Solution algorithms and heuristics for an energy efficient hybrid flowshop problem

The files contain psuedocodes of the main algorithm, the hyperheuristics, metaheuristics, and the modified exact algorithm for the scheduling problem. The main algorithm first decides on how the choice of machines at the stations with parallel processors would be made, before it decides the job orders followed in every stage. The hyperheuristic consists of six low level heuristics (LLH) which are combined with the metaheuristc by Nawaz, Enscore and Ham (NEH) and GA to form the Improved Hyper heuristic NEH (IHNEH) algorithm, and the Improved Hyper heuristic GA (IHGA) algorithm. Each of these two algorithms operate in three stages and are improved using the local search heurisitc, and share the first step, which is where the hyper heuristic is used to select a low-level heuristic for implementation. The second step makes use of the selected LLH and processing time vector derived thereby to create long replication cycles for the problem instances created, consisting of the combination of each sequencing rule (IHNEH and IHGA) and job size, followed by the neighbourhood search algorithms. The final step of the algorithm tests the effectiveness of the two solutions against the exact algorithm called the Branch and Bound (B&B) algorithm in terms of the value of makespan returned, the energy consumption level, and the running time taken to obtain the solutions. The first part of the datasets contains figures and tables of the average makespan, energy consumptions, and running times of the algorithms. The last part contains figures of the convergence of the genetic algorithm and the energy threshold reduction factors to validate. the choice of the particular parameter values over others in developing the algorithms

本数据集包含调度问题的主算法、超启发式、元启发式以及改进的精确算法的伪代码。主算法首先确定并行处理器的车站如何选择机器，然后决定各阶段的作业顺序。超启发式由六个低级启发式（LLH）组成，这些启发式与Nawaz、Enscore和Ham（NEH）的元启发式及遗传算法（GA）相结合，形成了改进的超启发式NEH（IHNEH）算法和改进的超启发式GA（IHGA）算法。这两种算法均分三阶段运行，并采用局部搜索启发式进行优化，且共享第一步，即使用超启发式选择一个低级启发式进行实施。第二步利用选定的LLH和处理时间向量，为生成的调度实例创建长复制周期，这些实例由每个排序规则（IHNEH和IHGA）与作业规模组合而成，随后进行邻域搜索算法。算法的最终步骤通过测试两种解决方案相对于分支定界（B&B）算法的精确度，即返回的最大完工时间、能耗水平和获取解决方案所需运行时间，来检验其有效性。数据集的前部分包含算法的平均最大完工时间、能耗和运行时间的图表和表格。最后一部分包含遗传算法收敛图和能量阈值降低因子图，以验证在算法开发过程中特定参数值相较于其他值的合理性。