Estimates of energy consumption in China using a self-adaptive multi-verse optimizer-based support vector machine with rolling cross-validation. (1st June 2018)

• Record Type: Journal Article
• Authors: Wang, Xiaoyu; Luo, Dongkun; Zhao, Xu; Sun, Zhu
• Is Part Of: Energy. Volume 152(2018)
• Page Start: 539
• Page End: 548
• Publication Date: 2018-06-01
• Subjects: China energy consumption forecast -- Self-adaptive -- Multi-verse optimizer -- Support vector machine -- Rolling cross-validation; Power resources -- Periodicals; Power (Mechanics) -- Periodicals; Energy consumption -- Periodicals; 333.7905
• DOI: 10.1016/j.energy.2018.03.120
• Languages: English
• ISSNs: 0360-5442

Abstract

Abstract: Primary energy plays a critical role in the socio-economic development of China, and accurate energy consumption forecasting can help the government to formulate energy policies. To do this, the present study aims to apply a self-adaptive multi-verse optimizer (AMVO) to optimize the parameters of the support vector machine (SVM). It employs a rolling cross-validation scheme to predict China's primary energy consumption in which the independent variables are gross domestic product (GDP) per capita, population, the urbanization rate, the share of the industry in GDP and coal's share of primary energy consumption. The results indicate that the hybrid AMVO-SVM model has higher precision than other models. Finally, we apply the hybrid AMVO-SVM model to predict the energy consumption of China between 2017 and 2030 in five scenarios. In the reference scenario, China's primary energy consumption will reach 4839.3 Mtce in 2020 and 5656.2 Mtce in 2030. Highlights: Proposed a self-adaptive multi-verse optimizer-based support vector machine model. Applied rolling cross-validation to improve prediction precision. The hybrid AMVO-SVM model has higher precision than other models. China's primary energy consumption is forecasted with the proposed AMVO-SVM model in five scenarios.