[研究生论坛]自适应多重Baldwin克隆选择算法

提出一种自适应多重Baldwin克隆选择算法(SAMBCSA)用以改善经典克隆选择算法(CLONALG)针对复杂多峰优化时精度不足的问题。通过学习不同抗体之间的优势差异信息引导免疫变异,挖掘潜在优势解区域方向,增强免疫进化能力。设计了免疫响应策略库,并引入基于统计的自适应学习框架以使算法可自适应选择不同的免疫响应策略应对不同的问题,增强算法普适性。分析了算法的时间复杂度,并说明了算法的收敛性。针对16个最新测试函数的仿真实验表明,对于多数复杂多峰优化问题,SAMBCSA的求解质量和收敛速度优于标准免疫克隆选择算法等已有算法。

Self Adaptive Multiple Baldwin Effect based Clonal Selection Algorithm

A multiple Baldwin effect based clonal selection algorithm (SAMBCSA) was proposed to deal with complex optimization problems. First, the ability of immune evolving was enhanced by the scheme of learning difference information between different antibodies to find the promising solution regions. Second, an immune response strategy pool and a statistical self adaptive learning framework were designed in order to improve the universality of the algorithm. Then, the time complexity and the convergence of the algorithm were analyzed. Finally, the experimental results on 16 state-of-the-art benchmark functions indicated that, compared with clonal selection algorithm and some other algorithms, SAMBCSA achieves good performance with high solution quality and good robustness on most problems.