A New Method for Oil-Immersed Transformers Fault Diagnosis Based on Evidential Reasoning Rule With Optimized Probabilistic Distributed

Abstract

Power transformers are important equipment in power systems. Most medium and large transformers are oil-immersed transformers, with the exception of some small-and medium-capacity transformers and gas transformers with special purpose. It is extremely critical to detect malfunction timely and accurately, because the reliability and safety of oil-immersed transformers can cause significant implications for power systems. Therefore, the fault diagnosis model based on the evidential reasoning (ER) rule with reference points in the Gaussian distribution form optimized by a constrained genetic optimization algorithm (GA), named GER-G model, is proposed in this paper. The GER-G fault diagnosis model can monitor and identify the degree of transformer faults in real time. First, the concentration of dissolved gases in the oil of oil-immersed transformers varies at different fault degrees, so the gases with high weight are selected using the maximum correlation-minimum redundancy (mRMR) to constitute the indicator system. Second, to solve the uncertainty existing in the fault information of transformers, the reference points of the ER rule are continuous probability distribution reference points described by Gaussian distribution form. Third, the constrained GA is proposed to adapt the Gaussian distribution. And the accuracy of fault diagnosis results can be enhanced by optimizing the grade parameters using constrained GA. Finally, the internal and external influences in the environment are quantified using perturbation analysis, the GER-G fault diagnosis model is applied to the dataset of dissolved gases in the oil, and the robustness and validity of the GER-G fault diagnosis model are validated. Meanwhile, the validation results also show that the GER-G fault diagnosis model possesses higher accuracy compared with other fault diagnosis methods.