Grid Fault Diagnosis Based on Information Entropy and Multi-source Information Fusion

Abstract

In order to solve the problem of misjudgment caused by the traditional power grid fault diagnosis methods, a new fusion diagnosis method is proposed based on the theory of multi-source information fusion. In this method, the fault degree of the power element is deduced by using the Bayesian network. Then, the time-domain singular spectrum entropy, frequency-domain power spectrum entropy and wavelet packet energy spectrum entropy of the electrical signals of each circuit after the failure are extracted, and these three characteristic quantities are taken as the fault support degree of the power components. Finally, the four fault degrees are normalized and classified as four evidence bodies in the D-S evidence theory for multi-feature fusion, which reduces the uncertainty brought by a single feature body. Simulation results show that the proposed method can obtain more reliable diagnosis results compared with the traditional methods.