区域农业面源污染灰色多元驱动预测与优化控制模型研究

With the gradual control of industrial point source pollution, agricultural non-point source pollution has become the main factor of water environment deterioration in our country. Agricultural non-point source pollution has the characteristics of uncertainty and uneasy monitoring, which makes end managing technology difficult to apply. To develop the strategy for emission reduction from source, it is necessary to accurately grasp and predict the evolution trend of agricultural non-point source pollution in social and economic development. As regional agricultural non-point source pollution's social and economic driving effects present the features of "few data and poor information" and structural uncertainty, driving factors' identification model based on grey spread spectrum analysis, and spatial pattern division model for agricultural non-point source pollution based on multivariate grey clustering theory can be established to solve driving factor' identification problem and space differentiation. A series of multivariable grey model can be built to study direct effect, interactive driving effect and environmental policy's impact effect of the driving factors, thus solve the multiple driving prediction problem of agricultural non-point source pollution. The research on measurement model of emission reduction potential under uncertain scenarios and multi-objective grey target decision model regarding controlling strategy choice can solve the optimal control problem of regional agricultural non-point source pollution. On these bases, the empirical study of agricultural non-point source pollution prediction and control in Jiangsu province is designed, which realizes the effective connection between theoretical innovation and practical application and provides reference for regional agricultural non-point source pollution control.

随着工业点源污染治理的逐步深入，农业面源污染已成为我国水环境恶化的主要因素。农业面源污染的不确定性、不易监测性等特点使得末端治理技术难以适用。为制定源头减排策略，需要准确把握和预测社会经济发展中农业面源污染的演变趋势。课题针对区域农业面源污染的社会经济驱动效应存在“少数据、贫信息”和结构不确定性的灰色特征问题，建立基于灰色扩频分析的驱动因素识别模型，依据多元灰聚类构建农业面源污染空间格局划分模型，解决污染的驱动因素辨识和空间分异性问题；构建系列多变量灰色模型研究驱动因素的直接效应、交互驱动效应和环境政策冲击效应，解决农业面源污染的多元驱动预测问题；研究不确定性情景模式下减排潜力测度模型和控制策略选择的多目标灰靶决策模型，解决区域农业面源污染优化控制问题；在此基础上，设计江苏省农业面源污染预测与控制的实证研究，实现理论创新与实际应用的有效对接，为区域农业面源污染治理提供参考。

随着农村经济的快速发展，农业面源污染问题逐渐显现。由于农业面源污染具有分散性、隐蔽性、随机性等特点，其监测、预测和控制具有较大的不确定性。 .为此，课题面向“少数据、贫信息”条件下区域农业面源污染的多元驱动预测和优化控制问题，对传统灰色关联分析、聚类、多变量离散模型和决策模型进行扩展研究。综合灰色关联分析和混沌理论构建了灰色扩频识别方法，形成了农业面源污染驱动因素识别模型；分别从接近性、相似性和综合性视角构建了多元序列灰色关联度模型，并建立了农业面源污染空间格局划分的灰色聚类方法；研究了含有时间窗、虚拟变量和交互效应的一系列离散灰色预测模型，深入揭示面源污染形成和演化的机理，并测度各类政策和措施的减排效应与减排潜力；构建含有灰信息的多目标灰靶决策模型，为面源污染控制策略的选择提供理论工具。 .采用构建的模型对全国面源污染数据进行分析，结果显示:农业总产值、畜牧业占比、人口规模等因素对于农业面源污染的驱动作用均具有较强的滞后性，面源污染的治理任务依然艰巨；目前，各类政策对于农业面源污染起到了较好的治理效果，但是“产业与经济”类政策减排效应并未充分发挥，同时各区域减排强度存在空间差异性，形成了以经济中心向外梯度递减的分布特征；课题研究还以江苏海安县为例，开展测土施肥这一具体治理措施的减排效应分析，结果表明，实行测土配方施肥技术可以增加农产品收入6744.87万元｡如在全省能够推广，每年可减少经济成本为3.80亿元、环境成本为2.57亿元、增加经济收益为48.85亿元。.课题研究所构建的灰色系统模型能够为面源污染的预测、治理决策提供理论支持和应用工具，课题案例研究所获得的相关参数和结论也能为面源污染治理政策制定提供参考和借鉴。