A tripartite evolutionary game approach to performance-based assessment of nuclear power plant physical protection system

Traditional static security assessments difficult to capture the dynamic strategic interactions and behavioral decision-making inherent in extreme nuclear security risks. To address this, we propose a tripartite evolutionary game model involving an attacker, a defender, and a regulator. Our model integrates Regret Theory to capture the regret aversion psychology in high-consequence, low-probability (HILP) environments and incorporates the EASI (Estimate of Adversary Sequence Interruption) model to quantify probability of successful intrusion. We derive replicator dynamics to establish threshold conditions for evolutionarily stable strategies (ESS). Using a Small Modular Reactor (SMR) case study, our 3D topological flow analysis demonstrates that optimized regulatory parameters reliably steer the system toward a globally stable, low-risk equilibrium. Conversely, inadequate incentives or extreme threat payoffs lead to high-dissipation traps like regulatory fatigue. This framework provides robust quantitative decision support for nuclear power plant Security-by-Design and multi-party security governance.