Nonlinear model of cascade failure in weighted complex networks considering overloaded edges

Chaoyang Chen, Zhao Yang, Jianxi Gao, H. Eugene StanleyScientific 2020 · DOI

TL;DR

This paper introduces a novel nonlinear model for cascade failures in complex networks that accounts for edge overload capacity and weighted interactions. By incorporating these realistic features, the model demonstrates improved robustness compared to existing methods, offering a more effective framework for understanding and mitigating cascading failures in real-world systems like power grids.

What Problem Does This Paper Address?

Real-world networks, such as power grids and transportation systems, are not static and exhibit complex behaviors that traditional cascade failure models often overlook. Specifically, components in these networks possess "redundant capacity," meaning they can handle loads exceeding their nominal capacity to some extent before failing. Furthermore, the interactions between components (edges) are not uniform; they have varying strengths. Existing models, like the Motter-Lai (ML) model, often simplify these aspects, leading to a less accurate representation of how cascading failures propagate and how networks can be made more resilient. This paper aims to bridge this gap by developing a more sophisticated model that captures edge overload capacity and weighted interactions, thereby providing a more realistic and effective tool for analyzing network robustness.

Methods

The core contribution of this paper is the development of a new nonlinear cascade failure model. This model introduces several key parameters to capture the nuances of real-world networks:

  • Overload Coefficient ($\delta$): This parameter quantifies the redundant capacity of an edge, allowing it to withstand loads beyond its initial capacity before failing.
  • Capacity Parameter ($\alpha$) and Load Parameter ($\beta$): These parameters define a nonlinear relationship between an edge's load and its capacity, moving beyond linear assumptions.
  • Weight Coefficient ($\omega$): This parameter represents the interaction strength between nodes, acknowledging that not all connections are equal.
  • Distribution Coefficient ($\psi$): This parameter influences the probability of failure when an edge is overloaded, reflecting varying sensitivities to overload.

The model builds upon the concept of load redistribution, where upon failure of an edge, its load is distributed to its neighbors. The paper employs a "weighted flow local redistribution principle," where the amount of load an adjacent edge shares is proportional to its weight. Theoretical analysis is used to derive conditions for cascade failure to stop, and simulations are conducted on various synthetic network models (scale-free, random, small-world) and a real-world power grid (US power grid) to evaluate the model's performance.

Datasets

The paper utilizes several network datasets for its simulations and analysis:

  • Synthetic Networks:
    • Erdos-Renyi (ER) random networks
    • Barabasi-Albert (BA) scale-free networks
    • Watts-Strogatz (WS) small-world networks
  • Real-world Networks:
    • IEEE 118 power grid
    • US power grid

These datasets represent a range of topological structures, from homogeneous to highly heterogeneous, allowing for a comprehensive evaluation of the proposed model's applicability across different network types.

Key Insights and Ideas

  • Realistic Overload Matters: The ability of edges to handle loads beyond their nominal capacity (overload) is crucial for network robustness. Ignoring this can lead to an underestimation of a network's resilience.
  • Nonlinearity is Key: The relationship between load and capacity in real networks is often nonlinear. The proposed model's nonlinear capacity parameterization captures this better than linear models, leading to more accurate cascade failure predictions.
  • Weighted Interactions Enhance Understanding: Considering the varying strengths of connections (weighted edges) provides a more nuanced view of how failures propagate and how different parts of the network contribute to overall stability.
  • Parameter Tuning for Robustness: The study highlights that judicious adjustment of the overload coefficient, capacity parameters, weight coefficient, and distribution coefficient can significantly improve a network's resistance to cascading failures.
  • Model Superiority: The proposed nonlinear model demonstrates superior robustness compared to existing models (like the ML model) when evaluated on real-world power grid data, especially under similar cost constraints.

Key Results

  • Improved Damage Resistance: The model shows that increasing the overload coefficient ($\delta$) within a certain range significantly improves a network's damage resistance. For instance, in the IEEE 118 power grid, not considering overload led to only six active components after an edge attack, while with an overload coefficient of 1.2, most nodes remained operational.
  • Network-Specific Sensitivity: The sensitivity to parameter changes varies across network topologies. BA scale-free networks, for example, are shown to be more sensitive to the overload coefficient than other network types.
  • Optimal Parameter Ranges: The research identifies optimal ranges for various parameters. For example, there appears to be a critical value for the capacity parameter ($\alpha$) around 1.0 for maximum robustness in some networks. Similarly, the distribution coefficient ($\psi$) has a significant impact up to a certain threshold (e.g., 2.5), beyond which further increases yield diminishing returns.
  • Cost-Robustness Trade-off: The study implicitly suggests that while increasing parameters like overload capacity can enhance robustness, there's a point of diminishing returns, and the associated "cost" (defined as the ratio of capacity to load) needs to be considered. The proposed model achieves better robustness at a similar cost compared to simpler models.

Limitations and Future Work

  • Parameter Tuning Complexity: While the paper demonstrates the impact of various parameters, finding the optimal combination for a specific real-world network might still be a complex task.
  • Theoretical Simplifications: The theoretical analysis makes some simplifying assumptions, such as ignoring degree-degree correlations in networks, which might not hold true for all real-world scenarios.
  • Future Work Suggestions: The authors suggest that their methodology provides an efficacious reference for controlling and preventing cascading failures. Future work could involve applying this model to a wider range of real-world networks and exploring more advanced control strategies based on the insights gained from parameter analysis. Further investigation into the interplay between different parameters and their combined effects on network robustness would also be valuable.

Summary updated

Abstract

Considering the elasticity of the real networks, the components in the network have a redundant capacity against the load, such as power grids, traffic networks and so on. Moreover, the interaction strength between nodes is often different. This paper proposes a novel nonlinear model of cascade failure in weighted complex networks considering overloaded edges to describe the redundant capacity for edges and capture the interaction strength of nodes. We fill this gap by studying a nonlinear weighted model of cascade failure with overloaded edges over synthetic and real weighted networks. The cascading failure model is constructed for the first time according to the overload coefficient, capacity parameter, weight coefficient, and distribution coefficient. Then through theoretical analysis, the conditions for stopping failure cascades are obtained, and the analysis shows the superiority of the constructed model. Finally, the cascading invulnerability is simulated in several typical network models and the US power grid. The results show that the model is a feasible and reasonable change of weight parameters, capacity coefficient, distribution coefficient, and overload coefficient can significantly improve the destructiveness of complex networks against cascade failure. Our methodology provides an efficacious reference for the control and prevention of cascading failures in many real networks.
BibTeX
@inproceedings{chen2020nonlinear,
  author = {Chaoyang Chen and Zhao Yang and Jianxi Gao and H. Eugene Stanley},
  title = {{Nonlinear model of cascade failure in weighted complex networks considering overloaded edges}},
  booktitle = {Scientific},
  year = {2020},
  doi = {10.1038/s41598-020-69775-5},
}
Figure