摘要
针对传统遗传算法的不足,提出一种基于离散系统状态空间模型的实数编码遗传算法(RGABS)。突破传统遗传算法的计算模式,将问题的求解过程表示为离散系统状态空间模型的动力学过程,通过构造遗传算子矩阵来确定搜索方向,通过选种池的选择操作体现优胜劣汰的自然选择机制,通过评估遗传算子矩阵的范数来考察算法的收敛性和收敛速度,从而改善算法性能。给出RGABS的计算流程及遗传算子矩阵应满足的条件,分析了遗传算子矩阵和选种池选择操作的作用及算法的收敛性和收敛速度。仿真实验结果表明:RGABS能够避免陷入局部最优,具有计算精度和计算效率高等特点。
In view of the shortcomings of traditional genetic algorithm,a real-coded genetic algorithm based on the state-space（RGABS）model of discrete system is presented.RGABS breaks through the calculation mode of traditional genetic algorithms and improves the computational efficiency by expressing the solving process as a dynamic process of the state-space model of discrete system,guiding the search direction by constructing genetic operator matrix,reflecting the natural selection of the fittest machine system by selection operation,and evaluating its convergence by matrix norm of genetic operator matrix.The calculation process of RGABS and the conditions of the genetic operator matrix are given,and the actions of genetic operator matrix and selection operation as well as the convergence of RGABS are analyzed.The simulation results show that by avoiding being trapped in the local optimum,RGABS has the characteristics of higher precision and higher speed.
出处
《山东科技大学学报（自然科学版）》
CAS
2015年第3期1-7,共7页
Journal of Shandong University of Science and Technology(Natural Science)
基金
国家自然科学基金项目(61473048)
关键词
状态空间模型
遗传算法
遗传算子矩阵
选择操作
收敛性
state-space model
genetic algorithm
genetic operator matrix
selection operation
convergence
