The classification of transient phenomena in the electrical power systems by using wavelet analysis and probabilistic neural networks

An investigation into the characterization and classification of power system transients, using advanced signal processing and pattern classification techniques has been realized. Automation of power system fault identification using information conveyed by the wavelet analysis of power system transients is proposed. In the system developed, which is intended to act as an artificial consultant to power systems operators, the implementation of wavelet analysis was for characterizing transients in power systems and to extract features from them. The Daubechies wavelet family used in this thesis decomposes the signal into details and approximations, which contain the high and low frequency content of the signal, respectively. The usefulness of this method in characterizing the transients as well as their combination is evaluated. As a classification method, the probabilistic neural network (PNN) has been used to identify the corresponding class of a transient. Because the wavelet detail coefficient for each type of simple fault is characteristic in nature, these coefficients are used in PNN for distinguishing between the waveforms and hence the faults. The performance of the system is evaluated on simulated transients in Matlab/Simulink Environement. For the simulated data, the PNN yielded an average accuracy of 99.4% with the training set and 96.7% with the testing set of data. The results show superior performance, both in the accuracy of the classification and selection of the waveform features used. Key Words: Power system transients, wavelet analysis, fault classification, probabilistic neural netwoks, fault recognition