The Neural Network Adaptive Filter Model Based on Wavelet Transform

Due to the problem that the noise in the noisy signal can not be predicted in many practical fields, we have proposed an adaptive filter based on wavelet transform method. As the adaptive filter has the characteristic of eliminating noise no use to predict the priori knowledge of the noise in the signal, we have taken the signal after the first wavelet threshold denoising as the main input of the adaptive filter, meanwhile taken the wavelet reconstruction coefficients after the second wavelet transform as the reference input of the adaptive filter. And a neural network adaptive filter model based on wavelet transform is constructed. The model has applied the Hopfield neural network to implement the adaptive filtering algorithm LMS, so as to improve the computation speed. The simulation results show that the neural network adaptive filter model based on wavelet transform can achieve the best denoising effect.