BP神经网络宜居城市可吸入颗粒物污染评估研究  

Assessment of Inhalable Particulate Matter Pollution in Livable Cities Based on BP Neural Network

作　　者：覃晓玲[1] 覃月凤 韦丽英[3] 梁建平[4] 陆杰 Qin Xiaoling;Qin Yuefeng;Wei Liying;Liang Jianping;Lu Jie(Hechi Meteorological Bureau,Hechi 547000,China;Guangxi Zhuang Autonomous Region Meteorological Bureau,Nanning 530022,China;Baise Meteorological Bureau,Baise 533000,China;Guilin Meteorological Bureau,Guilin 541001,China;Nanning Meteorological Bureau,Nanning 530022,China)

机构地区：[1]河池市气象局,广西河池547000 [2]广西壮族自治区气象台,广西南宁530022 [3]百色市气象局,广西百色533000 [4]桂林市气象局,广西桂林541001 [5]南宁市气象局,广西南宁530022

出　　处：《环境科学与管理》2022年第11期183-188,共6页Environmental Science and Management

摘　　要：针对因空气中影响因素过多、数据量杂乱,导致宜居城市可吸入颗粒物污染评估准确率低问题,提出一种基于BP神经网络的评估方法实现有效解决。采集历年城市可吸入颗粒物的准确数据,整理得到气温、降雨量、风速、降雪量等影响因素,采用谢尼诺夫模型计算其中影响程度最高的气温和风速因素。以此值为参考值,建立BP神经网络分布模型,通过关联度值归类污染物属性。参照国家污染标准与现阶段空气环境临界值,建立评价等级矩阵和实测数据矩阵,求得最终的评估结果。仿真实验证明,所提方法评估精准度高,耗用时间较少,应用性能较强。Aiming at the low accuracy of inhalable particulate pollution assessment in livable cities due to too many influencing factors and cluttered data in the air,an assessment method based on BP neural network was proposed to effectively solve the problem.Accurate data on inhalable particulate matter in cities over the years was collected,and the influencing factors such as temperature,rainfall,wind speed,snowfall,etc.were obtained.The most influential factors of temperature and wind speed were calculated using the Seninov model.Taking this value as a reference value,a BP neural network distribution model is established,and pollutant attributes are classified by the correlation value.With reference to the national pollution standard and the critical value of the air environment at this stage,an evaluation level matrix and a measured data matrix are established to obtain the final evaluation result.Simulation experiments show that the proposed method has high evaluation accuracy,consumes less time,and has strong application performance.

关 键 词：宜居城市可吸入颗粒物 BP神经网络 谢尼诺夫模型 绝对关联度 评价等级矩阵 

分 类 号：X311[环境科学与工程—环境工程]