Based On The Nonlinear Traction Transformer Fault Diagnosis System With Lun Lm Algorithm

This paper clarified the fault diagnosis status of fault diagnosis on electric locomotive traction transformer,and significance of this issues Analysing the operating environment of electric locomotive traction transformer, after discussing a variety of fault diagnosis on power transformer, we choose a transformer Gasses dissolved in oilmethod as a Theoretical basis to diagnose internal fault of traction transformer. Then we introduced the principle of gasses dissolved in oil, studied the relationship among the gas production,rate and the type of traction transformer failure. Introducing the gasses dissolved in oil method applied to traction transformer three-ratio method fault diagnosis theory and defects.Based on the analysis of characteristics of neural network, combined with traction transformer fault gases and features of fault types, a LM optimization method combined with nonlinear algorithm ethics,improved method called nonlinear homotopy LM method was obtained. This method can improve the ability of BP network to avoid the local minimum points, and convergence speed and accuracy. The theoretical and simulation analysis of various defects improved BP algorithm shows that the problem which the method still easily fall into local minimum has not larger improvement on Shortcoming we mentioned above. This article use nonlinear homotopy LM method on traction transformer fault diagnosis, through network model training and fault verification indicate that the proposed non-linear homotopy LM algorithm for solving the problem of traction transformer fault diagnosis has good results, better than the usual BP algorithm and the traditional three-ratio method. Finally, we researched the implementation in traction transformer fault diagnosis system and use IPC to implement nonlinear homotopy LMmethod with the combination of MATLAB and VB.