计算机工程与应用 ›› 2018, Vol. 54 ›› Issue (5): 156-164.DOI: 10.3778/j.issn.1002-8331.1707-0211

• 模式识别与人工智能 • 上一篇    下一篇

多目标灰狼优化算法的改进策略研究

崔明朗,杜海文,魏政磊,李  聪   

  1. 空军工程大学 航空航天工程学院,西安 710038
  • 出版日期:2018-03-01 发布日期:2018-03-13

Research on improved strategy for multi-objective grey wolf optimizer

CUI Minglang, DU Haiwen, WEI Zhenglei, LI Cong   

  1. College of Aeronautics and Astronautics, Air Force Engineering University, Xi’an 710038, China
  • Online:2018-03-01 Published:2018-03-13

摘要: 为了解决多目标灰狼优化算法(MOGWO)易陷入局部最优,稳定性差等缺点,基于对算法寻优时灰狼个体运动情况的分析,提出了两条改进策略:一是通过引入“观察”策略赋予灰狼个体自主探索的能力,以提高算法的优化效率和跳出局部最优的能力;二是改进控制参数调整策略,选用幂函数取代线性函数以提高算法的稳定性。然后对两条改进策略进行了可行性分析,提出了带观察策略的多目标灰狼算法并进行了算法复杂度分析。最后通过对6个不同特点测试函数的多次重复实验,结合GD与IGD两种通用评价指标,对原算法、改进后算法和多目标粒子群算法进行比较,从算法效率、寻优能力和稳定性等方面综合验证了算法改进的有效性和优越性。

关键词: 多目标灰狼算法, 观察策略, 控制参数, Pareto边界, 多目标优化评价方法

Abstract: For the problems of easily falling into local optimum and poor stability of the Multi-Objective Grey Wolf Optimizer(MOGWO), two improvement strategies are put forward by studying the movement of grey wolf individual at algorithm optimization process: One is adding the “survey process”, the grey wolf individual is endowed with the ability to explore independently and both the efficiency of algorithm and the ability of jumping out the local optimum solution are improved; the other is improving the adjustment strategy of control parameter. The power function is used to replace the linear function to improve the stability of the algorithm. Based on two universal evaluation methods of multi-objective optimization(Generational Distance and Inverted Generational Distance), 6 different test functions and 3 different algorithms(the original algorithm, the improved algorithm and the Multi-Objective Particle Swarm Optimization algorithm) are compared with the repeat experiments. The experimental results show the effectiveness and feasibility of the AS-MOGWO from efficiency, ability and stability.

Key words: Multi-Objective Grey Wolf Optimizer(MOGWO), survey strategy, control parameter, pareto optimal front, evaluation method of multi-objective optimization