Base The Improve Of PGSA-BP Neural Network In Power Transformer Fault Diagnosis Simulation Studing

The paper describes the mechanism and characteristics of the power transformer fault diagnosis, artificial intelligence technology in the transformer fault diagnosis, and domestic and international development overview, analysis of the non-linear relationship between the causes of phenomena and mechanism of the power transformer fault proposed PGSA transformer fault diagnosis method combined with neural networks.From the PGSA (Plant Growth Simulation Algorithm) itself algorithm principle characteristics and shortcomings of view, the introduction of cultural algorithm to improve cultural algorithm to improve the structure of the PGSA diagram and model design, after improvement PGSA neural network transformers the type of fault diagnosis provides theoretical support. Improved PGSA access to and knowledge of the search space related to the accumulation in the search process, and automatically control the search process to obtain the global optimal solution, and finally combined with neural networks applied to transformer fault diagnosis research. Text for the fault diagnosis model to establish the5-16-5structure of BP neural network fault diagnosis model to call MATLAB7.2neural network toolbox and experimental simulation. Experimental results show that the improvements the PGSA optimize BP neural network for transformer fault diagnosis can improve diagnostic speed and accuracy to overcome the lack of training time and accuracy of the traditional neural network, the method to improve the performance of the typical transformer fault diagnosis and effective online tracking direct discovery potential failure of the transformer has a good application value.In order to meet the requirements of the power transformer fault diagnosis, in addition, around the transformer electrical characteristics of the method of transformer winding internal fault diagnostic studies to establish the mathematical model of the transformer winding identification relationship, and establish the simulation circuit model experimental simulation in the Simulink. The results show that, the rate of change of the parameters before and after the use of a winding fault can probably figure out the winding whether the occurrence of faults and fault phase, fault diagnosis comprehensive practicality.