Dynamic Analysis of China's PM2.5 Concentration——Based on the Perspective of Functional Adaptive Clustering
time-varying characteristics,fine particulate matter,functional data analysis,clustering analysis
（3）能源与可持续绿色发展 > 5. 生态文明与低碳经济
摘要录用
勇 王 / 中国矿业大学
Fine particulate matter (PM2.5) has typical time-varying characteristics, and the traditional discrete analysis method cannot describe its dynamic change process, which is not sufficient for in-depth information mining. Therefore, it is necessary to bring it into the category of continuous dynamic function for analysis. Based on the continuous dynamic perspective, this paper studies the temporal characteristics and category patterns of PM2.5 concentration changes in 74 sample cities in China from December 2014 to August 2021. Firstly, the intrinsic function of the temporal trends of PM2.5 was reconstructed by the discrete data information of actual observation. Secondly, the dominant trend of PM2.5 change was determined by functional principal component analysis (FPCA). Furthermore, the classification model of PM2.5 change was objectively divided by clustering analysis. Finally, functional variance analysis was used to test the significance of PM2.5 change difference under different types of modes, and the potential energy transformation rule of the whole and each type of modes was analysed based on the periodic period division of PM2.5 change function. There are three types of PM2.5 dynamic change patterns in China. The characteristics of the three types of PM2.5 dynamic change patterns are significantly different in the initial stage, and the decrease rate is different. The analysis shows that the three types of PM2.5 dynamic change patterns are mainly affected by the differences of industrial structure, economic development level, geographical location and policy. At the end of the stage, PM2.5 levels of different modes are close to each other and tend to have no difference. This paper expands the research perspective of PM2.5 evolution law and classification model, analyses the potential influencing factors according to the dynamic change characteristics of PM2.5, and provides reference for decision-making departments to formulate real-time and effective air pollution control plans.