Cascading Failure Model Research Articles

Vulnerability Assessment of Cyber Physical Power System Based on Improved Cascading Failure Model

Vulnerability Assessment of Cyber Physical Power System Based on Improved Cascading Failure Model

Cascading failure modelling in global container shipping network using mass vessel trajectory data

Port plays a key role in maintaining traffic flows and the effectiveness of global maritime logistics. However, the vulnerability of the Global Container Shipping Network (GCSN) is likely to increase when a single port interruption entails failures in cascading when ports encounter situations like congestions, labor strikes or natural disasters. Such situations require the deployment of port protection measures and adjustments of shipping schedules. This paper introduces a cascading model, which employs extensive and worldwide vessel trajectory data to comprehensively analyze the occurrence of cascading failures within a GCSN. The principles behind the cascading failure model are that port failures are simulated and the maritime traffic is redistributed and equilibrated to other routes and ports. A Motter-Lai overload model is applied, complemented by a three-level balanced redistribution of the traffic flows according to the specific roles of the disrupted ports. Overall, this favors the analysis of the GCSN's vulnerability, reliability, potential risks, and possible impacts. It enables maritime authorities and decision-makers to optimize service routes and mitigate the GCSN's vulnerability.

The effect of strategic alliances on supply chain resilience under external shocks

ABSTRACT Enterprises opt for alliances to bolster supply chain resilience (SCR) against external shocks. Yet, studies on the impact of strategic alliances on SCR from an operations management perspective are still lacking. This study develops a supply chain network (SCN) cascading failure model that accommodates the presence of strategic alliances, drawing upon complex network theory and cascading failure theory. Using numerical simulations and empirical examples, the study delves into how enterprises can effectively enhance their SCR by forming strategic alliances and appropriately allocating redundant resources. The findings reveal that: (I) Strategic alliances enhance SCR; (II) When enterprises maintain a degree of redundancy, the strengthening effect of alliances on SCR is even more pronounced, particularly under significant external shocks; and (III) In the context of cascading failures, disruptions to smaller enterprises have a more pronounced effect on SCR. Notably, weak-weak alliances contribute more to SCR than strong-strong alliances.

Reliability analysis of urban road traffic network under targeted attack strategies considering traffic congestion diffusion

The cascading failures caused by traffic congestion diffusion may deteriorate traffic network reliability. Comprehending urban traffic congestion mechanisms is essential for road network planning and traffic management against cascading failures. To uncover this, the reliability of urban road traffic network (URTN) under cascading failure considering different attack strategies is analyzed. The cascading failure model is established based on the improved nonlinear load-capacity relationship. Five kinds of attack strategies including Strength Attack (SA), Betweenness Centrality Attack (BCA), Eigenvector Centrality Attack (ECA), Closeness Centrality Attack (CCA), and Random Attack (RA) are selected. In particular, the capacity affected by traffic congestion is considered, providing a new perspective for the study of traffic congestion diffusion. A state update equation for networks is proposed to simulate the network congestion diffusion. Finally, a case study is conducted by using the URTN of Shanghai as the background. The results show that the network will experience large-scale congestion when high-importance nodes are attacked. The congestion degree is the highest under CCA strategy, network efficiency is the lowest under ECA strategy, and traffic quality is the poorest under CCA strategy. As the congestion critical failure threshold decreases, the speed and scale of cascading failures caused by traffic congestion diffusion are greater. Maintaining proper traffic management and control capability can largely reduce the cascading effect to a great extent and improve the reliability of the network. The results can provide a research basis for traffic management to improve network reliability.

Spatio-temporal patterns and cascading risks of embodied energy flows in China

The increasingly complex interrelation between energy use and trade is challenging the secure management of energy. However, limited knowledge of the cascading risks of embodied energy in trade is understood. An embodied energy network of China's interprovincial trade was established based on China's multiregional input-output tables for 2012 and 2017. The pivotal nodes, main flows, and flow trajectories within the embodied energy network were identified using the complex network model. Subsequently, the cascading effects within the embodied energy network were simulated using a cascading failure model. The results indicate that a substantial portion of embodied energy flows is concentrated within a few provinces and sectors. Within the provincial-sectoral network, approximately 50 % of the edges carry 90 % of the embodied energy flows. Sectors exhibiting robust influence, significant intermediary capacity, and pronounced centrality assume pivotal positions within the network. Due to cascading effects, a supply crisis in one crucial province or sector can spread quickly, affecting the entire energy network. This study contributes to ensuring regional energy supply security and offers fresh insights into mitigating cascading risks in the energy system within and beyond China.

