分数阶Fourier变换极值搜索算法研究  

Study of Algorithm for Extremum Seeking in the Fractional Fourier Transform

作　　者：卫红凯[1] 王平波[1] 蔡志明[1] 姚万军[2] 

机构地区：[1]海军工程大学电子工程学院,湖北武汉430033 [2]海军兵种指挥学院,广东广州510430

出　　处：《电子学报》2010年第12期2949-2952,共4页Acta Electronica Sinica

基　　金：国防973项目(No.613660202);中国博士后科学基金项目(No.200902668;No.20090451532)

摘　　要：针对步进式搜索算法在分数阶Fourier变换域二维平面效率、精度低下的问题,通过推导分数阶Fourier域极值函数的一阶导数及其快速算法,将全局寻优效果好的混沌优化法和局部搜索能力强的多步拟牛顿法相结合,提出了分数阶Fourier变换极值混合优化算法.并对混合算法的全局收敛性进行了证明.最后,通过仿真实例,验证了混合算法的全局收敛性及其快速收敛能力.混合优化算法的收敛速度和精度均好于步进式搜索法、混沌优化法及步进式-多步拟牛顿混合法.In order to overcome the problem of step by step searching method in the two-dimensional Fractional Fourier domain, a hybrid algorithm combines chaos optimization method and multi-step Quasi-Newton method is proposed by deriving the first derivative and its fast computation algorithm. And the global convergence of the algorithm is proved and tested. The simulation result indicates that the presented method is preferable than that of direct searching method, chaos optimization algorithm and multi-step quasi-Newton method mixed with direct searching method considering computation efficiency and accuracy.

关 键 词：分数阶FOURIER变换 混沌优化 多步拟牛顿法 线性调频信号 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]