A developed hybrid forecasting system for energy consumption structure forecasting based on fuzzy time series and information granularity. (15th March 2021)

• Record Type: Journal Article
• Authors: Jiang, Ping; Yang, Hufang; Li, Hongmin; Wang, Ying
• Is Part Of: Energy. Volume 219(2021)
• Publication Date: 2021-03-15
• Subjects: Artificial intelligence -- Fuzzy time series -- Energy consumption structure -- Information granularity -- Improved optimization algorithm; Power resources -- Periodicals; Power (Mechanics) -- Periodicals; Energy consumption -- Periodicals; 333.7905
• DOI: 10.1016/j.energy.2020.119599
• Languages: English
• ISSNs: 0360-5442

Abstract

Abstract: The energy consumption structure has a crucial influence on the sustainable development of the economy and on the environment, and it has drawn the attention of scholars and managers. The forecasting of different types of energy consumption, especially small-sample forecasting, has been a challenging task because of the limitation of the sample size. Thus, in this study, a novel forecasting system based on fuzzy time series that is appropriate for small-sample forecasting was developed. Specifically, the fuzzy time series, which deals with the fuzzy set, is applied as the forecasting program. In fuzzy time series forecasting, the information granularity and fuzzy c-means clustering are utilized for fuzzification. Moreover, an improved chaotic electromagnetic field optimization algorithm is applied to search for the optimal parameters of the information granularity. The experiments and comparison verified that the proposed forecasting system has an excellent performance in energy consumption forecasting with great accuracy and stability, providing accurate forecasting for the energy consumption structure. Highlights: A novel hybrid forecasting system is developed for energy structure forecasting. Fuzzy time series is applied focusing on small sample forecasting. Information granularity is used to improve the interpretability of fuzzification. Improved optimization algorithm is developed to search the optimal parameters. Energy structure is calculated based on energy consumption forecasting results.