Study on weak signal detection in chaotic background based on prediction of GRNN

摘要

As a new radial basis neural network, GRNN is easier to train and has better ability of function approach. In this paper, based on the characteristics of GRNN, signals in various chaotic and noisy background are predicted. Simulation results prove that GRNN has the characteristics of fast learning speed, simple design and stable structure, and the trained GRNN can implement prediction and recomposition of chaotic time series because its better ability of function approach and higher predicting precision.