Identification of Abnormal Users based on Isolated Forests during Electricity Price Implementation

In the process of electricity price implementation, users' abnormal electricity consumption will seriously lead to income decrease of power company. Therefore, it is necessary to choose a method with excellent performance to intelligently identify abnormal users in massive electricity consumption data. For this purpose, principle of isolated forest algorithm as well as its modeling steps are firstly analyzed. Then, based on data from electric energy data acquisition system, attributions used in screening abnormal users are extracted, and abnormal user identification model based on isolated forest is constructed, some index used in evaluating identification effect of the model is also given. Lastly, with data from field system, the proposed model is verified to be valid. And compared with other approaches, isolated forest algorithm is with higher identification accuracy.