The Research And Application Of Multi-Objective Artificial Bee Colony Algorithm

In real life, many optimization questions are multi-objective and have NP-hard, so it is very significance how to quickly obtain optimal or quite satisfied solutions to improve productivity and promote social development. Due to the deterministic heuristic algorithm is overly dependent on the structure of the problem itself, and sometimes generate unreasonable solution or the solution’s quality is not ideal, so the current research focusing on the multi-objective optimization is mainly heuristic algorithms. As the recently proposed swarm intelligence algorithm, which is derved from the simulation of the mechanism of labor division and the foraging behavior of the colony, the artificial bee colony (ABC) algorithm has been applied in many optimization questions and archived good results. But the algorithm is now mainly used for single objective optimization problem, and its research for multi-objective optimization problem is just getting started. In this paper, an overall framework of multi-objective artificial bee colony algorithm is presented first. Then, different strategies for the various components of the framework are designed and nine multi-objective artificial bee colony algorithms are designed based on this framework. Finally, the multi-objective artificial bee colony algorithms are applied to three multi-objective optimization problems. the function optimization problem, the QoS based wireless network routing optimization problem and the QoS based service selection problem. In order to evaluate these algorithms, in the experimental part, we define a number of indicators to evaluate the performance of the algorithms from all aspects, we do convergence analysis, parameter adjustment, local optimization strategies on the performance of the algorithms, statistical analysis, etc. Then they are compared with other recently proposed related algorithms for these problems. The experimental results strongly suggested that the multi-objective artificial bee colony algorithms proposed in this paper can effectively solve the three problems.