Forecasting the energy intensity of industrial sector in China based on FCM-RS-SVM model.

Abstract

Analysis of industrial energy intensity is greatly significant in China specifically from the perspective of sector heterogeneity due to considerably different levels of energy utilization in various industrial sub-sectors. This study proposes a new methodology to forecast energy intensity in industrial sub-sectors, considering the complexity of the socioeconomic system. This research collects the data of 36 industrial sub-sectors in China and combines fuzzy C-means clustering (FCM), rough set (RS) and support vector machine (SVM) to predict the energy intensity of industrial sub-sectors in 2030. First, this method classifies all the industrial sub-sectors according to energy intensity level and identifies the main factors that affect the energy consumption of the industrial sub-sectors. Second, the resulting classification paves the way for specifying models to forecast energy consumption. Finally, scenario analysis predicts the energy intensity of each industrial sub-sector in 2030. This exploration has the following results. (1) Energy intensity has significantly different trends in various industrial sub-sectors. For example, industrial sub-sectors with low energy intensity mainly belong to the manufacturing industry (S06-S33). In contrast, the medium- and high-energy intensity categories mainly belong to the mining industry (S01-S05) and energy extraction and supply industry (S34-S36). (2) The critical factors affecting industrial energy consumption are fixed assets, R&D investment, and labor investment. (3) By 2030, the energy intensity has a downward trend in various industrial sub-sectors in China. The scenario analysis implies that China's energy intensity would reach the current world average level under the low-speed development scenario. Also, China's energy intensity would reach the current world advanced level under the medium-speed or high-speed development scenario.