论文部分内容阅读
采用高斯变异算子的进化规划算法存在早熟现象,根本原因是高斯变异产生的变异量较小,导致个体分量乃至整个个体不发生变异.文中从变异算子、个体分量值的计算和搜索空间三个方面改进了进化规划算法.设计了能产生较大变异量的离散余弦变换算子,并且采用动态比例变异法动态调整个体中的每个分量,多个体竞争策略扩大了算法的搜索空间.针对复杂采购业务模型,运用改进的进化规划算法求解.实验证明,改进的算法在求解精度上优于采用高斯变异和随机变异的进化规划算法,解决了进化规划算法的早熟问题.
The evolutionary planning algorithm using Gaussian mutation operator has premature phenomenon, the fundamental reason is that the variation caused by Gaussian mutation is small, which leads to no variation of individual component and even the whole individual.In this paper, from mutation operator, calculation of individual component value and search space This paper improves the algorithm of evolutionary programming and designs a discrete cosine transform operator that can generate a large amount of variation, and dynamically adjusts each component of the individual using a dynamic proportional-variation method, and multiple-body competition strategy expands the search space of the algorithm. The complex procurement business model is solved by the improved evolutionary programming algorithm.Experimental results show that the improved algorithm is superior to the evolutionary programming algorithm using Gaussian mutation and random variation in the solution accuracy and solves the premature problem of evolutionary programming algorithm.