Analysis of correlation between meteorological factors and short-term load forecasting based on machine learning

Abstract

The power system load is affected by various external factors, making the short-term load have the characteristics of uncertainty and randomness. There are many factors affecting power system load forecasting, and weather conditions have the most significant impact on load forecasting. Based on the existing literature, this paper proposes a method to analyze the correlation between single meteorological factors and system load, and then comprehensively consider the impact of all meteorological factors on system load. Considering weather, rainfall, humidity and other meteorological factors. The BP algorithm has a very strong nonlinear fitting ability and can theoretically fit any complex nonlinear mapping relationship. This paper uses Python to implement the BP algorithm considering multiple meteorological factors, and predicts the load of October in a certain area of South Network. The prediction results show that compared with the traditional input of all meteorological factors as BP model, and the correlation between meteorological factors and system analysis is not analyzed, the processing method of this paper improves the accuracy of load forecasting and accelerates the algorithm. The convergence speed and learning time are reduced, which greatly improves the efficiency and has a certain guiding effect on the application of load forecasting in the actual power grid.