基于多尺度灰元网络的中国大气污染区域协同治理研究

In order to study the Chinese air pollution problems having the regional and complex characteristics with incomplete data, short time series and limit sequences, a novel model called grey networks is designed to determine the regional coordinated policies for the Chinese air pollution problems. Firstly, the subject will analyze the status quo of air pollution and governance in China by expert interviews, field research and literature research. Then, the regional distribution of air pollution in China will be made by taking into account the geographical location, economic interaction, and industrial relations. Aim to analyze the air pollution data with the following characteristics, such as short monitoring time, limited amount, time lags and periodicity, three grey relational models for the panel data with time lags and periodicity will be developed to determine the factors causing air pollution in a specified region. At the same time, the subject will put forward the construction of grey correlation test model to verify the validity and stability of the results. Afterwards, considering that the air pollution is a complex multi-factor interaction process, and the interaction mechanism of air pollution between region and region is particularly complicated, the subject will combine the grey theory with the multi-network to propose the multi-layer complex grey network, the regional coordinated grey network and the spatial diffusion grey-element network, to analyze the regional air pollution system including the coordinated governance mechanism. Next, using the rolling prediction technique, grey multi-factor time delay rolling prediction model and catastrophe forecasting model will be put forward according to the two dimensions of data driven and policy oriented, so as to build fine atmospheric pollution early warning platform. Finally, the regional coordinated governance measures of multi subjects (government, enterprises and people) will be developed by simulating the effect of regional air pollution control in different scenarios, and the simulation model will be used to simulate the control effect and improve the efficiency of regional air pollution control, so as to reduce the obstruction of air pollution to economic system, social system and environmental system.

中国大气环境污染问题呈现出区域性复杂交互特征，且污染数据具有不完整性、时序长度短、数据量有限等灰色特征，因此构建多尺度灰元网络研究大气污染区域协同治理方案。首先通过专家访谈和文献研究等渠道，梳理中国大气环境污染发展与治理现状。然后考虑地理空间位置、经济互动、产业承接关系等方面，对中国大气污染进行区域划分。针对大气污染数据监测时间短、数据量有限并存在滞后性和周期性等特征，构建面板数据灰色关联模型、灰色面板滞后性和周期性关联模型，从人为因素和自然因素识别区域大气污染的影响因素，并构建灰色关联检验模型验证结果有效性。将灰色理论与多元网络结合，构建多层复杂灰元网络、区域协同灰元网络和空间扩散灰元网络，剖析区域大气污染系统及其协同治理机制。利用滚动预测思想，构建灰色多因素时滞滚动和灾变组合预测模型，搭建大气污染预警精细化平台。最后模拟不同情境下区域大气污染治理效果，制定多主体参与的区域协同治理措施。

日益严重的空气污染现象对国家经济社会安全、人民群众身心健康均造成了重大影响，已成为社会各界人士普遍关注的环境问题。课题以区域协同治理为研究切入点，对大气污染进行区域划分，并深度剖析其关键影响因素及作用机理、预判发展趋势、构建预警模型、提出区域协同治理手段，对有效控制大气污染现象和拓展灰色系统的评价与预测理论应用范围具有重要意义。本项目的主要研究成果体现在以下几个方面：.1）从大气污染与影响因素的时空特征出发，基于趋势变化相似性、空间溢滞性等视角，构建基于面板数据的灰色指标关联模型、面板数据的灰色综合关联分析模型、灰色时空关联模型，有力地促进了灰色关联模型群的完善，同时对大气污染进行关键影响因素识别提供方法支撑。.2）针对大气污染的复杂多变和不确定性特征，构建基于指标信息量的灰元可能度聚类模型、基于灰色可能度函数的灰元聚类方法，对大气污染协同治理区域进行划分。考虑到城市间大气污染扩散程度、协同发展的作用效果等不确定性，以灰元表示不确定量，建立三参数灰元的多阶段多属性不确定决策方法和基于三参数灰元的风险型多阶段多属性决策方法，研究各区域大气污染的风险程度及区域内城市间的协同关系。.3）根据大气污染振荡、多因素驱动特征，引入交互作用项，构建基于多因素耦合作用的多变量灰色预测模型群。给出了参数估计公式并提出了两个拓展模型，对拓展模型的适用范围进行研究，有效解决含有耦合交互关系的多因素影响的系统预测问题。鉴于驱动因素作用机制对模型精度的重要影响，通过引入驱动控制函数，多阶段识别起主导作用的驱动因素，构造多变量离散灰色幂模型的优化模型。.4）中国大气污染区域协同治理的对策。从协同治理的视角出发，针对各市大气污染的空间分布特征，划分大气污染协同治理区域，在对空气污染物来源、空气污染影响因素理论进行充分研究的基础上，利用灰关联技术量化分析了各类污染物浓度对空气质量的贡献程度。从区域治理构架、完善区域大气污染治理信息共享平台、开放大气污染治理市场化运营等多个层面提出对大气污染区域协同治理的建议。