Research On Transformer Fault Diagnosis Based On Online Sequential Extreme Learning Machine

The dissolved gas-in-oil analysis is one of the most commonly used method for fault diagosis of oil inunersed power transformers In this text,the online sequential extreme learning machine algorithm was applied in the fault diagnosis of transformer based on the analyses of current fault diagnosis methods for transformer, And combined partial discharge detector for discharging signal using D-S evidence theory fusion, the different diagnosis results.Analyzed the characteristics and shortcomings of the traditional methods and intelligent method in the fault diagnosis of transformer summarized,and general idea and implementation procedure of the extreme learning machine and D-S evidence theory.In view of the octual environment,transformer condition monitoring has achieved online monitoring online monitoring in some ways, Provide a method for transformer fault diagnosis based on online sequential extreme learninmachine. The affection of bidden layer activation function on diagnostic performance was investigated, and the implementation procedure of fault diagnosis was provided in. detail Experiments show that, comparing toSVM and ELM, the OS-ELM transformer fault diagnosis has more accuracy and less training time,and it has great application foreground.Be directed against the instability of the OS-ELM network outpuT, used ensemble pf online sequential extreme learning machine to optimlze it ,Gave the description of the EOS-ELM algorithm. Experiment show that,compoaring to OS-ELM the EOS-ELM algorithm is more stable.Only rely on a single method is very difficult to get the accurate diagnosis results in the oil immersed power transformer fanlt diagnosis Provided a synthesis diagnostic method based on OS-ELM and D-S evidence theory is aimed at the huge damage brought by discharge fault, and established the discharge fault diagnosis model. Through the analysis of examples,this model can improve the reliability of in diagnosed (he electro-discharge fault.