非线性多目标优化的和声分散搜索算法  

Harmony Scatter Search Algorithm for Nonlinear Multi-criteria Optimization

作　　者：刘思[1] 马良[1] 

机构地区：[1]上海理工大学管理学院,上海200093

出　　处：《小型微型计算机系统》2015年第7期1498-1503,共6页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(71401106)资助;上海市一流学科建设项目(S1201YLXK)资助;沪江基金项目(A14006)资助

摘　　要：针对带约束的高维非线性多目标优化问题,给出一种混合型的智能优化算法进行求解.在和声搜索算法搜索机制的基础上,通过吸收分枝定界的思想,加入了分散搜索算法的优化机制,由此改善了和声搜索算法求解多目标问题时效率不高、搜索方向有时盲目的不足,同时也增强了算法的全局搜索能力,丰富了解的多样性.算法在计算机上编程实现,并对DTLZ系列中的问题进行了求解测试,取得了在三目标问题下的Pareto最优面.通过解集的分散程度和间距等指标对结果进行了评估,并与已有算法的结果进行了对比,验证了所给算法在求解多目标优化问题上的优势.This paper proposed a hybrid intelligent optimization algorithm for high dimensional constrained non-linear multi-objective optimization problems. Based on the Harmony Search Algorithm and the idea of Branch and Bound strategy,the Scatter Search mechanism is used and modified to improve the effectiveness and efficiency of Harmony Search and strengthen the global searching ability and diversification of the algorithm. The algorithm is coded and implemented to test series of benchmark instances of DTLZ problems.The best Pareto frontier is reached for problem with 3 objectives and is compared with that of other results. The computational experiments showthe advantages of the proposed algorithm in solving the multi-criteria optimization problems.

关 键 词：多目标优化 和声搜索算法 分散搜索算法 PARETO最优 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]