A mobile node path optimization approach based on Q-learning to defend against cascading failures on static-mobile networks

The research on cascading failures in static networks has become relatively mature, and an increasing number of scholars have started to explore the network scenarios where mobile nodes and static nodes coexist. In order to enhance the resilience of static-mobile networks against cascading failures, an algorithm based on Q-learning for optimizing the path of the mobile node is proposed in this paper. This paper proposes a Q-learning-based algorithm for optimizing the path of the mobile node. To achieve this objective, a cascading failure model is established based on sequential interactions between mobile nodes and static nodes in this study. In this model, the motion paths of the mobile node are generated by the Q-learning algorithm. Based on this approach, extensive experiments are conducted, and the results demonstrate the following findings: 1) By increasing the adjustable load parameters of static nodes in the network, the occurrence of cascading failures is delayed, and the frequency of cascading failures is decreased. 2) Increasing the adjustable load parameter, capacity parameter, and network size of static nodes contributes to the network's resilience against cascading failures. 3) As the communication radius of the mobile node increases, the scale of failures in the static network initially increases and then decreases. 4) Different trajectories of the mobile node have a significant impact on network robustness, and paths generated based on Q-learning algorithm exhibit significantly improved network robustness compared to Gaussian-Markov mobility trajectories. The Q-learning algorithm is compared to the Ant Colony Optimization algorithm in terms of execution time, path length, and network robustness, and the Q-learning algorithm demonstrating favorable performance. These experimental results can be valuable for theoretical research on cascading failures in static-mobile networks.

Simulation system design for critical-node detection and robustness analysis of supply chains

ABSTRACT A well-developed manufacturer requires a stable supply chain network that is robust to crises. In this study, a simulation system was developed to build a specialised supply chain model and analyse the critical nodes and risks. It makes three major theoretical contributions to literature. First, a novel supply chain model and cascading failure model are developed with adjustable power-law degree distributions for companies to identify suitable models. Second, a novel approach to the critical-node detection problem, considering cascading failure, is proposed. Finally, based on the proposed algorithm, supply chain vulnerability is analysed under two types of attacks.

Inferring cascading effects of natural and built environments failures from social media videos in disasters

The objective of this study is to infer the cascading effects between disruptive events in disasters using content coding and imagery representation techniques on social media videos. It is well recognized that disaster events are cascading due to the dynamic interactions between natural and built environments. Understanding why and how the failures of natural and built environments such as road and vehicle damages cascade is critical for retrospective investigation and disaster adaptation. Social media provides high-volume and real-time data, which has been broadly used in disaster management. What's more, the immense availability of social media videos, which record the facts of disaster events with high spatial and temporal resolutions, provides a unique opportunity to review and uncover the cascading process in disasters. Here, we propose an analytical method from two theoretical perspectives: content understanding and the philosophy of cascading effects. First, we establish a content classification system to code social media videos with respect to specific disaster events. Then, an imagery representation approach is developed to capture the cascading effects based on the observed relations from the social media videos. To examine the application of these methods and to derive new insights for the cascading effects, we leverage empirical social media video data collected from Weibo, a popular Chinese microblogging platform, during Typhoon Lekima. The coding results show that social media videos contain a wealth of information regarding disaster impacts, enabling us to unveil the occurrences and processes of the natural and built environment failures. By establishing cascading relations among disaster failures, we can present a whole picture of the complex disaster events induced by typhoons. The proposed method and findings of this study can contribute to not only computational modeling of cascading failures within natural and built environment, but also informing the design and planning of cities in adapting to future disasters.

Effect of network topologies and attacking strategies on cascading failure model with power-law load redistribution

Various traffic networks play an important role in daily life and have different topological characteristics such as small-world and scale-free. The factors of traffic congestion, natural disasters and traffic accidents may induce cascading failure in which the load redistribution usually has the characteristic of power-law (that is to say, when a station is broken, the great majority of passengers would evacuate to the nearby stations, but the very few would evacuate to further). Based on them, this paper studies the load-capacity cascading failure model with power-law load redistribution using three kinds of attacking strategies (removing a single node according to the load, removing multiple nodes according to the load and removing multiple nodes randomly) for different network structures (random networks, WS small-world networks and BA scale-free networks). We confirm the robustness of our results by further performing simulations on real-world traffic networks. A quantitative description of network topologies and attacking strategies vs. the cascading failure model with power-law load redistribution is obtained by computer simulation. The results demonstrate that in contrast to removing the node according to the load, network robustness increases with the increase of the load control parameter α when randomly removing the node; when θ is small, the load tolerance parameter β is sensitive to the network robustness in random networks and WS small-world networks when removing a single node according to the load.

