基于BP神经网络的PM_(2.5)浓度值预测模型  

PM_(2.5) Concentration Prediction Model Based on BP Neural Network

作　　者：张丹宁 吴巧丽 张博[3] ZHANG Dan-ning;WU Qiao-li;ZHANG Bo(Shanxi Polytechnic College,Taiyuan,Shanxi,030006,China;Xi′an Institute for Innovative Earth Environment Research,Xi′an,Shaanxi,710000,China;Northwestern Polytechnical University,Xi′an,Shaanxi,710072,China)

机构地区：[1]山西职业技术学院,山西太原030006 [2]西安地球环境创新研究院,陕西西安710000 [3]西北工业大学光电与智能研究院,陕西西安710072

出　　处：《建材技术与应用》2023年第2期9-13,共5页Research and Application of Building Materials

摘　　要：PM_(2.5)对空气质量的恶劣影响和对生命健康的严重威胁日益引起了各界的关注。揭示PM_(2.5)时间分布规律,对其浓度进行有效预测,有助于大众及时采取防控措施和降低污染暴露强度。故以西安市为研究区,基于反向传播神经网络(BP神经网络),应用2014年1月1日至2017年11月4日的1400组大气污染物监测数据进行训练学习,并用2017年11月4日至2018年8月31日的300组数据进行测试和检验,最终建立了精度较高的PM_(2.5)浓度预测模型,用以预测次日PM_(2.5)浓度值,并针对偏差较大的预测结果,进行了成因分析和讨论。The adverse impact of PM_(2.5) on air quality and serious threat to life and health has attracted increasing attention from all walks of life.Revealing the temporal distribution of PM_(2.5) and effectively predicting its concentration will help the public take timely prevention and control measures and reduce pollution exposure intensity.In this study,Xi′an City was selected as the research area,based on back propagation neural network(BP neural network),1400 sets of air pollutant monitoring data from January 1,2014 to November 4,2017 were used for training and learning,and 300 sets of data from November 4,2017 to August 31,2018 were used for testing and verification.Finally,a prediction model of PM_(2.5) concentration with high accuracy was established to predict the next day′s PM_(2.5) concentration value,and the causes of the prediction results with large deviations were analyzed and discussed.

关 键 词：PM_(2.5)浓度值 预测模型 反向传播神经网络 成因分析 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] X513[自动化与计算机技术—控制科学与工程]