贝叶斯网络在水资源管理中的应用

为了解决水资源管理中具有不确定性的多目标决策问题，将贝叶斯网络方法引入水资源管理中。通过对实例系统中变量间相互关系的分析，构建描述变量间不确定性关系的贝叶斯网络模型，其中包括表示其依赖关系的有向无环图和表示其具体概率依赖程度的条件概率表，并在6个目标变量均达到预期目标的前提下进行概率推理。实例结果表明：当补偿款数额增加到500元/亩时，所有的目标变量均可达到最优，因此确定出政府应给农民补偿款的数额为500元/亩的合理水资源决策方案。贝叶斯网络以图模型的方式直观地表达了实例系统中变量之间的不确定性关系，概率推理的结果兼顾了环境效益以及农民的利益，使多个预期目标均达到了最优，有效地解决了水资源管理中具有不确定性的多目标决策问题。

Application of Bayesian Network in Water Resource Management

Bayesian network is applied in water resource management to deal with the uncertainty of multi-object decision-making problem. The relationship between variables is analyzed, and then Bayesian network model is constructed, including directed acyclic graph which describes the dependent relationship of variables and conditional probability tables which express the specific level of the dependency. On the premise that all the six objective variables achieve the intended goals, the probabilistic inference of Bayesian network is taken. Results of the case study show that the compensation amount increase to 500 yuan (RMB) per Mu（Mu≈666.666 7 m²）, all of the objective variables could achieve optimization.  So the reasonable water resources decision scheme that the government should give the compensation of 500 yuan (RMB) per Mu to the farmers was proposed. Bayesian network can intuitively express the uncertain relationship between variables in the case study. The probabilistic inference result takes into the environmental benefit as well as the farmer’s benefit, so the multi-objective variables could achieve optimization. It is an effective method to deal with the multi-objective decision-making problem with uncertainty of water resource management.