一般克隆选择算法的概率性收敛研究

Investigation of convergence in probability for generic clonal selection algorithm

为了丰富克隆选择算法的理论研究并将遗传算法与克隆选择算法的收敛属性进行比较,采用与研究遗传算法类似的方法研究一般克隆选择算法概率性收敛属性,得到了克隆选择算法以一个预先定义的概率δ找到全局最优解的进化代数上界,该上界是独立于优化问题的。另外,在概率性收敛的情况下,得出了克隆选择算法与遗传算法的进化代数上界的比较结果以及相关结论。

In order to enrich theoretical research on CSAs and contrast convergence properties between GAs and CSAs, this paper investigated the properties of the convergence in probability for generic clonal selection algorithms in a similar manner as performed previously, in literature, for genetic algorithms. Found problem independent upper bounds for the number of genera- tions required to guarantee that the solution of a global optimum problem with a defined probability δ. Furthermore, under the condition of the convergence in probability, derived the comparison results and conclusions about the upper bounds of the number of generations for CSAs and GAs.