Work

A novel model averaging forecasting method for electricity consumption using electricity business expansion data

Year: 2025

Type: article

Abstract: This paper introduces a novel model averaging method for electricity consumption forecasting that incorporates electricity business expansion application data. The proposed framework uses a two-step regression approach: first, electricity consumption is decomposed into long-term deterministic trends, temperature-related components, and short-term economic activity components. Second, a model averaging forecast is constructed for the short-term component using 15 forecasting variables derived from expansion application data, along with short-term GDP growth rates, electricity prices, and other relevant factors. Using data from Fujian Province, we compare the forecasting performance of models with and without expansion application data, employing both model selection and AIC-based model averaging techniques. Empirical results show that incorporating expansion application data improves forecasting accuracy by up to 14% in out-of-sample predictions. These findings underscore the value of expansion application data as a resource for enhancing electricity consumption forecasting and supporting more effective energy management strategies.

Source: Energy Reports

DOI: 10.1016/j.egyr.2025.03.024

Authors: Guoyao Wu, Zhiqiang Lan, Kun Zhou, Shiqing Huang

Institutions: State Grid Corporation of China (China), Xiamen University

Language: English

Topic: Energy Load and Power Forecasting

Subfield: Electrical and Electronic Engineering

Field: Engineering

Domain: Physical Sciences

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

Open Access status: gold

Estimated APC paid: $3,040