应用贝叶斯网络模型评价象山港养殖区富营养化风险

为了真实、准确地反映象山港养殖区海域的富营养化状况，以2003—2007年，5年间的象山港养殖区水质监测资料为训练数据并结合有关研究结果，采用K2结构学习算法，构建了基于贝叶斯网络的象山港养殖区水质富营养化风险评价模型，通过该模型得到了各变量间的因果关系及其影响强度。结果表明，试验数据显示准确性为90．18％，Kappa指数为0．8770，以上证明该方法是有效可行的，预测结果表明象山港养殖区水环境富营养化程度日趋严重。

The Model Based on Bayesian Network to Assess Eutrophic Risk in Xiangshan Bay

In order to show the eutrophication status of the Xiangshan bay truly and accurately. The monitoring data of Xiangshan bay from the year of 2003 to 2007 was taken as the training data and the prior knowledge was added. A bayesian network based on the K2 algorithm was constructed. The model can effectively express the causal relationship among the various indicators in the net-cage aquaculture environment, and the eutrophication of the Xiangshan bay can be predicted. The result showed that the appraisal accuracy reached 91.7%, Kappa is 0.877 0, which meant that this method is feasible. The forecasting results showed that the water environment of the aquaculture area in Xiangshan bay has been in eutrophic state.