垃圾能源回收系统规划的数据驱动型鲁棒性最优化方法研究

Waste-to-Energy (WtE) systems planning is the key component in developing the sustainable municipal waste management, it is also a complex class of optimization problems under uncertainty. Meanwhile, waste data plays an increasingly important role in enhancing the performance of modern waste management. There is, however, still much room for the optimization approaches development on how to utilize the waste data information in a more direct and effective manner to optimize the WtE systems performance, and simultaneously immunize the systems to the perturbations by the uncertain waste stream conditions. Therefore, this project aims to develop a set of data-driven robust optimization approaches for the WtE systems planning under uncertainty. The main contents consist of the following four parts: First, we investigate the structure of the problem, and build the base model in the form of two-stage stochastic program. Furthermore, extending the base model and applying the theory of multivariate statistics, we focus on the development of the data-driven two-stage distributionally robust optimization models for the WtE systems, where a new modelling paradigm of two-stage distributionally robust optimization will be proposed to achieve the computational tractability of the resulting models. Leveraging the data-driven robust models, we then develop the risk stress testing and analysis methods for the WtE systems. Finally, we discuss the case studies of the developed models and methods. The expected project achievement can not only provide an effective support in terms of modelling and methodology for the municipal waste management and planning in making a better use of the waste data resource, but also enriches the theory of distributionally robust optimization.

垃圾能源回收(WtE)系统规划是城市垃圾管理可持续性发展的关键一环，也是一类复杂的不确定性最优化问题。与此同时，垃圾数据对现代化垃圾管理起着越来越重要的作用。然而如何更直接有效地利用垃圾数据信息来优化WtE系统效能，同时抵抗垃圾状况不确定性对系统的冲击，相关最优化方法仍有待深入发展。因此，本项目将针对WtE系统规划建立一套数据驱动型鲁棒优化方法，并重点研究以下内容：首先分析WtE系统最优化问题的结构特征，并建立两阶段随机规划基本模型。进一步，拓展基本模型并结合多元统计理论，重点研究WtE系统的数据驱动型两阶段分布鲁棒优化模型，并提出一种两阶段分布鲁棒优化新型建模范式，以保证模型的高效求解。进而，基于所建立的数据驱动鲁棒性模型发展WtE系统压力测试分析方法。最后，探讨案例应用及分析。预期成果可为城市垃圾管理与规划更有效地利用数据资源提供有力的模型及方法论支持，同时也可丰富分布鲁棒优化理论。

在国家对垃圾管理高度重视的大背景下，本项目研究了垃圾能源再生(WtE)系统的优化决策问题。在垃圾状态不确定性下，为了更加高效精准的使用垃圾数据信息并得出鲁棒而有效的WtE系统规划方案，本项目研究了(1) 两阶段分布鲁棒优化建模求解新范式; (2) 数据驱动型WtE 系统两阶段分布鲁棒优化模型及求解方案; (3) WtE系统风险压力测试分析方法。 项目的相关成果目前已经发表于UTD24期刊POM等国际重要期刊上。本项目形成了一套针对垃圾状态不确定性的WtE系统规划高效数据驱动型鲁棒优化方法，该套方法有望在不久的将来作为核心模块被植入“垃圾管理系统”中，助力下一代可持续性数据驱动型“智慧垃圾管理”的发展。.