首页|基于贝叶斯网络理论的空气质量分析与预测

基于贝叶斯网络理论的空气质量分析与预测

扫码查看
原文链接
  • 万方数据
  • 维普
为科学构建空气质量监测体系,提升空气质量预测精度,基于海量监测数据的不确定性,提出了一种基于贝叶斯网络的方法来预测空气质量指数AQI及相应的等级.以合肥市为研究对象,首先利用朴素贝叶斯分类算法来预测空气质量等级,训练得到待验样本的分类准确率为 85%,由于该算法的条件独立性假设过于严格,进一步引入贝叶斯网络模型实证研究,基于后验概率分布训练得到预测结果.仿真实验表明待验样本的AQI预测平均绝对百分比误差为6.89%,空气质量等级分类准确率为 90.28%,说明贝叶斯网络具有良好的预测效果,能为空气质量预测预报提供技术支撑,助力城市空气质量改善.
Analysis and Prediction of Air Quality Based on Bayesian Network Theory
In order to scientifically construct the air quality monitoring system and improve the accuracy of air quality pre-diction,based on the uncertainty of massive monitoring data,the method based on Bayesian networks is proposed to predict the air quality index AQI and the corresponding level.This paper takes Hefei city as the research object.Firstly,the naive Bayesian classi-fication algorithm is used to predict the air quality level,and the classification accuracy of the tested samples obtained through train-ing is 85%.Because the conditional independence assumption of the algorithm is too strict,the Bayesian network model is further introduced for empirical study,and the prediction results are obtained based on the posterior probability distribution.The simulation experiment shows that the average absolute percentage error of AQI prediction for the tested samples is6.89%,and the accuracy of air quality grade classification is 90.28%.This indicates that Bayesian networks have good prediction performance,which can pro-vide technical support for air quality prediction and help the improvement of urban air quality.

air quality levelair quality indexnaive BayesBayesian Networkposterior probability

尤游

展开 >

安徽机电职业技术学院 公共基础教学部,安徽芜湖 241000

空气质量等级 空气质量指数 朴素贝叶斯 贝叶斯网络 后验概率

2021年安徽省高校自然科学研究重点项目2021年安徽省职业与成人教育学会教育教学研究规划重点课题2023年安徽省质量工程教学研究项目中青年教师培养行动青年骨干教师境内访学研修资助项目

KJ2021A1523Azcj20210172023jyxm1333JNFX2023175

2024

东莞理工学院学报
东莞理工学院

东莞理工学院学报

影响因子：0.265
ISSN：1009-0312
年,卷(期)：2024.31(5)