This paper proposes a new method for ground fault protection of distribution networks based on fuzzy measure fusion criterion. The fuzzy c-means clustering algorithm is used to characterize the historical characteristics of different operating states of protected feeders under various operating conditions. The sample set is divided into fault class and non-fault class. The sample similarity measure criterion is used to quantitatively analyze the similarity degree between the sample to be tested and different types of historical samples, forming a fuzzy measure fusion criterion matrix, and fuzzy by multi-level evaluation index system. The measurement fusion criterion matrix is used to judge, and finally a comprehensive criterion matrix is output as the protection criterion. The simulation results of PSCAD/EMTDC show that under the influence of non-linear load and other disturbance factors, the method can still accurately identify fault samples and is not affected by changes in system operation mode, and has strong robustness.

Distribution network, ground fault, fusion criterion, fault measure, multi-level evaluation system