Evolutionary Game Analysis and Capability Reconstruction of Coal Mine Disaster Emergency Governance Oriented Toward Collaborative Resilience

Coal mine disaster emergency governance requires sustained coordination among regulators, mining enterprises, and professional rescue forces. This study develops a budget-closed tripartite evolutionary game in which subsidy/support outflows and penalty-revenue incidence are represented explicitly. Replicator dynamics is interpreted as a mean-field approximation of repeated organizational learning, and a cross-product extension is used to examine higher-order synergy. Dynamic stability is assessed as local asymptotic stability rather than classical evolutionary stability. Resilience is quantified using measure-conditioned basin stability PH(ν), direction-conditioned critical disturbance magnitude Δcrit(d), and conditional recovery time Trec. The open form is retained only as a mechanism-isolation benchmark. Under full fiscal closure, the original high-incentive parameter set shifts the stable outcome to E7 = (0, 1, 1), indicating that enterprise and rescue coordination may persist while strengthened regulation is not self-sustaining. An illustrative closure-consistent parameterization restores E1/E8 bistability and yields PH(U([0, 1]3)) = 0.889, Δcrit(1, 1, 1) = 0.738, and Trec = 13.95 under the stated experiment. No locally asymptotically stable interior state arises in either the additive benchmark or the cross-product extension. Sensitivity checks further show that the numerical resilience values vary with the initial-state measure and disturbance direction. The findings suggest that collaborative resilience depends jointly on policy intensity, fiscal incidence, initial coordination conditions, and shock structure.