The feature extraction of plant electrical signal based on wavelet packet and neural network

To solve the difficulties of feature extraction of plant electrical signals and to realize effectively the classification of plant electrical signals, a method of plant electrical signal recognition which is combined with wavelet packet decomposition and the BP neural network was put forward in this paper. The method first decomposes wavelet packet of the plant signals, and puts the maximum of the eigenvalue of signal covariance matrix. The mean absolute value and zero-crossing rate as the eigenvalues, and then put the constructing feature set into the BP neural network, realizing different plant signal identification. The experimental results show that the optimal wavelet packet decomposition reduces the size of the BP neural network, and reduces the complexity algorithm, speeds up the network training time and classification speed. Therefore, this method has good recognition effect.