Line Loss Anomaly Identification of Low-Voltage- Station Based on Second-Order Clustering and Robust Random Cut Forest Algorithm

To accurately identify the abnormal line loss in the substation area and ensure the economic and stable operation of the distribution network, in allusion to the abnormal line loss in the substation area, based on second-order clustering and robust random cut forest (abbr. RRCF) algorithm a method to detect the abnormal line loss in the substation area was proposed. Firstly, by means of second order clustering the different operating conditions of the substation area were clustered and the line loss nodes under the same operating conditions were merged. Secondly, the nodal line loss data of all kinds of operating conditions was led into RRCF algorithm to conduct the analysis. By means of deleting and inserting sample nodes and computing the complexity of the evaluation model after inserting nodes, the score values of abnormal line loss nodes could be obtained, and further the nodes with abnormal line loss could be found out. Finally, the effectiveness and accuracy of the proposed method are verified by related examples.