Implementation of CMAC neural network control systems in residue number system

Abstract

The method and algorithm of cerebellar model arithmetic computer (CMAC) training of the many variables modular function is offered. The submitted method allows using CMAC neural network to implement proportional-integral-derivative controllers, functioning in residue number system. The hardware realization of CMAC on the basis of lookup tables is considered. The simulation results demonstrate the effectiveness of the proposed CMAC neural network controller in residue number system. Application of the proposed CMAC neural network allows reducing of hardware costs while maintaining high reliability and performance of controller