摘要
基于Lipschitz下界估值和分枝定界技术,给出了一维参数化小波滤波器逼近问题的全局最优算法。由于充分利用了滤波器逼近问题的特点,本方法将原来的Lipschitz算法的线性收敛速率提高为二次收敛速率。
Based on Litschitz infimum estimate and the technology of branch-bound,the global optimization algorithm for one-dimension finite wavelet filter approach is presented,Taking advantage of the peculiarity of this filter approach problem,the convergence rate of Lipschitz algorithm was improved from linear to second-order.
出处
《国防科技大学学报》
EI
CAS
CSCD
北大核心
2006年第6期58-62,共5页
Journal of National University of Defense Technology
基金
国家自然科学基金资助项目(60673090)
关键词
参数化正交小波滤波器
分枝定界
全局优化
formulated finite wavelet filters
technology of branch-bound
global optimization