Fault Diagnosis of Power Transformers Based on BP Network with Clonal Selection Algorithm

In this paper, a novel approach based on BP network (BPN) for fault diagnosis of power transformers is proposed. Optimization of BPN weights is achieved by using clonal selection algorithms (CSA). In addition, the mutation probability is adjusted adaptively according to the affinity of antibody. Compared with previous approaches, this one can avoid prematurity effectively, with good self-learning and self-memory ability. The experiment results show that the presented approach outperforms previous ones in both classification accuracy and computational efficiency.