Abstract: With the deepening of high-intensity human engineering activities on a global scale, the global air pollution trend has become more and more obvious. Extreme weather events frequently occur and cause a series of disasters, and a high degree of air pollution seriously has endangered human’s health. Air quality has become a more significant problem with the development of industrialization and urbanization. The remarkable economic growth has resulted in serious environmental issues. The emissions of air pollutants from industrial and motor vehicles are currently the most important environmental risk to human health. As one of the rapidly developing countries, China has experienced rapid economy growth for the past several decades. As a result, industrial activities, urban expansion and human engineering activities have produced high intensity of air pollutant emissions in China. Since 2010, China actively has implemented the Clean Air Action to deal with air pollution, and many pollutant emissions have decreased since then. To effectively assess air qualityand provide guidance for outdoor activities, the Ministry of Environmental Protection(MEP)adopted and developed an air quality index(AQI)system in 2012.
Despite improvements over several years, China has continued to cause air pollution with a sizeable economic aggregate and dense population. Additionally, the economic development model with high energy consumption and low efficiency is an important factor that leads to air pollution in China.Considering the above existing problems, the prevention and prediction of air pollution have become the focus of researchers all over the world. Nowadays, related researches are mainly divided into the precise prediction of AQI, which is important for early air quality warning and policymakers’ work. Many AQI forecasting models have been presented for recent years, including physical models, statistical models, and hybrid models. Among them, physical methods are more complicated and time consuming, and statistical models have been proved to outperform physical methods.
In order to improve the level of short-term air quality forecasting(SAQF)and reasonably predict AQI, this research uses the PSO-SVR algorithm to construct a new hybrid forecasting model based on the meteorological data of 495 cities across the country from 2017 to 2019.The research results show that: by constructing the non-equal-weight clustering hybrid PSO-SVR model of 9 cities, the average values of RMSE and MAPE of 9 representative cities are calculated, respectively, which are better than traditional SVR, GA-SVR, BPNN, XGBoost and LSTM models, verifying the superiority of the model proposed in this study and the practical value it brings,and integrating the indirect impact of industrialization and urbanization factors in the economic and social environment on the environment, and statistics of 9 cities. The quality prediction error tolerance rate is up to 70% within 10%. It is further verified that the model improves the prediction accuracy of air quality, so that the air quality index can better serve government managers and urban residents and other related groups. The reason why the hybrid PSO-SVR has a good fitting effect at high peaks is that the model not only performs unequal-weight dimensionality reduction processing on the data, but also takes into account the non-linearcharacteristics of the data.
Most previous researches have analyzed AQI spatiotemporal distribution using statistical methods and regression analysis to study the air quality. The goal of the current research is to analyze and determine the influence of the development of industrialization and urbanization on AQI. In this period of artificial intelligence, a hybrid PSO-SVR model for air quality forecasting could effectively improve the predicting accuracy and provide a new reference source for future air pollution management.
It is crucial to pay attention to and improve China’s prediction of air quality conditions and strengthen climate change research. Therefore, in order to enhance the adaptability and accuracy of the model proposed in this study, the next step will focus on studying the impact of industrialization and urbanization on the concentration of atmospheric pollutants, and propose targeted policies and suggestions for the development of green economy and society as much as possible.

Key words: air quality prediction; PSO-SVR; structural equation model; clustering hybrid

摘要： 为提高短期空气质量预测(SAQF)水平,合理预测空气质量指数(AQI),本研究基于全国495个城市的气象数据,运用PSO-SVR算法构建了一种新的混合预测模型。混合PSO-SVR不仅对数据进行了非等权的降维处理,又兼顾了数据的非线性特征,研究结果表明构建的非等权聚类混合PSO-SVR模型输出结果的RMSE和MAPE平均值优于传统SVR,GA-SVR,BPNN,XGBoost和LSTM模型,验证了本研究提出的模型优越性及带来的研究价值;探究经济社会环境中工业化和城市化因素对AQI造成的影响,统计9个城市空气质量预测误差率在10%以内的占比超过了70%,进一步验证该模型可提高空气质量的预测精度,从而使空气质量指数更好地服务政府管理者和城市居民等相关群体。

关键词: 空气质量预测, 粒子群-支持向量回归, 结构方程模型, 聚类混合