Abnormal Line Loss Data Detection and Correction Method

To reduce the energy loss caused by the line loss, it is necessary to accurately calculate the line loss rate. In the actual calculation process, the accurate calculation of the actual line loss is often difficult to achieve. One of the main problems are due to various reasons, the original data will inevitably accrue data loss and data confusion. To improve the calculation accuracy of the line loss rate, the paper proposes an abnormal line loss data detection and correction method. Aiming at the problem of abnormal power grid data collection, the paper first extracts abnormal power consumption data from historical data, uses AP clustering algorithm to classify abnormal data, extracts the abnormal power consumption behavior characteristics for each type of historical abnormal power consumption curve, and then compares real-time power consumption data with abnormal electricity behavior features, and determines whether abnormal data do occur according to the similarity theory. Finally, DNN algorithm is used to replace the abnormal data. Based on the model proposed, calculations show that the use of the DNN algorithm for data correction not only has a higher accuracy rate but also can better improve the accuracy of the calculation of the line loss rate of the distribution network compared to BP prediction and other regression algorithms.