基于多Agent建模的城市生活垃圾分类回收行为模拟与政策调控分析

The classification and recycling of household solid waste (HSW) has developed rapidly in many cities of China. Considering the waste separation and recycling management is a complex system engineering, how to co-ordinate in various stakeholders in the system and how to select a reasonable regulatory policy, that has become common problems for the policy-makers in the process of urban HSW management. First of all, combining with the methods of multi-agent based simulation and modeling (MABMS) technique, discrete selection modeling, social investigations and structural equation modeling (SEM), it will study the decision making mechanism of residents, resource recycling enterprises, environmental sanitation and government, and also study the mechanism of interactions between these agents in the HSW classification and recycling system. On this basis, the multi-agent based behavior simulation model of urban HSW classification and recycling system will be developed as a virtual policy control experiment platform. Then, it will select one sample from each of the developed, medium-developed and underdeveloped cities, and conduct empirical research using the developed model, to simulate micro behavioral responses of each agent and macro evolution process of the system. Meanwhile, the quantitative evaluation on the implementation effects of the metering charge of the HSW, subsidies of the recycling, improved education program and promotion of new designed recycling models will be conducted. Finally, the countermeasures and suggestions for optimizing the classification and recycling management of urban domestic waste will be put forward. This project will provide a scientific tool for systematic quantitative assessment of the effectiveness of pre-judgment policies, and provide decision-making basis for the implementation of refined management of urban HSW separation and recovery.

我国城市生活垃圾分类回收已进入高速发展期，面对这项复杂的系统工程，决策者如何统筹系统中各利益相关主体，合理选择调控政策，成为许多城市管理中面临的共性难题。首先，本项目将综合运用基于多主体的仿真建模技术、离散选择模型法、社会调查和结构方程统计分析等方法，研究城市垃圾分类回收系统中居民、资源回收企业、环卫企业和政府等主体的行为决策与交互作用机制，开发城市生活垃圾分类回收系统仿真模型，以此作为虚拟政策调控试验平台。然后，选择发达、中等发达和欠发达城市各1个样本，运用开发的模型开展实证研究，模拟垃圾计量收费、资源回收补贴、增加宣传教育渠道和新型回收模式推广等政策情景下，系统中各微观主体行为响应及宏观系统动态演化过程，定量预测、评估分析政策的实施效果。最后，提出优化城市生活垃圾分类回收管理的对策建议。本项目将为系统性定量评估预判政策的有效性提供科学工具，为开展城市垃圾分类回收精细化管理提供决策支持。

城市生活垃圾分类回收系统包含居民、企业和政府等多个主体，且主体间的相互作用机制较为复杂。目前，我国生活垃圾分类回收管理效果欠佳，政府在制定政策时缺乏对各利益相关主体的综合考虑。因此，研究城市生活垃圾分类回收系统中各主体的行为决策因子和交互作用机制，系统评估预判政策实施效果以选择有效的管理方案，是城市生活垃圾分类回收管理中的两个关键问题。. 本文首先通过对居民生活垃圾处理行为的社会调查，运用结构方程统计分析和离散选择模型方法，构建了居民垃圾处理行为决策模型；构建了企业主体行为决策利润函数，探讨了分类回收系统中各主体行为交互作用机制。在此基础上，采用多主体仿真建模方法，开发了城市生活垃圾分类回收系统多主体行为模拟模型，用以模拟政策刺激加入后系统中居民、资源回收企业、环卫企业等微观主体的行为响应和系统宏观演化过程；运用开发的模型对生活垃圾计量收费、增加宣传教育渠道、资源回收补贴和新型回收模式推广等政策的实施效果进行了预测模拟和定量化评价。. 研究发现，居民垃圾处理行为选择是4个主观因素和7个外部因素共同作用的结果，且外部因素对居民垃圾处理行为的综合影响是个体主观因素的近2倍。比较五种政策的实施效果：垃圾计量收费、资源回收补贴和增加宣传教育渠道三种政策组合实施，可有效调控居民、再生资源回收企业等微观主体行为，使得宏观系统涌现最优；情景模拟一年后，“不分类”居民由185万人减少到25万人，减少86%，“分类投放”居民由128万人减少到108万人，减少约16%，“分类交售”居民由105万人增长到281万人，增加约63%；人均垃圾日产生量由1.21 kg下降到1.04 kg，实现垃圾减量14%；资源回收率由33.4% 增加到66.5%。其他四种政策的实施效果比较排序为：新型回收模式推广情景 > 计量收费情景 > 资源回收补贴情景 > 增加宣传教育渠道情景。对居民主体实施垃圾计量收费、对回收企业主体给予资源回收补贴的经济激励政策，可有效促进两类主体形成良性互动。同时，建议加强城市分类回收配套设施规划建设，开展垃圾分类与资源回收系统融合发展，并因地制宜地推广新型回收模式。