多目标快速群搜索优化算法及在模型修正中的应用  

Multi-objective quick group search optimization algorithm and its application in model updating

作　　者：李世龙[1] 马立元[1] 李永军[1] 王天辉[1] 

机构地区：[1]军械工程学院导弹工程系,石家庄050003

出　　处：《振动与冲击》2015年第20期120-128,共9页Journal of Vibration and Shock

基　　金：军队科研资助项目([2012]80)

摘　　要：为解决群搜索算法在求解多目标优化问题时易陷于局部最优或过早收敛,限制其在复杂结构模型修正中的应用问题,提出改进的群搜索优化算法-多目标快速群搜索优化算法(MQGSO)。采用LPS搜索方法对发现者进行迭代更新,能使发现者更快到达最优位置,提升寻优效率;对追随者增加速度更新机制,考虑其自身历史最优信息以保证收敛精度,并在算法后期采用交叉变异策略增加追随者个体多样性,避免陷入局部最优;在游荡者迭代更新中引入分量变异控制策略,增加其搜索的随机性,提高算法的全局寻优性能。通过7个典型多目标优化测试函数及某发射台有限元模型修正实例,对算法性能进行验证分析。结果表明,与已有MPSO(Multi-objective Particle Swarm Optimization)及MBFO(Multi-objective Bacterial Foraging Optimization)两种算法相比,所提MQGSO算法搜索性能更强、收敛速度更快、计算精度更高,不失为求解复杂多目标优化问题的有效方法。Model updating is a typical multi-objective optimization problem. Comparing with other traditional swarm intelligent algorithms,the group search optimization（ GSO） algorithm has outstanding performance on solving complex optimization problems. However,it is not free from the problems such as being entrapped into a local optima and premature convergence when it is utilized in multi-objective optimization. A novel multi-objective quick group search optimizer by the name of MQGSO was proposed for solving these problems. To simplify the computation,the LPS method was introduced in updating producers. Meanwhile,the particle swarm optimization（ PSO） evolutionary strategy was adopted in updating scroungers to improve the convergence accuracy. Moreover,to restrict fallng into local optima,a crossover and mutation operation was introduced to increase the diversity of scroungers in each iteration. In addition,mutation probability,which can increase the randomicity of rangers,was introduced to enhance the global searching capability. Seven multi-objective minimization benchmark functions and a model updating case of a certain launch platform were used to evaluate the proposed MQGSO against MPSO and MBFO algorithms. The calculation results show that the MQGSO has a preferable convergence rate and accuracy,and it is an effective method for multi-objective optimization.

关 键 词：多目标优化 群搜索算法 PARETO最优解 模型修正 目标函数 

分 类 号：TB123[理学—工程力学] TU311[理学—力学]