冬季突发性PM_(2.5)重污染天气预警技术研究  

Early Warning Technology of Sudden PM_(2.5) Heavy Pollution in Winter

作　　者：王子钊 Wang Zizhao(Hunan Meteorological Bureau,Meteorological Service Center,Changsha 410007,China)

机构地区：[1]湖南省气象局气象服务中心,湖南长沙410007

出　　处：《环境科学与管理》2022年第10期134-138,共5页Environmental Science and Management

摘　　要：大范围的PM_(2.5)分布特征具有时空双重属性,为了及时应对冬季突发性重污染天气,提升预警准确性,提出一种冬季突发性PM_(2.5)重污染天气预警方法。通过ZigBee组网采集PM_(2.5)重污染天气数据,采用缺失值算法对数据归一化处理。引入卷积神经网络提取更深层次的突发性PM_(2.5)重污染特征,构建并列一维特征图,展开多尺度卷积操作,拼接与融合冬季突发性PM_(2.5)重污染的时空特征关系,实现PM_(2.5)重污染天气预警。仿真实验结果表明,所提方法可以有效提升PM_(2.5)重污染天气预警结果的准确性。The distribution of PM_(2.5) in a large range had dual attributes of time and space.In order to timely respond to sudden heavy pollution weather in winter and improve the accuracy of early warning,this paper proposed a method for early warning of sudden PMPM_(2.5) heavy pollution weather in winter.PM_(2.5) heavy pollution weather data were collected through ZigBee networking,and the missing value algorithm was used to normalize the data.The convolution neural network was introduced to extract the deeper features of the sudden PM_(2.5) heavy pollution.It constructed the parallel one-dimensional feature map,unfolded the multi-scale convolution operation,splice and fuse the temporal and spatial characteristics of the sudden PM_(2.5) heavy pollution in winter to realize PM_(2.5) heavy pollution weather warning.The simulation results showed that the proposed method could effectively improve the accuracy of PM_(2.5) heavy pollution weather warning results.

关 键 词：突发性污染 PM_(2.5) 重污染天气预警 卷积神经网络 遗传算法 

分 类 号：X181[环境科学与工程—环境科学]