【摘要】 将学习机制引入到变异算子中,定义了个体距离、相似性和邻域等概念,用距离反映个体间的差异程度,用相似性描述个体间对应基因位的类似程度,用邻域实现对种群按相似性分割。提出了基于相似性学习的自适应演化算法,从而使得变异算子具有了很强的导向性,避免了传统达尔文演化策略的半盲目性,使计算结果稳定地收敛到全局最优解。以下料问题为例,对算法进行数值试验,运算结果表明该算法能很好求解整数规划问题。更多还原

【Abstract】 A learning mechanism is introduced to evolutionary computation. The individual distance, comparability and neighborhoods are defined. The distance can reflect the difference between individuals, comparability is designed to reflect how close two individuals are, and neighborhoods are used to realize the division of population. A new adaptive evolutionary algorithm based on comparability learning featuring several characteristics is proposed. Firstly, this algorithm is capable for mutation operators to acquire insight jumps of the fitness. Secondly, it can avoid the semi-blind of conventional Darwinian-type evolutionary computation. Thirdly, it ensures a stable convergence of the algorithm into a global optimum. The optimal cutting problem is studied as an example to analyze the performance of this algorithm. The experimental results prove the advantages of this algorithm.更多还原