Integrated Energy-Efficient Scheduling of Distributed Flow Shop and Distribution Using a Q-Learning-Based Multi-Objective Cooperation Evolutionary Algorithm

Production and distribution are indispensable in a supply chain. Practically, finding an overall optimization of the two processes has become an urgent problem. This article addresses an integrated energy-efficient scheduling problem of distributed flow shop and distribution in the context of economic globalization and environment sustainability. First, a mixed integer programing model is formulized to achieve minimal maximum completion time, minimal total energy consumption and minimal maximum workload of factories. Second, a Q-learning-based multi-objective cooperation evolutionary algorithm with problem-specific knowledge is specially developed. A cooperation strategy is devised to search a solution space using three populations. Three heuristic rules are proposed to initialize the populations. A knowledge-based local search is devised to refine better individuals. A Q-learning-based approach is designed to adaptively choose premium search strategies at each iteration. Finally, the designed method is compared with three meta-heuristics in prior research and a mathematical programming solver CPLEX. The comparison experiments are conducted by using a group of benchmark instances and a real-world case. The outcomes confirm outstanding capacities of the developed model and algorithm for dealing with the considered problems.