Modern manufacturing systems build on an effective scheduling scheme that makes full use of the system resource to increase the production, in which an important aspect is how to minimize the makespan for a certain production task (i.e., the time that elapses from the start of work to the end) in order to achieve the economic profit. This can be a difficult problem, especially when the production flow is complicated and production tasks may suddenly change. As a consequence, exact approaches are not able to schedule the production in a short time. In this paper, an adaptive scheduling algorithm is proposed to address the makespan minimization in the dynamic job shop scheduling problem. Instead of a linear order, the directed acyclic graph is used to represent the complex precedence constraints among operations in jobs. Inspired by the heterogeneous earliest finish time (HEFT) algorithm, the adaptive scheduling algorithm can make some fast adaptations on the fly to accommodate new jobs which continuously arrive in a manufacturing system. The performance of the proposed adaptive HEFT algorithm is compared with other state-of-the-art algorithms and further heuristic methods for minimizing the makespan. Extensive experimental results demonstrate the high efficiency of the proposed approach.

中文翻译：

---

用于处理灵活作业车间调度问题的动态作业自适应调度算法

现代制造系统建立在有效的调度方案之上，充分利用系统资源来增加产量，其中一个重要方面是如何最小化某个生产任务的完工时间（即从工作开始经过的时间）到最后）以达到经济利益。这可能是一个难题，尤其是当生产流程复杂且生产任务可能突然发生变化时。因此，精确的方法无法在短时间内安排生产。在本文中，提出了一种自适应调度算法来解决动态作业车间调度问题中的完工时间最小化问题。有向无环图不是线性顺序，而是用于表示作业中操作之间的复杂优先级约束。受异构最早完成时间 (HEFT) 算法的启发，自适应调度算法可以即时进行一些快速调整，以适应不断到达制造系统的新作业。将所提出的自适应 HEFT 算法的性能与其他最先进的算法和进一步的启发式方法进行比较，以最小化完工时间。大量的实验结果证明了所提出方法的高效率。