Abstract
Power transformers are one of the most critical devices in power systems. It is responsible for voltage conversion, power distribution and transmission, and provides power services. Therefore, the normal operation of the transformer is an important guarantee for the safe, reliable, high quality and economical operation of the power system. It is necessary to minimize and reduce the occurrence of transformer failure and accident. The on-line monitoring and fault diagnosis of power equipment is not only the prerequisite for realizing the predictive maintenance of equipment, but also the key to ensure the safe operation of equipment. Although the analysis of dissolved gas in transformer oil is an important means of transformer insulation monitoring, the coexistence of two kinds of faults, such as discharge and overheat, can lead to a lower positive rate of diagnosis. In this paper, we use the basic particle swarm optimization algorithm to optimize the BP neural network DGA method, select the typical oil in the oil as a neural network input, and then use the trained particle swarm algorithm to optimize the neural network for transformer fault type diagnosis. The results show that the method has a good classification effect, which can solve the problem of difficult to distinguish the faults of the transformer when the discharge and overheat coexist. The positive rate of fault diagnosis is high.
