The neural network method of detecting weak harmonic signal embedded in chaos
Yang Bao-jun · Journal of China Institute of Communications · 2004
This paper presents a neural network method to detect weak harmonic signal embedded in chaotic 憂oise? Based on embedding theorem, the method utilizes the observed values of single variable of chaotic system to reconstruct phase space. The radial basis function neural network(RBFNN)is used to build one-step predictive model which has the same underlying dynamics as the chaotic background. Combining with a comb-filter, predictive error is processed so that weak harmonic signal is extracted from strong chaotic noise. The method can detect weak harmonic signal when the signal-noise-radio (SNR) gets-46dB.