Abstract
Currently, ensuring that power systems operate efficiently in stable and secure conditions has become a key challenge worldwide. Various unwanted events including injections and faults, especially within the generation and transmission domains are major causes of these instability menaces. The earlier operators can identify and accurately diagnose these unwanted events, the faster they can react and execute timely corrective measures to prevent large-scale blackouts and avoidable loss to lives and equipment. This paper presents a hybrid classification technique using support vector machine (SVM) with the evolutionary genetic algorithm (GA) model to detect and classify power system unwanted events in an accurate yet straightforward manner. In the proposed classification approach, the features of two large dimensional synchrophasor datasets are initially reduced using principal component analysis before they are weighted in their relevance and the dominant weights are heuristically identified using the genetic algorithm to boost classification results. Consequently, the weighted and dominant selected features by the GA are utilized to train the modelled linear SVM and radial basis function kernel SVM in classifying unwanted events. The performance of the proposed GA-SVM model was evaluated and compared with other models using key classification metrics. The high classification results from the proposed model validates the proposed method. The experimental results indicate that the proposed model can achieve an overall improvement in the classification rate of unwanted events in power systems and it showed that the application of the GA as the feature weighting tool offers significant improvement on classification performances.