Skip to Main Content
A combination approach of wavelet transformation and neural network is applied to realize transient power quality disturbance signals' recognition. Firstly, the mathematical models of five kinds of transient disturbance signals, such as voltage surging, voltage sag, voltage interruption, transient pulse and transient oscillation are founded. Then, using time-frequency characteristic of wavelet, the sample signal's feature vectors are extracted. At last these feature vectors are input into BP neural network. Using self-learning ability, the disturbance signals can be classified and recognized. The examples show that the method has a higher discrimination. It's effective to resolve transient power quality problem.