Dynamic feature extraction of power disturbance signal based on time-frequency technology

The growing concern for power quality issues from both utilities and power users is generated by proliferation of power electronic devices and nonlinear loads in power system network. Therefore, the techniques for power quality monitoring and power disturbance mitigation are capturing increasing attention. A novel approach for the power quality disturbances recognition using wavelet transform and neural network is proposed. The wavelet transform is used to complete feature extraction and can accurately localizes the characteristics of transient signal both in time and frequency domains. These feature vectors are input variables for neural network training and the neural network structure is designed for disturbance pattern recognition. Therefore, the wavelet network combines advantages of wavelet transformation for purposes of feature extraction and selection with the characteristic decision capabilities of neural network approaches. During the training process, the wavelet network learns adequate decision functions and arbitrarily complex decision regions defined by the weight coefficients. The simulation results demonstrate the proposed method gives a new way for signal analysis and pattern recognition of power quality disturbances.