A resilience-aware framework for state-conditioned attack-defence analysis and resource allocation in smart mining cyber-physical systems

This study develops a simulation-based framework that connects a corrected three-state fault-transition model, a sequence of myopic state-conditioned zero-sum attack-defence games and node-level resource allocation for smart mining cyber-physical systems. Normal, faulty and recovering states are separated by explicit entry, recovery and relapse rules. At three decision stages, the current state determines a complete 10 × 10 success-probability and payoff matrix, from which the attacker and defender mixed strategies are recomputed by linear programming. A deterministic convex benchmark then minimises total resource use subject to a raw-capability floor; a genetic algorithm is retained as a numerical benchmark. For the representative 107-node configuration, the deterministic allocation uses 67.3313 units to attain the required capability of 0.52, saving 32.7% relative to the 100-unit baseline. Across 30 genetic-algorithm runs, every run is feasible, the mean resource use is 67.4365 ± 0.0061 units and the mean gap from the deterministic optimum is 0.156%. The framework therefore supports state-conditioned strategy analysis and auditable resource planning.