CVT Measurement Error Range Recognition based on the FCM Clustering Algorithm

Traditional capacitive mutual inductor (CVT) detection methods require error comparison between standard mutual inductor and evaluated mutual inductor in case of power outage. Due to power outage and repeated disassembly and installation of mutual inductors, this method of detection is cumbersome and inefficient. For this reason, a CVT measurement error range identification method based on FCM clustering algorithm is presented, that is, under the condition of power outage, the error range level of CVT itself is identified by clustering analysis of the secondary-side data of CVT to lock the sensor that needs power outage inspection as soon as possible. First, the secondary-side CVT data is segmented by rolling time window method, then the standard secondary-side CVT data in rolling time window is processed by FCM clustering of two cluster centers, and the average distances between the two cluster centers are obtained. Finally, the FCM clustering of two cluster centers for the second-sided CVT data in the rolling time window is performed to get the standard ratio of the distance between the cluster centers in each time window to identify the error range level. On this basis, the CVT with higher error level is detected for power outage. Experiment shows that this method can effectively identify the error range rank of CVT, thereby significantly improving its detection efficiency.