Certified supervisory recovery and scheduling for automated manufacturing systems under diagnosis delay

Fault-tolerant operation of automated manufacturing systems becomes a coupled supervisory-recovery and scheduling problem when production decisions must be taken before diagnosis has uniquely resolved the failed component. An action that is admissible at the current state can alter which reconfigurations remain reachable after diagnosis, while a locally feasible reconfiguration can lose controllability, nonblockingness, observation consistency, or resource feasibility after plant-wide composition. This paper develops a game-contract-schedule framework for time-weighted discrete-event systems (DESs) that preserves admissibility across these changes in decision representation. A recoverability game first operates on support-hypothesis states and retains pre-diagnosis controls from which every observation-consistent fault hypothesis keeps access to an admissible recovery. Once diagnosis resolves the relevant fault, assume-guarantee contracts instantiate fault-partner options and certify their composition, with concrete recovered-plant verification used when interface abstraction cannot establish the required reflection properties. Event-operation closure then extracts the executable subset of the accepted recovered language, and partial-order scheduling provides optimal or bounded-suboptimal timing solutions within that certified subset. Computational studies isolate the corresponding failure modes, quantify loss of diagnosis-time recoverability and local-to-global incompatibility, and evaluate locality-dependent contract repair and executable-language-constrained scheduling.