A cascading failure model of cyber-coupled power system considering virus propagation

Cyber attacks from the information layer can have a serious impact on the operation of the cyber-coupled power system. In this paper, a practical cascading failure model used for investigating the impact of virus propagation on the cascading failure of cyber-coupled power system is proposed. The model takes into account the actual functionality of cyber-coupled system and the dynamic virus propagation in cyber network, and describes the effects of cyber node failure caused by virus infection and power flow overload on the cascading failure. Moreover, the impact of defense upgrade on the virus propagation is considered. Centralized control based on dispatching center and local regulation based on static power-frequency characteristic are introduced to describe the actual monitoring function of cyber system on the power grid. Then, the effects of defense upgrade strategy and modular community structure on the failure evolution are analyzed. The results indicate that the spread of virus destroys the monitoring function of cyber network on the power network and leads to more serious failure results. The effectiveness of defense upgrade strategies is related to the upgrade probability of information stations. In addition, the stronger community structure of cyber network can improve the robustness of cyber-coupled power system.

Dual nature of cyber–physical power systems and the mitigation strategies

Cyber–physical power system (CPPS) offers benefits in terms of enhanced grid operation and improved performance, but it also brings forth new threats to the power system posed by digital intelligence. In this paper, we propose a general cascading failure model to analyze the vulnerability of CPPS when facing various attack modes (physical attacks and information attacks). In addition, we introduce an all-type connected islands (ACI) mechanism in model to reflect the fact that a well-working island should contain all types of nodes. Then we conduct experiments under different scenarios, such as different attack modes, different coupling strategies, and different power flow calculation models. Our findings demonstrate that, on the one hand, the introduction of ACI enlarges the cascading failure size in CPPS. On the other hand, the double-edged sword effect of CPPS exists in various scenarios. Furthermore, given the potential threat of information attacks, we propose a fused centrality index from the perspectives of structure and function, and apply it to two mitigation strategies before and during the cascading process. Experimental results show that these strategies are effective in mitigating the damages of the information attacks.

Identifying the critical nodes in multi-modal transportation network with a traffic demand-based computational method

In the last decade, there has been an abundance of research on identifying critical nodes in single-modal transportation network, but less for multi-modal transportation network (MTN). In addition, none of them consider the effect of interactions between nodes in a transportation corridor on node importance. For this purpose, a modified weighted k-shell (MWKS) model is proposed for identifying the critical nodes of the MTN, which combines the network topology and traffic demand. It is modified by the differential contribution of the multi-order neighbor nodes embodied by the passenger volume and the transportation corridor idea. Then the effectiveness of MWKS is verified with an example of a real transportation system using susceptible-infected (SI) model and cascading failure model. The analysis proves that MWKS is suitable for identifying critical nodes in the MTN. Furthermore, the proposed contribution attenuation coefficient based on the idea of the transportation corridor outperforms the topology-based and constant. It demonstrates the necessity of capturing the difference in the contribution of each neighbor node based on traffic demand when calculating the node importance. The results can provide a research basis for traffic planning to improve network reliability.

Topological analysis, endogenous mechanisms, and supply risk propagation in the polycrystalline silicon trade dependency network

