Title:
ARTIFICIAL NEURAL CIRCUIT NETWORK TYPE FUZZY CONTROLLER
Document Type and Number:
Japanese Patent JPH02292602
Kind Code:
A
Abstract:
PURPOSE:To efficiently construct a fuzzy control system with high performance and to improve control performance by simultaneously adjusting membership functions used for a condition part and a conclusion part. CONSTITUTION:#1 and #2 state variables are inputted from an input part 1 and a prescribed value from a generation part 2. The arithmetic neurone part 4 of an artificial neural circuit network type fuzzy controller individually multiplies a couple of adjustable synapse loads corresponding to a parameter showing the central values of the membership functions of respective fuzzy control rule condition parts 3, and shapes with respect to inputs, and the part 4 operates assignment degrees for respective membership functions. A neurone part 5 operates adaptation for respective control parts 3 and a neuron part 6 operates the total sum of the areas of fuzzy sets for respective fuzzy variables. A non-fuzzy part 8 obtains control variable values and an error operation part 9 calculates the error of the control variable values and a control variable value being a criterion and obtains a feedback signal. Consequently, an error signal is controlled so that it becomes the smallest.
Inventors:
NISHIGUCHI EI
SUEMITSU KEIJI
SUEMITSU KEIJI
Application Number:
JP11203389A
Publication Date:
December 04, 1990
Filing Date:
May 02, 1989
Export Citation:
Assignee:
NIPPON KOKAN KK
International Classes:
G05B13/02; G06F9/44; G06F15/18; G06N3/00; G06N7/02; G06N99/00; (IPC1-7): G05B13/02; G06F9/44; G06F15/18
Attorney, Agent or Firm:
Muneharu Sasaki (1 person outside)
Previous Patent: IDENTIFICATION DEVICE FOR CONTROL SYSTEM
Next Patent: METHOD FOR CONTROLLING ORDER INSTRUCTION CARD
Next Patent: METHOD FOR CONTROLLING ORDER INSTRUCTION CARD