Correlation Analysis between Temperature and Electricity Consumption Behavior Based on Random Matrix Theory

Electric power companies have accumulated a large number of statistical data on load characteristics, and the quality and quantity of data have been continuously improved. In this situation, how to effectively extract the correlation characteristics between data and conduct effective correlation analysis of big data is of great significance. This paper proposes a correlation analysis method based on the random matrix theory to study data statistical correlation between temperature and electricity consumption behavior. Firstly, the augmented matrix is served as the data source using load data and temperature data. Then, based on the real-time split window analysis method and the random matrix theory, the statistical characteristics of the split window matrix elements are analyzed. Finally, a quarter time span is applied to compare the influence of temperature on the behavior of electricity consumption. The linear eigenvalue statistics, taking the mean spectral radius as an example, and the pearson correlation coefficient are employed as correlation quantification indicators. According to the actual measurement load data of a certain city power grid, the adaptability and effectiveness of the method are verified, and the approach has a good robust performance.