Design of FIR Digital Filter Using CFNN

摘要

Aimed at the problem of designing FIR digital filter which is approximating ideal complex frequency response, this paper proposes a neural network (NN) method of designing FIR digital filter based on energy function optimum theory. The method constructs a complex feedforward neural network (CFNN) according to the complex frequency response FIR digital filter, presents the Lyapunov function by using error energy function (EEF) representing the designing performance, and proves the convergence of the presented network algorithm. By exemplifying the design of complex non-linear phase FIR digital filter, it shows that the algorithm proposed in this paper is effective and meets the good generalized ability.