Transformer Fault Prediction Based on GA and Variable Weight Gray Verhulst Model

Abstract

A new prediction method combined variable weight Gray Verhulst model and gray integrated relation grade was proposed in this paper to solve the problem of power transformer fault prediction. Because power transformer gases concentration sequence was S-shaped, Gray Verhulst model was chosen to forecast the gases concentrations. Variable weight Gray Verhulst model was proposed based on 2 improved Gray Verhulst models with 2 difference select rules of parameter p in background function. Genetic algorithm chose parameter p for variable weight Gray Verhulst model. Powers transformer fault diagnosis using gray integrated relation grade had 93.7% diagnostic accuracy. Experiments on power transformer fault prediction show that Variable weight Gray Verhulst model had higher prediction accuracy, and, the fault prediction method proposed in this paper has the same forecasting result with the true values and is reliable and effective.