A load identification algorithm based on SVM

Load identification technique can identify the different types of loads in a power system. This paper presents a load identification algorithm based on support vector machine (SVM), which adopts the one-against-one method which combines multiple SVMs to build a multi-classifier to deal with load identification because SVM is only a binary-classifier. In this paper, the SVM-based multi-classifier is trained by extracted characteristic quantities normalized, during which the genetic algorithm (GA) is used to optimize the SVM parameters for the highest recognition accuracy. The experimental results show the validity of the SVM-based multi-classifier.