Design of FIR digital filter based on improved neural network

Abstract

This paper designs a FIR digital filter algorithm based on improved neural network. In the process of neural network learning, the algorithm can adjust the learning rate adaptively according to the error value between the frequency response of neural network output and the ideal frequency response, so that the algorithm can avoid the slow convergence or oscillation caused by fixed learning rate. In this paper, four FIR digital filter design examples are given. The simulation results show that the FIR filter design algorithm designed in this paper can achieve faster convergence speed and higher stopband attenuation than the neural network under fixed learning rate, which is an effective design method of FIR digital filter.