Hybrid Simulation in Flexible Job Shop Sequencing Systems

This research studies one of the most complex and important problems in the area of sequencing: flexible job shop systems. There is a current demand for the robustness of the model, which must be effective against unexpected everyday problems that affect the efficiency of the initial sequence defined. In addition, stochastic time and multi-objective search must be considered. Therefore, this research will perform the sequencing through simulation, which should sequence the orders at the beginning of the process and re-sequence when necessary. To do this, the model will need to use agents within Discrete Event Simulation, forming a Hybrid model. The objective is to evaluate how the agent can help in responding to these events, improving robustness and making it more accessible to users, bringing a friendly interface to insert modifications. Comparisons were made of the results obtained with and without re-sequencing. First, with the agent using only one logic and then with adjustments in the sequence during the production of the batches, seeking to respond to events. It is emphasized that the scheduling guarantees that the 'Manager' agent reduces the makespan and increases the utilization of the machines simultaneously.