A novel energy consumption forecasting model combining an optimized DGM (1, 1) model with interval grey numbers. (20th August 2019)

• Record Type: Journal Article
• Authors: Ye, Jing; Dang, Yaoguo; Ding, Song; Yang, Yingjie
• Is Part Of: Journal of cleaner production. Volume 229(2019)
• Page Start: 256
• Page End: 267
• Publication Date: 2019-08-20
• Subjects: Energy consumption -- Grey system theory -- Interval number -- Prediction -- Electricity; Factory and trade waste -- Management -- Periodicals; Manufactures -- Environmental aspects -- Periodicals; Déchets industriels -- Gestion -- Périodiques; Usines -- Aspect de l'environnement -- Périodiques; 628.5
• DOI: 10.1016/j.jclepro.2019.04.336
• Languages: English
• ISSNs: 0959-6526

Abstract

Abstract: Since energy consumption (EC) is becoming an important issue for sustainable development in the world, it has a practical significance to predict EC effectively. However, there are two main uncertainty factors affecting the accuracy of a region's EC prediction. Firstly, with the ongoing rapid changes in society, the consumption amounts can be non-smooth or even fluctuating during a long time period, which makes it difficult to investigate the sequence's trend in order to forecast. Secondly, in a given region, it is difficult to express the consumption amount as a real number, as there are different development levels in the region, which would be more suitably described as interval numbers. Most traditional prediction models for energy consumption forecasting deal with long-term real numbers. It is seldom found to discover research that focuses specifically on uncertain EC data. To this end, a novel energy consumption forecasting model has been established by expressing ECs in a region as interval grey numbers combining with the optimized discrete grey model (DGM(1, 1)) in Grey System Theory (GST). To prove the effectiveness of the method, per capita annual electricity consumption in southern Jiangsu of China is selected as an example. The results show that the proposed model reveals the best accuracy for the short data sequences (the average fitting error is only 2.19% and the average three-step forecasting error is less than 4%) compared with three GM models and four classical statistical models. By extension, any fields of EC, such as petroleum consumption, natural gas consumption, can also be predicted using this novel model. As the sustained growth in EC of China's, it is of great significance to predict EC accurately to manage serious energy security and environmental pollution problems, as well as formulating relevant energy policies by the government. Highlights: The data of regional electricity consumption are denoted as interval grey numbers. Limited information in interval grey numbers is fully transformed. The initial condition of original DGM is optimized. The forecasting results verify the proposed model's effectiveness by comparing with other seven models.