Determination of partial discharge development stage of oil-paper insulation based on sparse decomposition considering the effect of aging

Abstract

To realize the accurate determination of the partial discharge development stage of an oil-immersed power transformer by a method based on sparse decomposition is proposed in the insulation system of an oil-immersed power transformer. First, statistical parameters are extracted from the PD pulse signals to construct a complete atomic library. According to the sparse representation principle, the preliminary determination of the discharge stage can be realized by ignoring the influence of aging; secondly, to solve the influence of feature parameter correlation on the determination result, the statistical feature parameters are sparsely reconstructed to realize the ordering of the effectiveness of the statistical feature parameters; lastly, to take into account the influence of the aging of the insulating cardboard, the plausible weights of the aging factor are calculated, and the number of votes for the development stage of partial discharge is determined according to the Borda voting mechanism. Two typical discharge defect models are designed so that the PDs on which the work focuses are superficial, and the measured signals are used to verify the validity of this paper's method. The results show that the highest accuracy is 56.2% when ignoring the influence of insulation cardboard aging, and the accuracy of the decision is less than 75% when considering the influence of aging without sparse reconstruction of statistical feature parameters; the method in this paper has good recognition effect, and the recognition accuracy is improved by 38.6% compared with that of ignoring the influence of aging of the insulation cardboard, and the average can reach 94.2%.