High-purity polycrystalline silicon, as a core raw material in the photovoltaic industry, has a trade structure whose robustness affects the supply security of the entire photovoltaic industry. Using social network analysis methods and dependence indicators, this study constructs a Polycrystalline Silicon Trade Dependency Network (PSTDN) from 1995 to 2019, and performs descriptive statistical analysis and trade community divisions on the network. The Temporal Exponential Random Graph Model (TERGM) is used to explore the factors influencing the formation of trade dependencies. Based on the cascading failure model, a Photovoltaic Multilayer Network Supply Risk Propagation Model (PMNSRPM) is constructed to study the risk propagation process of polycrystalline silicon supply interruption as a raw material in the photovoltaic industry. Finally, using the risk propagation range and dependency degree, the study measures the importance of nations comparatively. The results show that (1) the PSTDN exhibits reciprocity, geographical clustering, and weak network convergence; (2) the evolution of the PSTDN's community has gone through three stages: the rising expansion phase from 1995 to 2007, the crisis recession phase from 2008 to 2010, and the competitive contraction phase from 2010 to 2019; (3) the evolution of the PSTDN shows strong reciprocity effects, transitive effects, stabilization effects, and lag effects as endogenous mechanisms; (4) the average risk propagation range is continuously increasing along the supply chain, while there is no significant difference in the risk propagation rate along the supply chain. The risk propagation rate of different risk sources shows a long-tail effect, with the risk propagation rate of leading countries like the United States and China growing rapidly over time. The United States can infect China, but China cannot infect the United States and Western Europe; (5) The risk propagation range and the dependency degree present a positive correlation ranging between 0.5 and 0.7, which is not as highly positive as initially anticipated, and the correlation between them shows a downward trend. This study explores the statistical characteristics of the PSTDN, community evolution, and evolutionary factors, demonstrates the importance of polycrystalline silicon supply security for the production of the photovoltaic industry, quantifies the risk propagation capabilities of different countries, and identifies a group of countries that actually have a high risk propagation range but tend to be underestimated due to their low dependency degree in normal trade network. This can help enhance the early warning capabilities for photovoltaic supply security risks of various countries.

Cascade failure analysis of power-information interdependent network

Modern information technology has been constantly evolving, transforming the traditional power grid into a network that couples both power and information layers. Understanding the cascade failure behavior of such power communication interdependent networks is essential for effectively controlling catastrophic network failures, preventing system collapse, and ensuring normal network operation. This research can contribute to the development of tools to predict and prevent such failures, and restore normal network functions in a timely manner. This paper focuses on the modeling method and cascading fault analysis of the power-information double-layer coupling network. We construct power information interdependent networks based on IEEE30 system and England39 system, and evaluate the cascade failure results using load distribution cascade failure model and HITS algorithm. The evaluation criteria include network efficiency, residual network size, and residual network load. By analyzing these parameters, we can gain insights into the performance of the power-information interdependent networks during cascade failures. Through simulation results, we demonstrate that the type i attack proposed in this paper renders the network structure unstable and less robust compared to the degree attack, intermediate attack, and random attack. These findings provide valuable references for developing strategies to mitigate the cascading failure of power-information interdependent networks.

Research on Chaotic Characteristics of Cascade Failure in Rail Transit Networks considering Nonlinear Load Fluctuations

The stable and efficient operation of rail transit networks (RTNs) is critical for the integrated development of metropolitan areas. However, numerous studies have indicated that RTNs are prone to large‐scale cascading failures when subjected to disturbances. To address the limitations of traditional cascading failure models, this paper proposes an innovative cascading failure model for metropolitan areas RTNs, which incorporates nonlinear load fluctuations and the bounded rationality of passengers. This model aims to capture the cascading failure characteristics of RTNs with chaotic properties under 12 combination strategies. A single‐ and dual‐parameter coupling analysis of chaotic evolution parameters and prospect theory parameters are conducted. Taking the RTN in the Chengdu metropolitan area as an example, both the static characteristics and cascading failure features of the network are analyzed. The findings reveal the following: (i) the RTN is a assortativity network and lacks small‐world and scale‐free properties. (ii) During network disturbances, a higher level of passenger familiarity with the network increases the likelihood of large‐scale cascading failures. (iii) When passengers tend to avoid risks, stations with higher carrying capacity are more prone to failures. This study holds significant implications for ensuring the stable and reliable operation of rail transit systems within metropolitan areas.

