A neural network used for PD pattern recognition in large turbine generators with genetic algorithm

The paper introduces a new neural network with a genetic algorithm for use in on-line partial discharge (PD) monitoring of large turbine generators. The disadvantages of the BP algorithm which is generally used for PD pattern recognition are discussed. As a result of these shortcomings, a new neural network is presented using a genetic algorithm. To ensure the suitability of the chosen network for PD recognition, the PD patterns which generally occur in large generators were simulated and tested. Test results show that the new neural network can meet the need of PD pattern recognition in large generators satisfactory.