基于网络搜索信息的农村水环境质量灰色预测模型

针对农村水环境直接监测数据相对缺乏、间接数据难以有效引入的问题,综合灰色关联分析和预测模型,提出一种基于网络搜索信息的农村水环境质量灰色预测模型。首先,综合考虑数据重要性和可获得性,确定农村水环境相关的搜索关键词清单;然后,采用主成分分析法提取搜索关键词的主要特征,构建初始网络搜索变量,并利用灰色绝对关联度衡量各初始网络搜索变量与水环境质量之间的关联程度。在此基础上,构建不同频率数据的多变量离散灰色模型,将强关联变量的降频数据作为多变量离散灰色模型的驱动因素,从而构建基于网络搜索信息的农村水环境质灰色预测模型。实例分析结果表明,相对于传统灰色模型,引入网络搜索信息可以提高农村水环境预测精度,为农村水环境治理提供决策支持。

Grey Forecasting Model of Rural Water Environment Quality Based on Online Searching Information

Water quality prediction is one of the key points in the prevention and control of rural water pollution. However, rural water environment lacks direct monitoring data, and non-direct data is difficult to introduce as variable effectively. A grey forecasting model of rural water environment quality based on online searching information is proposed by integrating grey incidence and prediction model. First, the relationship between rural water quality and online searching information is theorically analyzed, and demonstrated by the online searching data of Lanzhou water pollution event in 2014. Secondly, an initial list of searching keywords associated with rural water environment is collected according to Pressure-State-Response model. Then, by comprehensively contemplating importance and availability of the data, a final list of searching key words is determined. Furthermore, the main features of searching keywords are extracted to construct initial online searching variables by principal component analysis. Thirdly, utilizing grey absolute incidence to evaluate degree of association between each online searching variable and water environment quality. Fourthly, a new multivariable discrete grey forecasting model for different frequency data is proposed. Consequently, strongly associated variables are selected as driving factors of proposed model to construct rural water environment quality model. In case analysis, the data of the Chemical Oxygen Demand (COD) in Wuzhou, Guangxi Zhuang Autonomous Region is collected for 15 weeks. The proposed model is applied to forecast the COD with introducing the internet searching information. The performance of the model is compared with traditional grey forecasting model. The results show that introducing online searching information can significantly improve the accuracy of grey forecasting model of rural water environment quality. This study can provide decision support for rural water environment management. Meanwhile, a new approach for prediction of rural water environment is established.