求解多目标优化问题的改进布谷鸟搜索算法

Improved cuckoo search algorithm for multi-objective optimization problems

针对求解多目标优化问题,提出一种改进的多目标布谷鸟搜索算法(IMOCS).相比于标准多目标布谷鸟搜索算法(MOCS),IMOCS在莱维飞行中使用动态自适应的步长控制量,并基于层级和拥挤度距离选择下一次莱维飞行的种群.为了验证算法的有效性,通过在测试实例(SCH,ZDT系列,LZ)计算所求Pareto前沿与真实Pareto前沿的广义距离和所求Pareto前沿的多样性来测试IMOCS的性能.结果表明,与MOCS,NSPSO,NSGA-II比较,IMOCS所求的广义距离更小,即由IMOCS所求Pareto前沿更加接近于真实Pareto前沿,同时IMOCS的Pareto前沿分布更加广泛和均匀,即多样性更好.

In trying to solve multi-objective optimization problems,an improved multi-objective cuckoo search algorithm(IMOCS)was introduced.Compared with the standard multi-objective cuckoo search algorithm(MOCS),the IMOCS had two improvements,used a dynamic adaptive step-size control amount in Lévy flight;chose the next Lévy flight population based on the level and crowding distance.To verify the effectiveness of the IMOCS,test instances(SCH,ZDT series,LZ)were used to evaluate the performance:the generalized distance between the obtained Pareto front and the true Pareto front,and the diversity of the obtained Pareto front.Our results and comparison with the MOCS,NSPSO and NSGA-II showed that the IMOCS obtains a smaller generalized distance,which meant the IMOCSs Pareto front is closer to the true Pareto front,and at the same time its Pareto front distributes broader and more uniform,that diversity is better.