Research On Fault Diagnosis Of Power Transformer Based On Gas Analysis In Oil

Power transformers,as the key equipment of transmission and distribution systems,have excellent performance and will directly affect the power users.Therefore,it is of great significance to minimize the failure of power transformers.In the current research method based on dissolved gas analysis in oil,the nonlinear mapping from characteristic gas to fault type is mainly established.This method can diagnose the latent fault of the transformer in time and improve its operating efficiency through detection and analysis.Based on the characteristics of power transformer fault diagnosis and existing problems,this paper uses intelligent algorithm to realize the transformer fault diagnosis model construction and experimental simulation analysis based on dissolved gas analysis in oil.In this paper,the principle of gas generation in oil and its relationship with fault types are studied.Based on this,a fault diagnosis model based on extreme learning machine is established.The model vector,sample data,activation function and related parameters are selected.And the analysis,the experimental simulation of transformer fault diagnosis based on the algorithm model is realized.The results show that the algorithm has certain diagnostic effects,but the stability of the results is not good.Aiming at this,this paper introduces the idea of kernel function,and establishes a fault diagnosis method based on kernel limit learning machine.This method uses kernel function instead of hidden layer output matrix to avoid random generation of input weight and threshold.The problem of instability of the algorithm is solved,but it also leads to the problem that the diagnostic model is more sensitive to parameters.In view of this,based on the advantages of particle swarm optimization algorithm and artificial immune algorithm,the two algorithms are combined and used for parameter optimization of nuclear extreme learning machine,and the power transformer fault based on immune particle swarm optimization kernel limit learning machine is established.Diagnostic model,which has certain innovation in fault diagnosis of transformers.Finally,this paper uses the MATLAB platform to complete the experimental simulation based on the immune particle swarm optimization kernel limit learning machine diagnosis model,and compares the simulation results with the extreme learning machine and other diagnostic algorithms.The results show that the optimized limit is optimized.The learning machine diagnosis algorithm has a great improvement in output accuracy and time,and has a good diagnostic effect.