A new fault diagnosis method of vibrating of hearings was proposed on the basis of neural network based on wavelet packet-characteristic entropy(WP-CE).Firstly,three layers wavelet packet decomposition of the acquired vibrating signals of hearings was performed and the wavelet packet-characteristic entropy was extracted;then the eigenvector of wavelet packet of the vibrating signals was constructed,the three layers BP neural network were trained to implement the intelligent fault diagnosis by taking this eigenvector as fault sample.The simulation result from the proposed method is effective and feasible.