Abstract: In order to improve the global search capability of Immune Evolutionary Algorithm（IEA） and reduce its computational complexity, a multi-method collaborative optimization algorithm（MCIEA） is proposed. IEA is improved. Considering individual specificity, antibody population is divided into three subgroups：elite, ordinary, and inferior. By applying Gaussian mutation, uniform mutation, and regeneration to these subgroups respectively, the global search capability of the algorithm is enhanced. In the Pattern Search Method（PSM）, greedy descent is used instead of alternate operation as the move strategy of detection and pattern, which accelerates the convergence speed. PSM is embedded in the process of IEA as the local optimizer and uses the evolutionary information of IEA to set the parameters, e.g. initial point and steps, so as to realize the collaboration between IEA and PSM. Several classical test functions and a layout optimization example of satellite electronic equipment are used to verify the proposed method. The results show that the optimization capability and convergence efficiency of MCIEA are more advanced than IEA, and the computational complexity is decreased significantly.

Key words: multi-method collaboration, immune evolutionary algorithm, multi-population, pattern search method

摘要： 为提高免疫进化算法的全局寻优能力并降低计算复杂度，提出了多方法协作免疫进化算法。对免疫进化算法进行了改进。考虑抗体个体差异性，将抗体种群划分为精英、普通和劣等子群，对其分别执行高斯变异、均匀变异和消亡更新等差别化操作，增强了算法全局搜索能力。模式搜索法的探测和模式移动策略由单步交替改为贪婪下降，加快了算法收敛速度。将模式搜索法作为局部搜索工具嵌入免疫进化流程，同时采用免疫进化信息指导模式搜索法的初始点和参数设置，实现多方法协作优化。采用经典测试函数和某星载电子设备布局优化问题对算法进行了测试，测试结果表明算法寻优能力和收敛速度优于免疫进化算法，计算复杂度有显著下降。

关键词: 多方法协作, 免疫进化算法, 多种群, 模式搜索法