Abstract:
In order to solve the problem of conventional CMOS analog circuits to achieve high computing precision of fuzzy neural network problem,a kind of multi-voltage input fuzzy neural network model was proposed.The proposed network included Gaussian function circuit,minimization circuit and a defuzzifier circuit,to achieve a fuzzy circuit performance optimization.Each unit of Analog circuits is improved to achieve high-precision,high-speed fuzzy neural network,and to achieve a nonlinear function approximation through.All unit circuits and whole system were implemented under TSMC 0.18 μm standard technology.Simulation results show that the proposed circuit has the characteristics of high precision,simple structure,convenient adjustment and expansion,and can realize the control of fuzzy neural network integrity under operation voltage of 1.8 V.