Abstract: Analyzing the development and exploration performance of the standard Grey Wolf Optimizer（GWO）, a hybrid grey wolf optimizer（MAR-GWO） is proposed, the search domain is fully extended. In MAR-GWO, the initiative searching operators are introduced for Wolf Alfa, Beta, and Delta respectively to advance more on effect and speed, the eliminating-reconstructing mechanism is applied for poor search wolves to improve the search efficiency and the mutation with a given probability is adopted for excellent search wolves to increase the individual diversity and to protect it against the local optimum. To verify the validity of MAR-GWO, the experiment about 13 global optimization problems are done, and GWO, GWO-EPD（another improved GWO）, PSO, and EA are involved as comparisons. The result shows its success rate is greater, its convergence speed is faster, and it is not easy to fall into the local optimum, meaning MAR-GWO has a strong competitiveness in intelligent algorithms.

Key words: grey wolf optimizer, initiative search, eliminating-reconstructing mechanism, mutation

摘要： 在分析标准苍狼优化算法（GWO）的开发与探索性能基础上，提出了一种混合苍狼优化算法（MAR- GWO），搜索域得到了全面的扩展，其中针对[α、][β、][δ]领导层苍狼，引入自主搜索行为来加大其优化力度与促进速度的提高，对性能较差搜索狼采取淘汰重组机制以提高搜索效率，又采取概率差分变异行为增加了个体多样性，从而避免局部最优。为了验证MAR-GWO算法有效性，对13个全局优化问题进行实验，分别与GWO、GWO-EPD（改进的苍狼优化算法）、PSO、EA等算法进行了对比测试，从实验结果来看，MAR-GWO算法寻优成功率相对较高、收敛速度快，不易陷入局部最优，在智能算法中具有很强的竞争力。

关键词: 苍狼优化算法, 自主搜索, 淘汰重组机制, 变异