Abstract
In the context of the allocation of electricity big data, in order to solve the problem that the power consumption anomaly detection algorithm is less efficient due to the diverse sources, various types, and large data volume of the power consumption data, this paper proposes a power anomaly data detection algorithm based on user tag and random forest. By analyzing the power data, each data is marked with a power type, including a normal type, a less bur type, a large downward shift type, and a multi-burr type. Based on the tags, three types of user power data tags, such as basic information tags, environmental information tags, and power information tags, are constructed as attribute sets of user power data. Finally, a user data anomaly data analysis algorithm based on user tag and random forest is proposed. In the experimental part, the traditional the random forest algorithm is compared with the proposed algorithm, and the performance index of the algorithm in both accuracy and false alarm rate is better than the traditional random forest algorithm.