Supply chain research based on complex network theory

Supply chain is a chain structure formed by the sequential processes of production and distribution, spanning from raw material suppliers to end customers. An efficient and reliable supply chain is of great significance in enhancing enterprise’s market competitiveness and promoting sustainable social and economic development. The supply chain includes the interconnected flows of materials, resources, capital, and information across various stages, including procurement, production, warehousing, distribution, customer service, information management, and financial management. By representing the various participants in the supply chain as nodes and their interactions—such as the logistics, capital flow, information flow, and other interactions—as edges, the supply chain can be described and characterized as a complex network. In recent years, using complex network theory and methods to model and analyze supply chains has attracted increasing attention from researchers. This paper systematically reviews the supply chain research based on complex network theory, providing an in-depth analysis of supply chain networks in terms of network construction, structural properties, and management characteristics. First, this paper reviews two kinds of approaches to constructing supply chain network: empirical data-based approach and network model-based approach. In the empirical data-based research, scholars use common supply chain databases or integrate multiple data sources to identify the supply chain participants and clarify their attributes, behaviors, and interactions. Alternatively, the research based on network models employs the Barabási–Albert (BA) model, incorporating factors such as node distance, fitness, and edge weights, or uses hypergraph models to construct supply chain networks. Next, this paper summarizes the research on the structural properties of supply chain networks, focusing on their topological structure, key node identification, community detection, and vulnerability analysis. Relevant studies explore the topological structure of supply chain networks, uncovering the connections between nodes, hierarchical structures, and information flow paths between nodes. By analyzing factors such as node centrality, connection strength, and flow paths, the key nodes within the supply chain network are identified. Community detection algorithms are used to investigate the relationships between different structural parts and to analyze the positional structure, cooperative relationships, and interaction modes. Furthermore, quantitative evaluation indicators and management strategies are proposed for the robustness and resilience of supply chain networks. Further research has explored the management characteristics of supply chain networks, including risk propagation and competition game. Relevant studies have employed three main methods—epidemic model, cascading failure model, and agent-based model—to construct risk propagation models, simulate the spread of disruption risks, and analyze the mechanisms, paths, and extent of risk propagation within supply chain networks. These studies provide valuable insights for developing risk prevention and mitigation strategies. In addition, the game theory has been used to investigate the cooperative competition, resource allocation, and strategy selection among enterprises within the supply chain network. This paper reviews the research contents and emerging trends in supply chain studies based on complex network methods. It demonstrates the effectiveness and applicability of complex network theory in supply chain network research, discusses key challenges, such as how to obtain accurate, comprehensive, and timely supply chain network data, proposes standardized data processing methods, and determines the attributes of supply chain network nodes and the strength of their relationships. Furthermore, research on the structure of supply chain network has not yet fully captured the unique characteristics of supply chain networks. Existing models and methods for vulnerability assessment often fail to consider the dynamic and nonlinear characteristics of supply chain networks. Research on risk propagation in supply chains has not sufficiently integrated empirical data, overlooking the diversity of risk sources and the complexity of propagation paths. The asymmetry and incompleteness of information in supply chain networks, as well as multiple sources of uncertainty, make the prediction and analysis of multi-party decision-making behavior more complex. This paper also outlines several key directions for future research. One direction involves using high-order network theory to model interactions among multiple nodes and to describe the dynamics of multi-agent interactions within supply chain networks. Furthermore, integrating long short-term memory (LSTM) methods to process long-term dependence in time-series data can enhance the analysis of network structure evolution and improve the prediction of future states. The application of reinforcement learning algorithms can also adaptively adjust network structures and strategies according to changing conditions and demands, thereby improving the adaptability and response speed of supply chain networks in emergency situations. This paper aims to provide valuable insights for supplying chain research and promoting the development and application of complex network methods in this field.

Cascading failure resilience analysis and recovery of automotive manufacturing supply chain networks considering enterprise roles

With the rapid development of global trade, the member enterprises in the automotive manufacturing supply chain are widely spread around the world. As a result, the scale of the automotive manufacturing supply chain network has become increasingly large, and the structure has become more complex. At the same time, the connection between enterprises is becoming closer. On one hand, the close business relationships between enterprises and the vast network structure accelerate the operation efficiency of the automotive manufacturing supply chain network. On the other hand, it also exposes the entire network system to high risks of cascading failures when facing disruptions due to enterprise production interruptions. Therefore, this work aims to establish a realistic cascading failure model for the automotive manufacturing supply chain network to explore its cascading failure characteristics. Based on this model, a two-stage recovery strategy is proposed to enhance the network’s cascading failure resilience. Experimental results demonstrate the following findings: (1) underload nodes that emerge after load redistribution are the main cause of cascading failure resilience degradation in automotive manufacturing supply chain networks; (2) cascading failure resilience is negatively correlated with the load-exponential coefficient, the capacity lower-limit coefficient, and the attenuation-tolerance coefficient; (3) among four different types of attacks, cascading failures caused by the high eigenvector centrality attack strategy and the high degree attack strategy impose the most severe damage on the automotive manufacturing supply chain network; (4) the proposed two-stage recovery strategy can significantly improve the cascading failure resilience of the automotive manufacturing supply chain network.

