Parameter optimisation of support vector machine using mutant particle swarm optimisation for diagnosis of metal-oxide surge arrester conditions

This paper proposes an enhanced support vector machine (SVM), whose parameters are optimised by a novel mutant particle swarm optimisation (mutant PSO) algorithm to identify metal-oxide surge arrester conditions. The total leakage current and its resistive component under different arrester conditions are obtained and then are inputted into a multilayer SVM for the purpose of fault identification. Then, a mutant PSO-based technique is investigated to increase the classification accuracy as well as the training speed of the SVM classifier. The proposed technique has been tested on an actual data set obtained from Taipower Company to monitor five arrester operating conditions, including normal (N), pre-fault (A), tracking (T), abnormal (U) and degradation (D). Furthermore, to demonstrate the effectiveness of the proposed mutant PSO, the obtained results are compared to those obtained by using cross-validation method, genetic algorithm and particle swarm optimisation.