Hybrid Simulation and Multi-Agent Decision Support for Bottleneck Optimization in Metallurgical Production

This study proposes a hybrid simulation and multi-agent decision support framework aimed at optimizing complex resource conversion processes in industrial and organizational systems. Focusing on metallurgical production logistics, we integrate discrete event simulation, agent-based modeling, and critical path analysis to address bottlenecks caused by shared resource conflicts. The proposed method is implemented in two systems: the BPsim modeling suite and the Automated Metallurgical Control System. For constructing the multi-agent model system of resource conversion processes, this paper adopts Petri nets, queuing theory, system dynamics, and hybrid automata theory. The main object of this model is the resource converter, whose primary function is to transform inputs (resources) into outputs (products) using discrete or continuous operations. To address resource occupation and bottleneck elimination issues in metallurgical processes, we employ operational analysis of probabilistic networks, critical path method, network planning methods, and agent-based planning theory. Combining operational analysis, agent-based planning, and real-time simulation hybrid bottleneck mitigation methods. The results show that a converter feeding interval of 20 minutes maximizes continuous casting machine utilization while preventing steel solidification delays. The framework's adaptability can also be extended to supply chains, production planning, and other resource-constrained systems.