Transformer Fault Diagnosis Method Based on Information Fusion

Abstract

According to the characteristics and current situation of power transformer fault diagnosis , information fusion technology is introduced into this field of fault diagnosis of power transformer. By studying the general framework of information fusion fault diagnosis process, combined with the dissolved gas and electrical tests data,it is proposed a fault diagnosis method of information fusion which is based on fuzzy coding boundary and Bias regularization Levenberg Marquardt (LM) network. The algorithm uses a Bias approach to determine the hyper parameters, making the neural network adaptively adjust the parameter in the training process and getting the optimization parameters of the objective function . Combined with fuzzy coding boundary for feature attribute reduction to improve the accuracy of fault diagnosis .The contrast analysis of twice fusion results shows that the fault diagnosis model of the transformer accurate rate is 89.83%. Finally , it proves the validity and practicability of this new diagnosis method. DOI : http://dx.doi.org/10.11591/telkomnika.v12i5.4054