Research Of Transformer Faults Diagnosis Based On Multi-Agent

In electrical system power transformer is the critical equipment. Its performance directly affects the safe operation and dependability of electrical system. It is very important to realize transformer running condition and find incipient faults as early as possible for transformer condition maintenance.With the evolution of domestic power industry, the numbers of the power transformer surge. the data of transformer more and more huge, power transformer fault diagnosis is more and more complex.The analysis to dissolved gases in a transformer is useful to the diagnosis of the transformer faults. Due to the shortcomings of the randomness and uncertainty of power transformer fault diagnosis data , the benefits of the Bayesian network classifiers and the features of multi-Agent, this paper introduces a multi-Agent system diagnosis model. This model include three diagnosis-Agents, a management-Agent and fusion-Agent .In this model the Diagnosis-Agent are established according to the obtained gas and based on NB,SB and TAN, which are three different kinds of Bayesian classifier algorithm . Those diagnosis-Agents are regulated and controlled by the Management-Agent, which purpose to achieve the fault diagnosis in consultation under all diagnosis-Agents. The fusion-Agent will give the final transformer fault types under the control situation which is from the Management-Agent's message, diagnosis result which should be from all the Diagnosis-Agents'diagnostic result, different probability of each diagnosis-Agent and numbers of different diagnosis-Agent runs .Finally, the result of practical sample verifies the accuracy of the proposed idea.