Implementation of nonlinear fuzzy models using microcontrollers

Abstract

This paper presents an algorithm for implementation of fuzzy systems using inexpensive general-purpose hardware. The two major difficulties in fuzzy system implementation using 16 bit fixed point micro processors/controllers is in handling floating point arithmetic and the determination of the floating-point membership value. Existing membership functions (mfs), do not satisfy simultaneously ease in optimization and low end hardware implementation, hence new membership function that satisfies the two contradicting requirements have been proposed by us [1]. An algorithm for hardware implementation of fuzzy systems using fixed-point micro-controllers is proposed. The worst-case execution time is computed as a function of the number of inputs and the number of rules. Optimized fuzzy model of a benchmark system has been coded and implemented using Intel 8XC196KC micro controller.