Reliability analysis for complex electromechanical multi-state systems utilizing universal generating function techniques

The large-scale adoption of complex electromechanical systems (CMESs) renders it challenging to perform reliability assessments for such systems. In terms of the reliability evaluation of CMESs, previous studies generally assume that only binary states exist within the systems. However, owing to the stochastic failures and various operating characteristics of the system components, not only the good/bad states but also the performance levels within the system should be taken into account when analyzing the reliability of CMESs. Hence, we abstract the CMES as a multi-layer and multi-state network in which minimum maintenance units (MMUs) and three types of connections constitute nodes and edges in different layers. Furthermore, multi-state models for different MMUs in the CMES are developed based on universal generating functions (UGFs). Aggregation operators are then utilized to aggregate these UGFs to obtain the multi-state models. Subsequently, a cascading failure model is introduced to incorporate MMUs into the system reliability assessment. Thus, the effects of the operating mechanism and topological attributes can be considered in the reliability evaluation of the CEMS. Finally, the CRHX high-speed train system is used as an example to demonstrate the applicability and effectiveness of the proposed method in the reliability assessment of the CMES.

Cascading failure mechanism of major drainage system in mountainous city: Taking the basin of the main urban area of Chongqing as an example, China

Failure of major drainage systems is almost always accompanied by waterlogging. Understanding the operation and failure mechanism of major drainage systems is an important prerequisite for understanding and managing waterlogging. The natural and artificial environments in mountainous cities are more complex, and the major drainage system, as an open system for excessive surface runoff discharge and storage, is more prone to local overload and cascading failure. Existing theories and methods are difficult to explain and analyze the cascading failure characteristics and principles of major drainage systems in mountainous cities. Based on the complex network technology and the principle of runoff balance, the cascade failure model of major drainage system was constructed, and the characteristics and laws of cascade failure under six kinds of terrain, six kinds of short duration rainstorm and two kinds of surface environments were quantitatively analyzed. Research has found that the cascading failure of major drainage systems and waterlogging in mountainous cities should be the same phenomenon from different perspectives, and the simulated matching degree between the two can reach 79.5%. Meanwhile, 456 potential waterlogging locations were discovered in the study area, which is much larger than the actual 73 locations. When urban construction and rainfall conditions change, these unstable potential waterlogging locations are easily apparent. Among them, terrain is the main factor restricting the cascading failure of major drainage systems. Low lying valley areas often quickly accumulate a large amount of overflow, while flat areas are prone to overflow spreading to the surrounding areas; Slope areas are not prone to overflow, but are prone to local flooding. The change of short duration rainstorm and the difference of surface environmental openness will further interfere with the cascade failure of the major drainage system. The longer the return period of the short duration rainstorm or the increase of the peak value of the rainfall pattern, the greater the discharge and storage pressure of the major drainage system per unit time, resulting in intensified waterlogging. The open and unobstructed surface environment will lead to the small-scale spread of waterlogging, but this will significantly reduce the overall degree of waterlogging in the late rainstorm. The study revealed the spatiotemporal principles and basic laws of cascading failure of major drainage systems in mountainous cities. The conclusion is helpful for mountainous cities to formulate urban construction management strategies and climate change adaptation strategies, and to cope with current and future potential waterlogging disasters.

Resilience improvement of cyber-physical supply chain networks considering cascading failures with mixed failure modes

Failure mode analysis and recovery strategy improvement are critical to maintain the resilience of cyber-physical supply chain networks (CPSCNs). However, current literature mostly focuses on single failure mode and single-stage recovery strategy for CPSCNs, which may not be the case for real-world CPSCNs. In this paper, a new cascading failure model with four mixed failure modes and a joint improvement model for two-stage recovery regarding CPSCNs are proposed. A CPSCN model consisting of a weighted undirected cyber-supply network and a weighted directed physical-supply network is established to investigate the mixed cascading failures. Given the recovery requirements of practical contracts, the recovery time constraint on the enterprises in CPSCNs is considered in the joint improvement model of two-stage recovery. Based on the case studies of synthetic networks and a European supply chain network, the effects of some key factors, such as network topology and node coupling modes, on the resilience of CPSCNs are discussed. The performance of the proposed recovery strategy is compared with two traditional network recovery strategies, and the results show that the proposed strategy can help restore network performance more effectively. This work provides insights for CPSCNs recovery with resilience improvement against future cascading failures.