Objective With the advancement of information technology and regional integration, urban networks are more vulnerable to various disaster disturbances, posing serious challenges to population mobility, information transmission, industrial collaboration and innovative cooperation. Urban network resilience is an important issue in regional resilience research. The term reflects the ability of urban network systems to develop, strengthen, resist, and recover quickly from disaster disturbances, through the collaboration of urban networks. The current research on urban network resilience primarily focuses on network structure and network function, seldom considering systematic review of evaluation methods for urban network resilience. Therefore, this research comprehensively summarizes the evaluation methods for urban network resilience from the perspective of network elements.

Methods Based on the bibliometric method, this research analyzes the previous research on urban network resilience, revealing the research hotspots and evolution trends in this field. By following the workflow of network type – network characteristics – evaluation methods, the research constructs an evaluation framework of urban network resilience based on network elements.

Results More and more scholars pay attention to the evaluation methods for urban network resilience. Firstly, the characteristics of multiple urban networks and their disaster application scenarios are quite different. The transportation network focuses on the mobility of population flow and accessibility of infrastructure. The information network considers the promptness and diversity of disaster risk information transmission. The Economic network focuses on the self-sufficiency and scale of capital supply. The innovation network emphasizes the asymmetry and mediation of knowledge cooperation. Natural disasters, public health events and accidents often restrict population mobility. In this research, the transportation network is selected for resilience evaluation. Economic and innovation networks are selected to reflect the stability of industrial cooperation and technological exchange in the face of long-term disasters, such as the economic crisis, the COVID-19 epidemic, and socio-economic pressures. The information network is selected for exploring the risk perception of urban residents to various disaster disturbances. Secondly, the evaluation methods for urban network resilience based on four network elements have different advantages. The evaluation method for urban network resilience based on network node can identify the key nodes with positive influence or negative disaster transmission ability in urban networks. The evaluation method for urban network resilience based on network connection can assess the connection strength and dependency relationships between different nodes. The evaluation method for urban network resilience based on network structure can explore the urban networks with different morphological characteristics and topological structures. The evaluation method for urban network resilience based on network function can realize the function assessment by simulating multiple disaster disturbance scenarios. Thirdly, this research proposes an evaluation framework for urban network resilience based on network elements, aiming to achieve a breakthrough in network resilience evaluation from “single network evaluation” to “multiple network evaluation”. This evaluation framework involves three stages. In the first stage, when selecting the type of urban network, the intensity of disaster disturbances on urban network is considered. The urban networks include transportation networks, information networks, economic networks and innovation networks. In the second stage, the influence path of disaster disturbances on urban network characteristics is considered, and appropriate urban network characteristics are selected. In the third stage, the evaluation methods focus on four network elements, including network node, network connection, network structure and network function. When choosing the evaluation methods for urban network resilience, the types, attributes and characteristics of urban networks are considered. However, the research on urban network resilience faces limitations. 1) Little attention has been paid to the disaster propagation ability of network nodes, and the diffusion mechanism of disaster disturbances needs to be further analyzed. 2) The complex effects of connection type, connection direction and topological feature on spatial effects need to be explored. 3) Social network analysis is the main evaluation method for network structure. A scientific and unified evaluation framework has not yet formed. 4) The simulation results of network function cannot sufficiently represent the disaster disturbances in the real world.

Conclusion There exists a large amount of research on urban network resilience to resist single disaster disturbance. The research fields include urban planning, geography, disaster science, etc.. Some research directions need to be deepened. 1) Evaluation method based on network node. Network propagation model, agent-based model and other model methods need to be emphasized in future research, in order to simulate the dynamic diffusion process of information, virus, and population. The important nodes in urban network that have both resource control function and disaster adaptation ability should be identified. 2) Evaluation method based on network connection. The evaluation method of spatial effect of network connection should be improved by combining the centrality, agglomeration, transmission and other topological indicators. 3) Evaluation method based on network structure. It is necessary to integrate macro-scale and micro-scale evaluation methods, so as to effectively compare the evaluation results at different scales. 4) Evaluation method based on network function. Deep learning methods such as recurrent neural network model and long short-term memory network model should be adopted. It is necessary to establish a network function simulation model under multi-disaster scenarios to improve the accuracy of research results.