An improved neural network algorithm for classifying the transmission line faults

This study introduces a new concept of artificial intelligence based algorithm for classifying the faults in power system networks. This classification identifies the exact type and zone of the fault. The algorithm is based on unique type of neural network specially developed to deal with a large set of highly dimensional input data. An improvement of the algorithm is proposed by implementing various steps of input signal preprocessing, through the selection of parameters for analog filtering, and values for the data window and sampling frequency. In addition, an advanced technique for classification of the test patterns is discussed and the main advantages compared to previously used nearest neighbor classifier are shown.