Effective constrained optimization algorithms have been proposed for engineering problems recently.It is common to consider constraint violation and optimization algorithm as two separate parts.In this study,a pbest s...Effective constrained optimization algorithms have been proposed for engineering problems recently.It is common to consider constraint violation and optimization algorithm as two separate parts.In this study,a pbest selection mechanism is proposed to integrate the current mutation strategy in constrained optimization problems.Based on the improved pbest selection method,an adaptive differential evolution approach is proposed,which helps the population jump out of the infeasible region.If all the individuals are infeasible,the top 5%of infeasible individuals are selected.In addition,a modified truncatedε-level method is proposed to avoid trapping in infeasible regions.The proposed adaptive differential evolution approach with an improvedεconstraint processmechanism(IεJADE)is examined on CEC 2006 and CEC 2010 constrained benchmark function series.Besides,a standard IEEE-30 bus test system is studied on the efficiency of the IεJADE.The numerical analysis verifies the IεJADE algorithm is effective in comparisonwith other effective algorithms.展开更多
针对常规最大类间方差法在多阈值图像分割中存在的运算量大、计算时间长、分割精度较低等问题,该文提出一种基于改进的自适应差分演化(JADE)算法的2维Otsu多阈值分割法。首先,为增强初始化种群的质量、提升控制参数的适应性,将混沌映射...针对常规最大类间方差法在多阈值图像分割中存在的运算量大、计算时间长、分割精度较低等问题,该文提出一种基于改进的自适应差分演化(JADE)算法的2维Otsu多阈值分割法。首先,为增强初始化种群的质量、提升控制参数的适应性,将混沌映射机制融入到JADE算法中;进而,通过该改进算法求解2维Otsu多阈值图像的最佳分割阈值;最终,将该算法与差分进化(DE), JADE,改进正弦参数自适应的差分进化(LSHADE-cn Ep Sin)以及增强的适应性微分变换差分进化(EFADE) 4种算法的2维Otsu多阈值图像分割进行比较。实验结果表明,与其它4种算法相比,基于改进JADE算法的2维Otsu多阈值图像分割在分割速度以及精度上均有较明显的改善。展开更多
将智能算法和数值迭代法相结合,构造一种组合式算法——自适应差分进化算法-Newton迭代(Adaptive differential evolution and Newton iteration,ADE-Newton)算法。以3-RPS并联机构为研究对象,详细阐述利用ADE-Newton算法求并联机构位...将智能算法和数值迭代法相结合,构造一种组合式算法——自适应差分进化算法-Newton迭代(Adaptive differential evolution and Newton iteration,ADE-Newton)算法。以3-RPS并联机构为研究对象,详细阐述利用ADE-Newton算法求并联机构位置正解的原理和步骤。为了验证ADENewton算法的有效性和通用性,给出3-RPS并联机构在不同驱动杆长条件下的数值算例。仿真结果表明,ADE-Newton算法能够以较高的效率求得不同驱动杆长条件下的全部高精度位置正解。还比较了ADE-Newton、人工蜂群、粒子群和差分进化算法求3-RPS并联机构位置正解的性能,比较结果显示,ADE-Newton算法的计算效率、收敛精度、稳健性以及可靠性高于对比算法。展开更多
基金supported by National Natural Science Foundation of China under Grant Nos.52005447,72271222,71371170,71871203,L1924063Zhejiang Provincial Natural Science Foundation of China underGrant No.LQ21E050014Foundation of Zhejiang Education Committee under Grant No.Y201840056.
文摘Effective constrained optimization algorithms have been proposed for engineering problems recently.It is common to consider constraint violation and optimization algorithm as two separate parts.In this study,a pbest selection mechanism is proposed to integrate the current mutation strategy in constrained optimization problems.Based on the improved pbest selection method,an adaptive differential evolution approach is proposed,which helps the population jump out of the infeasible region.If all the individuals are infeasible,the top 5%of infeasible individuals are selected.In addition,a modified truncatedε-level method is proposed to avoid trapping in infeasible regions.The proposed adaptive differential evolution approach with an improvedεconstraint processmechanism(IεJADE)is examined on CEC 2006 and CEC 2010 constrained benchmark function series.Besides,a standard IEEE-30 bus test system is studied on the efficiency of the IεJADE.The numerical analysis verifies the IεJADE algorithm is effective in comparisonwith other effective algorithms.
文摘针对常规最大类间方差法在多阈值图像分割中存在的运算量大、计算时间长、分割精度较低等问题,该文提出一种基于改进的自适应差分演化(JADE)算法的2维Otsu多阈值分割法。首先,为增强初始化种群的质量、提升控制参数的适应性,将混沌映射机制融入到JADE算法中;进而,通过该改进算法求解2维Otsu多阈值图像的最佳分割阈值;最终,将该算法与差分进化(DE), JADE,改进正弦参数自适应的差分进化(LSHADE-cn Ep Sin)以及增强的适应性微分变换差分进化(EFADE) 4种算法的2维Otsu多阈值图像分割进行比较。实验结果表明,与其它4种算法相比,基于改进JADE算法的2维Otsu多阈值图像分割在分割速度以及精度上均有较明显的改善。
文摘将智能算法和数值迭代法相结合,构造一种组合式算法——自适应差分进化算法-Newton迭代(Adaptive differential evolution and Newton iteration,ADE-Newton)算法。以3-RPS并联机构为研究对象,详细阐述利用ADE-Newton算法求并联机构位置正解的原理和步骤。为了验证ADENewton算法的有效性和通用性,给出3-RPS并联机构在不同驱动杆长条件下的数值算例。仿真结果表明,ADE-Newton算法能够以较高的效率求得不同驱动杆长条件下的全部高精度位置正解。还比较了ADE-Newton、人工蜂群、粒子群和差分进化算法求3-RPS并联机构位置正解的性能,比较结果显示,ADE-Newton算法的计算效率、收敛精度、稳健性以及可靠性高于对比算法。
