge to an expert system constituting tool by generating an instance distribution table indicating the distribution state of output data to input data of instance data and generating the fuzzy knowledge from this table. CONSTITUTION:An instance distribution calculating part 11 reads out instance data from an instance data file 12 to totalizes instance data and displays an instance distribution diagram on a display device 13. An operator indicates the instance distribution calculating part 11 to section the input/output data by a keyboard 14 or a mouse 15 so that the distribution of instances can be explained well, and the instance distribution table is outputted to an instance distribution file 16. A rule extracting part 17 takes the instance distribution table as the input from the instance distribution file 16 and analyzes the frequency of instances to generate fuzzy rules and membership functions and outputs them to a fuzzy rule file 18. Thus, the fuzzy knowledge of the expert system constituting tool is efficiently generated from past instances.