Risk response decision model using risk scenario decomposition for large-scale problems

Effective risk response decision-making in complex contexts often necessitates addressing multiple interacting risk events under resource constraints. However, existing models have predominantly focused on single-type or pairwise interactions and incorporated only a limited range of influencing factors, thus facing substantial challenges in accurately capturing high-dimensional, heterogeneous interactions while maintaining computational tractability in large-scale problems. To overcome these limitations, this paper proposes a more reliable risk response decision-making framework capable of delivering precise solutions for large-scale decision problems. Specifically, the main contributions of this study encompass the following three aspects: (1) A comprehensive risk response decision model is developed to concurrently integrate multiple types of risk interactions and critical factors, thereby better reflecting real-world conditions and enhancing decision quality. (2) To mitigate the computational burden arising from the exponential growth of nonlinear terms, an equivalent formulation based on risk scenario decomposition is introduced, enabling the precise resolution of large-scale instances. (3) A Dynamic Infeasibility-Guided Genetic Algorithm is designed to efficiently generate high-quality solutions for extra-large-scale problems. Experimental evaluations demonstrate that the proposed algorithm consistently outperforms traditional approaches in terms of solution quality, stability, and computational efficiency across thousand-dimensional problem instances, underscoring its strong performance and scalability. This research provides a scalable and effective solution framework for complex risk response decision-making, offering valuable insights for large-scale risk management applications.