基于MapReduce模型带任务分割的平行机调度优化

研究一类基于MapReduce模型的两阶段平行机调度问题.该模型中的每个工件包含Map和Reduce两道工序,前一工序的任务可以划分并同步加工,而后一工序不可划分,结合工件的到达时间、交货时间等约束,以最大完工时间和总延迟时间的加权和作为优化目标构建混合整数规划模型,设计采用差分变异策略和逐维角度扰动机制的改进鲸鱼优化算法求解模型.数值仿真实验结果表明,所设计的算法相对于经典的鲸鱼优化算法、粒子群算法的求解效果有显著的提升,验证了模型和所设计算法的有效性.

Parallel machine scheduling with splitting jobs in MapReduce system

Based on the MapReduce model, a two-phase parallel machine scheduling problem is studied. In the model, each job consists of two operations named Map and Reduce. The Map operation can be split and processed simultaneously, while the Reduce shall be processed on a single machine. Considering the arrival time, and due date of each job, we establish a mixed integer linear programming (MILP) model, aiming at minimizing the weighted makespan and total tardiness. An improved whale optimization algorithm (IWOA) is proposed, which uses differential perturbation and dimension-by-dimension Levy perturbation to obtain a near-optimal solution. The numerical results show that the IWOA outperforms both the particle swarm optimization and the whale optimization algorithms for the considered problem.