基于分解机制的多目标蝙蝠算法

Multi-objective Bat Algorithm Based on Decomposition

在分析蝙蝠算法性能基础上,将蝙蝠算法融入分解机制,提出了一种基于分解机制的多目标蝙蝠算法。为了进一步提高算法的多样性,将差分进化策略引入算法中。对14个具有复杂Pareto前沿的多目标优化问题(LZ-09系列和ZDT系列)测试不同邻域规模对算法性能的影响,结果表明新算法的邻域规模为20时性能最优;将其与MOEA/D-DE和NSGA-II算法进行对比分析,结果显示该算法的分布性、收敛性和多样性均优于另外两种算法。为了验证其求解含有约束问题的性能,将其应用于滑动轴承多目标优化设计问题中,获得的Pareto前沿分布均匀,表明算法具有工程实用性,是求解复杂高维多目标问题的有效方法。

The bat algorithm was integrated into decomposition mechanism on the basis of its evaluation and a multi-objective bat algorithm based on decomposition（MOBA/D） was proposed.In order to improve the algorithm diversity,the differential evolutionary strategy was introduced into MOBA/D.The performances of MOBA/D on 14 multi-objective optimization problems were tested,which included family benchmark functions of LZ-09 and ZDT with different neighborhood scales effect on the performance of the algorithm.The result indicated that MOBA/D had the best performance with neighborhood size of 20.Compared with MOEA/D-DE and NSGA-II,the simulation results showed that MOBA/D can obtain a more uniform distribution of Pareto solution set and better convergence as well as diversity than those of state-of-the-art multi-objective metaheuristics.For further performance analysis of MOBA/D on constraint problem,the optimization design of sliding bearing was solved to demonstrate the feasibility and effectiveness.The good performance on convergence and diversity of the obtained Pareto set demonstrated that MOBA/D was suitable for engineering practice,which was an effective way for solving complex and high dimensional multi-objective optimization problems.