基于BP神经网络的 PM_2.5 污染气象风险等级分类方法研究

打开文本图片集

文章编号：1674-6139（2026）04-0048-06

中图分类号：X511文献标志码：B

Study on Classification Method of PM2.5 Pollution MeteorologicalRiskLevelsBasedon BPNeuralNetwork

Cao Yang1，²，Ren Hong²，Zhao Xiaoli1，Cheng Xiang1，Liu Weihua'

（1.Sichuan Meteorological Disaster Prevention Technology Center，SichuanProvincial Ecological Meteorology and Satelite Remote Sensing Center，Chengdu 610072，China；2.Air Environmental Modelingand Pollution Controlling KeyLaboratory of Sichuan Higher Education Institutes， Chengdu University of Information Technology，Chengdu 610225，China）

Abstract：A BP-neural-network-based classification method for meteorological risk of PM2.5 pollution were studied based on the observation dataof ground PM2.5 concentrationsand meteorological elements，whichcan comprehensivelyreflect the impact of meteorological conditions on the occurrence of PM2，5 pollution.The results indicate that：the sampling time for the model sample datasets was established by analyzing the PM2.5 polution characteristics in the study area，identifying January，February，November，and December as the months with the highest frequency of PM2，5 pollution events.The determination of the model's input meteorological elements was achieved byanalyzing the changes and impacts of meteorological factors during the PM2.5 pollution process.The key meteorological elements affctingwinterairqualitywereelativehmiditygound-levelwindspeed，precipitation，andmixinglayerheight.TheBneurale work-based classification model for meteorological risk of PM2，5 pollution was established based on meteorological parameters，had been effectivelyimplementdinSanProvinc.elasifationrultsodatheassfaonacuacyfoall1citcded 60% ，with17 cities over 70% ，and5 cities over 80% ·

Key words： PM2.5 pollution；meteorological conditions；BP neural network

前言

中国进入经济快速发展阶段，但越来越严重的空气污染问题引起社会广泛关注。（剩余6255字）