基于贝叶斯网络的雾霾重点污染源排放预测  

Emission prediction for haze's key pollution sources based on Bayesian network

作　　者：童英华[1] 田立勤[1,2] 李靖 TONG Ying-hua;TIAN Li-qin;LI Jing(School of Computer,Qinghai Normal University,Xining 810008,China;School of Computer Science,North China Institute of Science and Technology,Beijing 101601,China)

机构地区：[1]青海师范大学计算机学院,青海西宁810008 [2]华北科技学院计算机学院,北京101601

出　　处：《计算机工程与设计》2018年第9期2894-2901,共8页Computer Engineering and Design

基　　金：国家自然科学基金项目(61472137;61403290);教育部"春晖计划"基金项目(Z2016083;Z2016109);青海省重点研发基金项目(2016-SF-130;2016-ZJ-920Q;2014-ZJ-908);河北省物联网数据采集与处理工程技术研究中心基金项目(3142016020)

摘　　要：为有效解决雾霾重点污染源监测信息异常问题,提出多条件下基于贝叶斯网络反演的雾霾重点污染源排放预测模型。构建污染源排放预测指标体系,对预测指标进行规范化处理;在基于相对熵最小优化模型基础上,给出预测指标的权重,以量化的方式给出污染源的污染等级,保证预测结果的可靠性和准确性。通过实例验证了所提算法的有效性,理论分析和系统测试结果表明,该方法具有较高的精度。To solve the problem of abnormal data during the monitoring period,a predictive model to forecast the rate of emission from haze’s key pollution sources under multiple conditions was proposed.Inversion based Bayesian network was used.A predictive index system was built before hand.Relative entropy minimization model was used to assign weights to the indexes.The pollution levels were stored quantitatively.Reliability and accuracy of the results were ensured.The proposed algorithm was applied on the practical application.The results were analyzed and presented.The proposed method shows improved precision and high accuracy in results.

关 键 词：雾霾污染源 贝叶斯网络 反演 规范化 最优权重 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] TP181[自动化与计算机技术—计算机科学与技术]