This paper presents a new generic architecture for the implementation of an intelligent simulation-based real-time control in large-scale discrete-events systems. The structure of the proposed control methodology is mainly inspired from the model-based predictive control scheme. Fundamental characteristics of real-time applications, timeliness, concurrency and reactiveness, are clearly specified and embedded to ensure the conformity of our online control architecture with the real-time systems standards. The feasibility of this new architecture is demonstrated through a complete implementation of a truck control system in an emulated mine transportation environment. In the experimentations that were conducted, results showed the capability of the provided closed-loop control in dynamic environment to achieve efficient real-time truck dispatching. We also present the used methodology that allows the integration of the intelligent metaheuristics search for an effective real-time fleet management, even under operational time constraints. The novelty of this work lays in the usage of concepts from computer science and software engineering in complex real-life problems generally addressed by the operation research community.