摘要
提出一种非线性动态寻优的最小二乘算法(LMS),以高效地估计正弦信号幅值、相位和直流偏置.利用随机逼近理论中的Kiefer-Wolfowitz定理,提高了非线性梯度法搜索LMS估计量极值的收敛速度.以整周期内采样均值为迭代初值,迭代步长的选择根据各参数的差异分别选取,并采用修正因子以加快迭代的收敛速度,设置最大步长防止算法溢出.仿真表明,在计算量相当的情况下,该算法参数估计精度明显优于传统LMS.
An improved leastsquare method (LMS) algorithm of estimating the amplitude, phase and dcoffset of a knownfrequency sine time sequence in engineering was proposed. It is based on the nonlinear dynamic optimizaiton theory. Because of the nonlinear relationship between the sine time sequence and the amplitude and phase to be estimated, it is necessary to use a dynamic nonlinear optimization algorithm for the solutions of leastsquare algorithm. The practical iterative algorithm is an improvement of KieferWolfowitz algorithm. In addition, the principle of choosing the initial value and steplength considering the difference of all parameters, and how to improve the rate of convergence and the security of the algorithm were discussed. The simulation results exhibit that the proposed algorithm gives superior resolution to the traditional LMS in the same computation amount.
出处
《上海交通大学学报》
EI
CAS
CSCD
北大核心
2003年第10期1613-1615,共3页
Journal of Shanghai Jiaotong University
关键词
非线性寻优
最小二乘法
正弦信号
non-linear optimization
least-square method(LMS)
sine wave
