基于LIBSVM的PM2.5浓度预测模型  

Influence on Input Parameters of PM2.5 Concentration Prediction Model Based on LIBSVM

作　　者：谢申汝 钱彬彬[1,2] 杨宝华[1,2] 

机构地区：[1]安徽农业大学信息与计算机学院,安徽合肥230036 [2]农业部农业物联网技术集成与应用重点实验室,安徽合肥230036

出　　处：《洛阳理工学院学报（自然科学版）》2017年第2期9-12,共4页Journal of Luoyang Institute of Science and Technology：Natural Science Edition

基　　金：安徽省高校省级自然科学研究重点项目(KJ2016A837)

摘　　要：建立基于LIBSVM的PM2.5浓度预测模型,对合肥市5个监测点的PM2.5小时平均浓度值进行预测,分析了不同污染物浓度和不同天气状况下的预测误差。结果表明:LIBSVM模型对5个监测点的PM2.5预测结果稳定,平均绝对误差为4.763 1 ug/m^3;在输入参数污染物浓度较低和不利于污染物扩散的条件下预测误差较小,在输入参数污染物浓度较大和有利于污染物扩散的条件下预测误差较大。LIBSVM模型能够很好地对PM2.5浓度进行预测,且输入参数对于模型的预测效果具有较大影响。PM2.5 concentration prediction model is constructed by use of LIBSVM. Concentration of PM2.5 from the five monitoring stations in Hefei is predicted and the prediction error under the different pollutants concentration and different weather conditions are analyzed. Results show that prediction model of PM2.5 is stable. If the pollutant concentration of input parameters is low and it is not conducive to the spread of pollutants, the prediction error is small. If the pollutant concentration of input parameters is larger and it is conducive to the spread of pollutants, the prediction error is bigger. So the LIBSVM model can well forecast PM2.5 concentrations, but the input parameter has great influence on the model prediction.

关 键 词：PM2.5 小时平均浓度 LIBSVM 预测模型 

分 类 号：X823[环境科学与工程—环境工程]