Towards Predictive and Reactive Job Shop Scheduling: A Hybrid Approach in Flexible Manufacturing Systems

In the contemporary manufacturing landscape, unpredictable shop-floor disturbances and shortened product life cycles significantly challenge traditional scheduling methods, often leading to missed due dates and inefficient resource utilization. This study addresses these challenges by proposing a robust predictive–reactive scheduling framework designed for flexible manufacturing systems. The methodology utilizes a hybrid simulation–optimization approach, integrating discrete-event simulation to map real-time system states with a genetic-algorithm-based optimizer to identify high-performing scheduling policies. The framework is uniquely capable of dynamically generating and evaluating alternative dispatch sequences, enabling rapid recovery from operational disruptions such as equipment failure or order changes with minimal deviation from planned workflows. To enhance optimization efficiency, the model incorporates a permutation-based path generator and complexity filters that constrain the search space for the genetic algorithm. Case-study experiments conducted within a reconfigurable manufacturing context demonstrate that the proposed approach significantly improves performance, reducing makespan by up to 23% in standard scheduling scenarios and by up to 38% during rescheduling events, while simultaneously maintaining high system utilization. These findings underscore the effectiveness of integrating simulation with heuristic optimization to provide responsive, real-time control in complex production environments.