A short‐term electricity consumption forecasting approach based on feature processing and hybrid modelling. Issue 10 (26th February 2022)

• Record Type: Journal Article
• Authors: Wei, Minjie; Wen, Mi; Luo, Junran
• Is Part Of: IET generation, transmission & distribution. Volume 16:Issue 10(2022)
• Page Start: 2003
• Page End: 2015
• Publication Date: 2022-02-26
• Subjects: Electric power production -- Periodicals; Electric power transmission -- Periodicals; Electric power distribution -- Periodicals; 621.3105
• DOI: 10.1049/gtd2.12409
• Languages: English
• ISSNs: 1751-8687

Abstract

Abstract: The short‐term electricity consumption forecasting can help to ensure the safe and reliable operation of the power system. Power companies usually need to report the electricity consumption of the current month five to seven days in advance and make a power generation plan for the next month. The existing studies are usually lack of appropriate feature selection methods and hard to achieve satisfactory results. This paper proposes a short‐term electricity consumption forecasting approach based on feature processing and hybrid modelling. The maximum information coefficient (MIC) is employed to analyse the feature correlation, the electricity consumption curves are converted to several sub‐sequences of different frequency bands by the variational mode decomposition (VMD) to describe signal characteristics accurately, a hybrid model based on bidirectional gated recurrent unit (BiGRU) is innovated to extract the temporal and spatial features of the data and capture the contextual information from the complete time series, attention mechanism is used to do extract useful information and assign weights to make forecast. Compared with several benchmark methods, the proposed approach achieves better electricity consumption curve fitting and higher forecasting accuracy with the increase of forecasting step size.