摘要
应用贝叶斯网络的逆向推理功能对供水工程水质风险进行计算,分析推理出供水工程水质风险的主要因素及其风险概率,可为供水工程安全运行提供理论依据。在收集历史资料、结合专家知识的基础上,对供水工程水质风险因素进行了识别,确定了引发水质风险的变量节点,构建变量节点之间的水质风险贝叶斯网络有向无环拓扑图,应用MATLAB对目标水质风险进行了分析计算。多因子联合影响下的水质风险概率,以调水源头水质和污水排放组合情况进行讨论,调水源头水质为Ⅱ类水时,对应污水排放为点源、非点源、不排放的情况下,供水水质级别大于Ⅲ类的概率分别为0.964 3、0.971 7、0.489 6,可见点源污染和非点源污染导致水质风险发生的概率增大,污水不排放时水质风险发生的概率将减小。贝叶斯网络具有处理多状态复杂模型以及双向推理的优点,相比传统的分析方法,其能够有效地进行水质风险原因的推理及其水质风险原因概率的定量计算。
Backward inference function of Bayesian network is applied to calculation of the water quality risk of water supply project,analysis and reason of the main factors and risk probability of water quality risk,which can provide a theoretical basis for the safe operation of the water supply project.On the basis of historical data collection,combined with expert knowledge,risk factors for the water quality risk of water supply project were identified,and variable nodes of the risk of the water quality risk were determined,Bayesian network directed acyclic topology graph about the water quality risk between variable nodes was established,and using MATLAB to analyze and calculate the risk of target water quality.Water quality risk probability under joint influence of multi-factor,combination between water quality of water transfer and sewage discharge,is calculated.When water quality of water transfer sourceisⅡ,corresponding to sewage discharge are point source pollution,nonpoint pollution and sewage no emissions,the probability which the water level is greater than classⅢ are 0.964 3,0.971 7and 0.489 6.This shows that point source pollution and nonpoint pollution lead to the water quality risk increased.Water quality risk probability will decrease with sewage no emissions.Bayesian network has advantage of handling complex multi-state model bidirectional inference.Compared to traditional methods,Bayesian network can effectively reason the causes of water quality risk and quantitatively calculate probability of water quality risk causes.
出处
《黑龙江大学工程学报》
2015年第1期5-11,共7页
Journal of Engineering of Heilongjiang University
基金
高等学校博士学科点专项科研基金博导类资助课题(20126118110015)
中央财政支持地方高校发展专项资金特色重点学科项目资助
关键词
供水工程
水质风险
贝叶斯网络
风险分析
概率
water supply project
water quality risk
Bayesian network
risk analysis
probability
