An Optimum Design of Linear Phase FIR Filters by the Generalized Brain-State-in-a-Box Neural Network Model

摘要

In this paper, we used a neural network model to find, an optimum design for a linear phase FIR digital filters coefficients. We used GBSB neural network model. The GBSB can guarantee convergence to an equilibrium point of the Lyapunov energy function as well as the fast computational speed. We utilized WLS error function to formulate an energy function of GBSB neural network. When the dynamic equation achieves its stable configuration, the output of the GBSB yields the optimal filter coefficients.