Fault location based on FastICA and fuzzy C-means clustering for single-phase-to-ground fault in the compensated distribution network

A new method for fault location during a single-phase-to-ground fault in the compensated distribution network is proposed in this paper. The method makes full use of the difference between transient-state fault information and the difference between steady-state fault information in two FTUs (feeder terminal units) of each section to determine faulty location. In this method, the transient-state component and steady-state component of zero-sequence current measured by each FTU are extracted by FastICA (fast independent component analysis) algorithm and are represented by the transient-state factor and steady-state factor, which are defined in this paper. Then, the difference between transient-state factors and the difference between steady-state factors in two FTUs of each section are calculated. The two differences corresponding to a certain section are seen as a feature vector, which is given as one input to the fuzzy C-means (FCM) algorithm. All feature vectors will be partitioned into two clusters, one of which represents a faulty section, and the other of which represents a healthy section. As a result, faulty location will be determined. Simulation results show the effectiveness and reliability of the proposed method under different fault conditions.