摘要 |
PURPOSE:To provide a tuning device of membership functions for automatically tuning conclusion membership functions individually every learning data in a short period of time. CONSTITUTION:This device is provided with an error calculation part 62 for calculating the difference between the output result of fuzzy inference using singletons for the conclusion membership functions and an output expected value for an input value, a multiplier 63 for calculating a distance proportionate to the ratio of the heights of the respective conclusion membership functions to the total sum of the heights of all the conclusion membership functions in the defuzzy operation of inference and an adder 65 for calculating the position of the conclusion membership function based on the calculated results of a deviation calculation means and a distance calculation means.
|