基于动态朴素贝叶斯分类器的明渠水华风险评估模型
发布时间：2018-07-23 18:54
【摘要】：水华风险不仅是水利工程规划时需要考虑的环境问题,也是水利设施运营时不能忽视的监测项目。为了提高明渠水化风险等级预测的准确率,针对水华成因的不确定性和发展的时序性,基于动态朴素贝叶斯网络分类器提出一种应用于明渠的水华风险评估模型。模型用水华风险等级结点对应藻叶绿素a(Chla)的浓度,并考虑了9项影响水藻生长的因素。采用主成分分析法,处理专家咨询结果,进行参数的设计。在苏州河道北门桥2011年6月初至9月初观测的53例连续监测数据上,与基于朴素贝叶斯网络分类器的评估模型进行比较实验。混淆矩阵显示对中等风险情况的预测识别率提高了15.625%,单尾配对t检验表明在显著性水平0.05时,两模型预测识别率差异显著。考虑了时序特征的基于动态贝叶斯网络分类器的评估模型对明渠中等水化风险的预测识别率提高显著。
[Abstract]:Shui Hua risk is not only an environmental problem to be considered in the planning of water conservancy projects, but also a monitoring project which can not be ignored in the operation of water conservancy facilities. In order to improve the accuracy of prediction of open channel hydration risk class, a Shui Hua risk assessment model based on dynamic naive Bayesian network classifier is proposed for the uncertainty of Shui Hua cause and development time series. The water bloom risk level of the model corresponds to the concentration of chlorophyll a (Chla) of algae, and nine factors affecting the growth of algae were considered. The principal component analysis method was used to deal with the expert consultation results and the parameters were designed. In this paper, 53 consecutive monitoring data of Beimen Bridge in Suzhou River from the beginning of June to the beginning of September 2011 are compared with the evaluation model based on naive Bayesian network classifier. The confounding matrix showed that the prediction and recognition rate of medium risk cases was increased by 15.625%, and the single tail paired t test showed that there was a significant difference in prediction recognition rate between the two models at the significant level of 0.05. The evaluation model based on dynamic Bayesian network classifier with time series features is used to improve the prediction and recognition rate of open channel moderate hydration risk significantly.
【作者单位】： 河南工程学院;
【基金】：国家自然科学基金项目(U1304702) 河南省科技厅软科学项目(152400410480) 河南工程学院博士基金(D2015026)~~
【分类号】：TP18;TV672;X824
，
本文编号：2140338
[Abstract]:Shui Hua risk is not only an environmental problem to be considered in the planning of water conservancy projects, but also a monitoring project which can not be ignored in the operation of water conservancy facilities. In order to improve the accuracy of prediction of open channel hydration risk class, a Shui Hua risk assessment model based on dynamic naive Bayesian network classifier is proposed for the uncertainty of Shui Hua cause and development time series. The water bloom risk level of the model corresponds to the concentration of chlorophyll a (Chla) of algae, and nine factors affecting the growth of algae were considered. The principal component analysis method was used to deal with the expert consultation results and the parameters were designed. In this paper, 53 consecutive monitoring data of Beimen Bridge in Suzhou River from the beginning of June to the beginning of September 2011 are compared with the evaluation model based on naive Bayesian network classifier. The confounding matrix showed that the prediction and recognition rate of medium risk cases was increased by 15.625%, and the single tail paired t test showed that there was a significant difference in prediction recognition rate between the two models at the significant level of 0.05. The evaluation model based on dynamic Bayesian network classifier with time series features is used to improve the prediction and recognition rate of open channel moderate hydration risk significantly.
【作者单位】： 河南工程学院;
【基金】：国家自然科学基金项目(U1304702) 河南省科技厅软科学项目(152400410480) 河南工程学院博士基金(D2015026)~~
【分类号】：TP18;TV672;X824
，
本文编号：2140338
本文链接：https://www.wllwen.com/kejilunwen/shuiwenshuili/2140338.html
