Medium-long term load forecasting method considering industry correlation for power management

Abstract

Medium-long term load forecasting is recognized as the basis of economic dispatch in power systems, and high-precision load forecasting results can provide effective support for power management. Therefore, a medium-long term load forecasting method based on the deep learning algorithm is proposed with industry correlation considered. First, the missing monthly load data is supplemented based on the iterative interpolation method, and min–max normalization method is used for data preprocessing. On this basis, the Pearson correlation coefficient is adopted for quantifying the industry correlation and selecting the strongly related industries for load forecasting. Then, a machine learning model long-short term memory (LSTM) is utilized to forecast the medium-long term industry load with the historical data of related industries. Finally, the feasibility and accuracy of the proposed method are illustrated by the case studies on 2 major industries and 30 related industries of Jiangsu Province, China. The load forecasting error shown in the simulation results of the method proposed is kept within 5%, which is much better than other traditional methods.

Keywords

• Industry correlation
• Medium-long term load forecasting
• Long-short term memory (LSTM)
• Pearson correlation coefficient