求解多峰函数优化问题的免疫量子进化算法

Quantum Evolutionary Algorithm Based on Immune Theory for Multi-Modal Function Optimization

提出了一种求解多峰函数优化问题的免疫量子进化算法,该算法依据小生境机制将量子表达的初始种群划分为子群组,再对每个子群组利用免疫特性的局域搜索能力包括抗体的克隆选择、记忆细胞产生、免疫细胞交叉变异、抗体的促进与抑制等进化机制,找出局域最优解。最终算法可保持所有优化解。算法综合了量子计算的天然并行性和免疫算法的充分自适应性,它比传统的进化算法具有更好的种群多样性,更快的收敛速度,更有效的全局和局域寻优能力;证明了算法的收敛性,最后通过仿真实验表明了该算法的优越性。

A novel quantum evolutionary algorithm based immune mechanism （MIQEA） for solving function optimization containing multiple global optima was proposed. By niche methods the original population was divided into subpopulations automatically, and then local search was carried by the immune mechanism in which antibody can be clone selected, immune cell can accomplish cross--mutation, memory cells can be produced and similar antibodies can be suppressed for all subpopulations, each subpopulation can obtain optimal solutions. The algorithm can maintain all optimal solutions. The quantum evolutionary algorithm with intrinsic parallelism is integrated with adaptive immune dynamic model, it not only can maintain quite nicely the population diversity than the classical evolutionary algorithm, but also can help to accelerate the convergence speed and has been able to get the global optimal and sub--optimal solutions rapidly. The convergence of the MIQEA was proved; its superiority is shown by some simulation experiments.