Abstract: The standard grey wolf optimization (GWO) algorithm has a few disadvantages of slow convergence, low solving precision and high possibility of being trapped in local optimum. To overcome these disadvantages of GWO algorithm, this paper proposes a hybrid GWO (HGWO) algorithm based on opposition-based learning strategy and Rosenbrock local search method. In the proposed hybrid algorithm, opposition-based learning strategy is introduced to generate initial population, which strengthens the diversity of population. Rosenbrock local search method is applied to the current best individual, which improves the convergence speed and local search ability of GWO algorithm. Elite opposition-based learning approach is used to avoid premature convergence of GWO algorithm. The experimental results of 6 well-known benchmark functions show that the proposed HGWO algorithm has strong convergence and high precision.

Key words: grey wolf optimization algorithm, Rosenbrock search, opposition-based learning

摘要： 针对标准灰狼优化（grey wolf optimization，GWO）算法存在后期收敛速度慢，求解精度不高，易出现早熟收敛现象等问题，提出了一种基于对立学习策略和Rosenbrock局部搜索的混合灰狼优化（hybrid GWO， HGWO）算法。该算法首先采用对立学习策略取代随机初始化生成初始种群，以保证群体的多样性；然后对当前群体中最优个体进行Rosenbrock局部搜索，以增强局部搜索能力和加快收敛速度；最后为了避免算法出现早熟收敛现象，利用精英对立学习方法产生精英对立个体。对6个标准测试函数进行仿真实验，并与其他算法进行比较，结果表明，HGWO算法收敛速度快，求解精度高。

关键词: 灰狼优化算法, Rosenbrock搜索, 对立学习