VHDL implementation of very high-speed integer fuzzy controller

For high-speed fuzzy control systems, an important problem is the improvement of speed for the fuzzy inference, particularly in the consequent pan and the defuzzification stage. This paper introduces an algorithm to map real values of the fuzzy membership functions in the consequent pan onto an integer grid, as well as a method of eliminating the unnecessary operations of the zero items in the defuzzification stage, allowing a center of gravity method to be implemented with only integer additions and one integer division. A VHDL implementation of the system is presented. The proposed system shows approximately an order of magnitude increase in speed as compared with conventional methods while introducing only a minimal error and can be used in many fuzzy controller applications.