This paper develops a multi-operator based differential evolution with a communication strategy (MCDE) being integrated with a sequential Tabu Search (MCDE/TS) to solve the job shop scheduling problem (JSSP) with the objective of minimizing makespan. The three variants of DE which are implemented in the proposed algorithm evolve as independent sub-populations, which relate to a communication strategy that maintains the diversity and quality of each sub-population by employing a proposed mixed selection strategy to avoid premature convergence. The best solution order obtained from MCDE is then passed to Tabu Search (TS) and the evolution process is continued, creating neighbour solutions with N7 neighbourhood structure. This algorithm ensures the population diversity with curving the premature convergence but experiences faster convergence. The design of experiment for parameter tuning is employed for the best combination of the proposed algorithm’s parameter. The performance of the proposed MCDE/TS algorithm is evaluated against a number of state-of-the-art algorithms to show its competence in solving 122 standard benchmark instances.

本文开发了一种基于多操作员的差分进化，将通信策略（MCDE）与顺序禁忌搜索（MCDE / TS）集成在一起，以解决工期调度问题（JSSP），目的是最大程度地缩短工期。在提出的算法中实现的DE的三个变体演变为独立的子种群，这涉及一种通信策略，该通信策略通过采用提议的混合选择策略来避免过早收敛，从而保持每个子种群的多样性和质量。然后将从MCDE获得的最佳解决方案顺序传递给Tabu Search（TS），并继续进行演化过程，从而创建具有N7邻域结构的邻居解决方案。该算法通过消除过早收敛来确保种群多样性，但收敛速度更快。为了优化所提出算法的参数，采用了参数调整实验设计。针对众多最新算法对所提出的MCDE / TS算法的性能进行了评估，以显示其在解决122个标准基准实例方面的能力。