A Reconstruction Method for Missing Data of Electricity Users Using Extremely Randomized Tree

Data assets of electricity users are very important to the production, operation and maintenance of power grid, but data loss may occur during the collection, communication, and storage of power data. Therefore, this paper proposes a cyclic regression reconstruction method based on extremely randomized tree for missing data of electricity users. This method transforms the problem of missing data reconstruction into a regression problem, and introduces a regression reconstruction model based on extremely randomized tree, which overcomes the shortcomings of slow training and poor generalization ability of the deep forest method. Moreover, missing samples are sequentially reconstructed in a specific order, and the error stacking problem of sequential regression is solved through cyclic regression reconstruction. Finally, simulation verification is carried out by taking the actual measurement data set of a certain electricity users in China as an example. The results show that the proposed method has higher reconstruction accuracy and shorter simulation time than existing reconstruction methods, meaning that it is easy to popularize in engineering applications.