Work

Extreme weather load forecasting method for power systems based on extreme gradient lifting algorithm

Year: 2026

Type: article

Abstract: In order to reduce power load forecasting errors and ensure stable operation of power systems, an extreme weather load forecasting method based on the extreme gradient boosting algorithm is proposed. We analyze the composition of power system load. Extreme weather factors are considered the main factors affecting the load characteristics of the power system from the perspective of weather factors, and the correlation between meteorological factors and power load is analyzed. The Box–Jenkins method is used to construct the power system load time series based on the analysis results. The extreme gradient boosting algorithm is employed for load prediction by analyzing load time-series data along with meteorological factors, thereby enhancing the precision of load forecasting. The experimental results demonstrate that the proposed method achieves a maximum average absolute error of just 1.54 and a root mean square error of only 3.31. The difference between the predicted values and the actual load values is minimal, with the largest deviation being a mere 0.7 MW. Moreover, under extreme weather scenarios, the relative error of the predictions remains below 5.0%. These results comprehensively attest to the excellent predictive capability of the proposed approach.

Source: AIP Advances

DOI: 10.1063/5.0310755

Authors: Guang Yang, Wenqin Peng, Hong Chang, Zhiquan Tan, Xueying Yu

Institution: China Southern Power Grid (China)

Language: English

Topic: Energy Load and Power Forecasting

Subfield: Electrical and Electronic Engineering

Field: Engineering

Domain: Physical Sciences

SDGs: Climate action, Affordable and clean energy, Industry, innovation and infrastructure

Open Access status: gold

Estimated APC paid: $1,350