城市流动性网络中的动力学模型与韧性理论研究

随着城市化的不断推进，城市面临的各类风险也在增加，为了建设具有较强风险抵抗能力的新型城市，推进城市韧性的研究是应有之义。现阶段针对城市韧性的分析主要集中在韧性概念的完善和评估指标的建立，面临缺少有效的建模方法、难以分析动态过程等问题。课题选取城市流动性网络切入，借助城市大数据揭示人类行为模式，融合复杂网络分析技术，突破城市韧性建模和动态分析的关键技术瓶颈，形成城市流动性网络的韧性模型与基础理论。首先，通过人类移动性的实证数据，对城市人类活动进行规律分析和机制总结；其次，应用符号回归等机器学习方法，构建符合城市属性的典型动力学微分方程；再次，结合“抵抗、恢复、适应、转变”的城市韧性概念框架，考察城市流动网络模型的动力学平衡点；最后，对现有的网络韧性算法进行延拓。课题旨在建立城市移动性网络的动力学模型，回应国家对于提升城市品质的需求，应对新型冠状病毒肺炎疫情的挑战，服务城市可持续发展建设。

With the continuous advancement of urbanization, the risks faced by cities are also increasing. In order to build a new type of city with strong risk resistance, it is necessary to promote research on urban resilience. At this stage, the analysis of urban resilience mainly focuses on the improvement of the framework and the establishment of the evaluation system, facing problems such as the lack of effective modeling and the difficulty in analyzing the dynamic process of urban mobility. With the help of the human behavior patterns revealed by urban big data and the complex network analysis technology, the subject would break through the key technical bottlenecks of urban resilience modeling and dynamic analysis, and forms the resilience model and basic theory of urban mobility networks. Firstly, through the empirical data of human mobility, the subject would analyze the law and summarize the mechanism of urban human activities; secondly, the subject would construct a typical dynamic differential equation in line with urban attributes through machine learning methods like symbolic regression; thirdly, combined with the "resistance, recovery, adaptation, transformation" conceptual framework of urban resilience, the subject would investigate the dynamic equilibrium point of the urban mobility network model; finally, the subject would extend the existing network resilience algorithm. The project aims to establish a dynamic model of urban mobility, respond to the country's demand for urban quality, respond to the challenge of the COVID-19 pandemic, and serve the construction of sustainable urban development.