Evaluation studies of combined wavelet and neural network applications in high voltage transmission line protection

This paper presents a critical evaluation of the combination of the wavelet transform and neural network techniques into high voltage transmission line protection. These techniques are used to enhance the ability to discriminate between different types of fault at any point of the protected line, relying on only single-ended measurement of the induced transient information of the fault. The main objective is to locate the faults and discriminate the fault type; this has been done by analysing the three phase voltage and current signals measured with CTs and VTs. For signal analysis, the wavelet transform has been used due to the ability to decompose a signal from low frequency to high frequency without losing the information in the time domain. Neural networks have been introduced as a powerful tool for accurate pattern classification. A Modular Neural Network (MNN) has been used to optimize the structure of an artificial neural network for a short training time and hardware implementation.