Dynamic flexible job shop scheduling algorithm for multi-task collaborative optimization

Due to customer demand or market changes, production orders in intelligent manufacturing workshops become uncertain. Based on the above issues, an order detection framework is constructed to detect different types of order changes. Different dynamic response mechanisms are designed for different types of order changes. The scheduling of jobs in discrete manufacturing has composability, resulting in a huge search space. Previous research methods that used a single population to solve scheduling schemes could not fully explore the search space. Considering the characteristics of multitasking in job shop scheduling, this study designs an auxiliary task collaborative optimization algorithm (ATCOA) to solve the optimal rescheduling schemes. To escape from the situation of optimizing local optima in the main task, a knowledge transfer probability model based on the main task is adopted to determine population communication between tasks. A multitask knowledge transfer strategy is proposed for exchanging individual information between tasks to improve the diversity distribution ability of optimization algorithm. To evaluate the effectiveness of the ATCOA algorithm, it is compared with other algorithms on the constructed dynamic order test cases. In the case of order cancellation and insertion, ATCOA obtained 10 minimum inverted generation distance and maximum spread metric values and 9 hypervolume values, respectively. ATCOA has improved completion efficiency by 8.9% compared to scheduling rules. In engineering simulation cases, the ATCOA algorithm improved workload deviation by 41.9% compared to other algorithms. The experimental results show that the ATCOA algorithm is more efficient and stable than other algorithms.