基于多尺度动力学的空气质量指数特征分析

Characteristics Analysis of Air Quality Index Based on Multi-scale Dynamics

为了从不同尺度上挖掘空气质量指数及空气污染物浓度时间序列的动力特征,文章提出了一种多尺度趋势样本熵,该方法能有效处理具有强烈趋势的非线性时间序列。运用该方法研究了湖南省14个市州的AQI及PM_(2.5)和SO_(2)的动力学结构,发现AQI与PM_(2.5)的样本熵最接近,表明PM_(2.5)序列与AQI的演变趋势相似。此外,发现14个市州的3个时间序列在5~9 d的尺度上呈现最大的样本熵,表明以5~9 d为周期考虑的序列动力结构隐藏的信息量最大。进一步利用熵值对14个市州春夏秋冬4个季节进行了聚类,为寻找共同的污染源提供了依据,并为这些城市提出了空气污染防治的可行建议。

To discover the dynamic characteristics of air quality index(AQI)and air pollutant concentration time series from different scales,a multi-scale trend sample entropy is proposed.This method can effectively process nonlinear time series with strong trends.With this method,the dynamic structure of AQI and PM_(2.5)and SO_(2)in 14 cities in Hunan Province are researched.It is found that the sample entropy of AQI and PM_(2.5)are the closest,indicating that the evolution trend of PM_(2.5)sequence and AQI is similar.In addition,it is found that the three time series of 14 cities show the largest sample entropy value at the scale of 5~9 days,indicating that the amount of information,which hidden by the dynamic structure of the sequence considered with a period of 5~9 days,is the largest.Furtherly,by clustering the entropy value of 14 cities in the seasons of spring,summer,autumn and winter,a basis for finding common pollution sources will be provided.Finally,feasible suggestions for air pollution prevention and control are put forward for these cities.