An adaptive multi-population differential artificial bee colony algorithm for many-objective service composition in cloud manufacturing

Several conflicting criteria must be optimized simultaneously during the service composition and optimal selection (SCOS) in cloud manufacturing, among which tradeoff optimization regarding the quality of the composite services is a key issue in successful implementation of manufacturing tasks. This study improves the artificial bee colony (ABC) algorithm by introducing a synergetic mechanism for food source perturbation, a new diversity maintenance strategy, and a novel computing resources allocation scheme to handle complicated many-objective SCOS problems. Specifically, differential evolution (DE) operators with distinct search behaviors are integrated into the ABC updating equation to enhance the level of information exchange between the foraging bees, and the control parameters for reproduction operators are adapted independently. Meanwhile, a scalarization based approach with active diversity promotion is used to enhance the selection pressure. In this proposal, multiple size adjustable subpopulations evolve with distinct reproduction operators according to the utility of the generating offspring so that more computational resources will be allocated to the better performing reproduction operators. Experiments for addressing benchmark test instances and SCOS problems indicate that the proposed algorithm has a competitive performance and scalability behavior compared with contesting algorithms.