A New Approach for Power-Saving Analysis in Consumer Side Based on Big Data Mining

With the development of smart grid, power companies have accumulated massive amounts of data including running data of electricity distribution and multi-dimensional data about meteorology, users' file and social economy. In this article, we propose a power-saving analysis method based on big data mining. Firstly, all users belonging to the same industry should be extracted, then high-dimension data would be clustered based on the indexes describing the characteristic of electricity consumption. Secondly, the benchmark user will be selected based on the comprehensive assessment. So each user's power-saving potential can be calculated compared with the benchmark user, then the power-saving sensitivity coefficients can be calculated by using partial least squares regression, which leads to effective energy-saving strategies. Finally, the electricity saving prediction model is established by the XGBoost algorithm. The analysis process is performed with field data and a power-saving analysis platform is applied in Shanghai demonstrating the method's effectiveness.