一种改进的多目标正余弦优化算法  

Improved Multi-objective Sine Cosine Optimization Algorithm

作　　者：王万良[1] 李伟琨[1] 王宇乐[1] 王铮 WANG Wan-liang;LI Wei-kun;WANG Yu-le;WANG Zheng(Key Laboratory of Big Data,Vision Institute,College of Computer Science,Zhejiang University of Technology,Hangzhou 310023,China)

机构地区：[1]浙江工业大学计算机学院视觉研究所大数据重点实验室

出　　处：《小型微型计算机系统》2019年第10期2102-2108,共7页Journal of Chinese Computer Systems

基　　金：国家自然基金面上项目(51875524,61873240)资助

摘　　要：随着技术的不断发展,设备的不断更新,现实中的优化问题逐渐从简单的单目标问题向复杂的多目标问题转换,这就需要算法可以高效地搜索空间从而获得最佳的解决方案.鉴于此,提出一种改进的多目标正余弦优化算法.该算法继承了原有单目标算法的优良性能,此外,结合反向学习机制提出了一种全新的初始化方法来代替原有的随机初始化方法.并通过引入网格坐标,提出一种基于密度筛选的新机制来有效的搜索空间,从而使算法获得更为精确的解决方案.最后,该算法通过与其他五种流行多目标算法在一系列基准测试函数上进行对比实验来验证其良好性能,此外,该算法还与4个流行算法在实际的工业优化设计问题上进了验证与分析,结果表明该算法不论是在测试函数上还是在处理该实际问题上都具有良好的性能与潜力.With the continuous development of technology and the update of equipment,the optimization problem in reality gradually transforms from the simple single-objective problem to the complex multi-objective problem,which requires the algorithm to search the space efficiently to obtain the best solution. In viewof this,an improved multi-objective sines cosines optimization algorithm is proposed. This algorithm inherits the excellent performance of the original single-object algorithm,and integrates two new mechanisms of opposition-based learning and grid density selection to effectively search space,thus obtaining more accurate solutions. In addition,the algorithm is compared with other five popular multi-objective algorithms in a series of benchmark functions to verify its good performance. Finally,the algorithm is verified and analyzed in four popular algorithms in the actual industrial optimization design problem. The results show that the algorithm has good performance and potential both in the test function and in dealing with this actual problem.

关 键 词：正余弦算法 多目标优化 工业问题 反向学习 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]