Study On Prediction Models Of Power Transformer Fault Based On Genetic Algorithm

Power transformers are key elements in power system. The operation reliability of power transformer is related to the safety and stability of power system directly. Fault of one power transformer may cause long interruptions in supply, costly repairs and loss revenue. It's important that forceasting the faults in power transformer.In this paper, the power transformer interior fault diagnosis technique based on the dissolved gas in oil analysis and the principles of genetic algorithm are analyzed. The forecasting models for power transformer interior fault are proposed based on the grey prediction model. The genetic algorithm is applied to estimating optimum coefficients of this forecasting model. The research achievements are as followed:Studying the shortages and the improved methods of GM(1,1) grey prediction model, considering the characteristic of the transformer chromatographic data, bring forward the method for converting a series of data which are sampled in different interval into a series of data in the same interval. The weakening operator is applied to reconstruct the transformer chromatographic data for attenuating or eliminating the influence of randomicity. The improved prediction model for power transformer interior fault is constructeded. The applicability of this model is wider than GM(1,1) grey prediction model.The forecasting model based on genetic algorithm for power transformer interior fault is proposed. Programs for modeling and forecasting based on MATLAB are programed in this paper. It is simple and effective. The results of prediction examples show that this forecasting model is effective for forecasting the transformer interior fault.The sources and the classes of error of this prediction model are analyzed. Some effective methods for verifying the grade of this prediction model adequacy are presented. The results of examples in this paper show that the model is more precise than the other prediction models in this paper because the genetic algorithm is applied to estimating optimum coefficients of the model.The results of examples in this paper show that the genetic algorithm is effective for achieving optimum coefficients of this model. The prediction model is effective for forecasting the tendency of gas dissolved in power transformer oil. It is feasible and the forecasting precision of this model is remarkable.