一种结合次梯度的粒子群全局优化算法
为优化不可微且非凸的连续目标函数,提出了结合次梯度的粒子群全局优化算法（SGPSO）。在优化算法中,首次提出利用次梯度方向来更新粒子群算法中粒子的搜索速度方案。加上与粒子相互间的通信机制配合,改进方案提高了寻得全局最优的机率。进一步地,在次梯度迭代过程中,提出其中的步长函数需要满足关于次梯度幅值是低阶无穷小且关于迭代时刻是递减的充分条件保证序列稳定收敛。最后,针对标准库给出了SGPSO的实验和比较以验证其有效性,结果表明提出的算法能很好地实现目标函数的全局优化,且收敛效果更好。...
Saved in:
| Published in | 计算机应用研究 Vol. 32; no. 4; pp. 1007 - 1010 |
|---|---|
| Main Author | |
| Format | Journal Article |
| Language | Chinese |
| Published |
深圳信息职业技术学院软件学院,广东深圳518172%厦门大学信息科学与技术学院,福建厦门,361005
2015
深圳市可视媒体处理与传输重点实验室,广东深圳518172 |
| Subjects | |
| Online Access | Get full text |
| ISSN | 1001-3695 |
| DOI | 10.3969/j.issn.1001-3695.2015.04.011 |
Cover
| Summary: | 为优化不可微且非凸的连续目标函数,提出了结合次梯度的粒子群全局优化算法（SGPSO）。在优化算法中,首次提出利用次梯度方向来更新粒子群算法中粒子的搜索速度方案。加上与粒子相互间的通信机制配合,改进方案提高了寻得全局最优的机率。进一步地,在次梯度迭代过程中,提出其中的步长函数需要满足关于次梯度幅值是低阶无穷小且关于迭代时刻是递减的充分条件保证序列稳定收敛。最后,针对标准库给出了SGPSO的实验和比较以验证其有效性,结果表明提出的算法能很好地实现目标函数的全局优化,且收敛效果更好。 |
|---|---|
| Bibliography: | 51-1196/TP This paper proposed an approach of subgradient integrated into particle swarm optimizer（ SGPSO） for globally optimizing continuous objective function. In minimization,it proposed a revision for the manner of velocity update with the direction of subgradient to search for the local minima of a given non-differentiable and non-convex objective function. Thus,it combined with communications among particles,this revision would offer more chances to obtain the global minima. Furthermore,in the part of subgradient iteration,it suggested that the step function should be a lower order infinitesimal with respect to subgradient magnitude as well as be a decreasing function with respect to iteration time. In the end,experiments and comparisons of the proposed SGPSO on benchmark problems validate its performance with better effectiveness and efficiency. XU Zhi-liang, ZENG De-lu, ZHANG Yun-sheng （ 1. Shenzhen Key Laboratory of Visual Media Processing ＆ Transmission, Shenzhen Guangdong 518172, China ; 2. School of |
| ISSN: | 1001-3695 |
| DOI: | 10.3969/j.issn.1001-3695.2015.04.011 |