運用機器學習方法評估能源使用對空污影響: 臺中市 案例分析

In terms of environmental protection, ower dispatch has to justify the trade-offs between air pollution (PM2.5 and air pollutants of the others) and stable power supply. The main purpose of this research is to explore the intensity of the factors affecting PM2.5 by means of machine learning and big data methods, scientific discussion and rational exploration. As the results of this research show, there is a weak positive correlation between power generation and PM2.5, and the traffic volume presents a greater risk sensitivity uncertainty of PM2.5 in the sight of machine learning model and scatter diagram trend of the research. As the seasonal data scatter diagram shows, although Taiwan’s coal-fired generation has a downward trend during the winter (December to February), PM2.5 has an upward trend in the same period. The phenomenon may be attributed to the climate diffusion effect of foreign pollutants in the winter, but the impacts of climatic factors on PM2.5 depends on further research and verification.