Design and implementation of an adaptive neuro-fuzzy inference system on an FPGA used for nonlinear function generation

Abstract

This paper presents a digital system architecture for a two-input one-output zero order ANFIS (Adaptive Neuro-Fuzzy Inference System) and its implementation on an FPGA (Field Programmable Gate Array) using VHDL (VHSIC Hardware Description Language). The designed system is used for nonlinear function generation. First, a nonlinear function is chosen and off-line training is carried out using MATLAB ANFIS to obtain the premise and consequence parameters of the fuzzy rules. Then, these parameters are converted to a binary fixed-point representation and are stored in read-only memories of the VHDL code. Finally, simulations are performed to verify the system operation and to evaluate the system response time for given input data.