【摘要】 针对基本遗传算法在解决实际最优化问题时可能存在的收敛速度慢、易于陷入局部最优解等问题,提出从适应度排序标定和混沌算子嵌入两方面进行算法性能的改进。前者避免了传统适应度值计算方法中较小目标函数值导致的具有过大适应度值的个体对种群进化方向的影响,使得种群始终能够保持恒定的进化压差促进最优解的搜索;后者则增强了遗传算法局部搜索的能力,从而提高了近似最优解向全局最优解转化的可能性。在此基础上,建立了一种基于混沌遗传算法的光纤布拉格光栅(FBG)轴向非均匀应变分布重构方法,仿真算例表明,混沌遗传算法有效改善了非均匀应变分布重构算法的收敛性能,提高了重构的精度。讨论了算法中相关参数的设置对非均匀应变分布重构精度的影响。更多还原

【Abstract】 Aiming at the problem of slowly convergence and tending to get into local best solution when conventional genetic algorithm is applied to solve practical optimization problems,the genetic algorithm are improved by fitness ranking scheme and embeding chaos operator into the algorithm.Fitness ranking scheme can avoid the influence of individuals with big fitness impact on the direction of population evolving and make the population keep a constant evolutionary pressure for searching the best solution;Chaos operator can enhance the ability of local searching of the genetic algorithm,thereby improve the possibility of transforming the approximate solution into the best solution.On this basis,a method based on chaos genetic algorithm is presented to reconstruct the axial non-uniform strain profile distribution of fiber Bragg grating.The simulation result shows that chaos genetic algorithm improves the convergence capability of non-uniform strain profile distribution reconstruction algorithm.Furthermore,the effect factors of the reconstruction accuracy are also discussed.更多还原