摘要
通过改进传统仿射投影算法(APA)的约束条件,提出了一种新的变步长仿射投影算法(VSS-APA)。该算法具有随估计误差的变化自动调整步长的特点,同时克服了传统APA不能充分反映更新后的自适应滤波器系数应该尽可能逼近未知系统参数的缺点。在算法实现中,提出了简单有效的遗忘因子选择规则,以实现对误差能量的准确估计。最后将该算法成功应用于系统辨识。理论分析与计算机仿真表明,该算法有更快的收敛速率和更小的稳态均方误差。
By improving the constrain conditions of traditional affine projection algorithm （APA）, a new variable step size APA （VSS-APA） was proposed. The step size of this algorithm is adjusted automatically according to the change of the estimation error. This algorithm avoids the shortage of traditional APA, which can not reflect the request that the updated coefficients of adaptive filter should represent the best fit to the parameters of unknown system. In order to estimate the power of error more accurately, a simple and effective rule about the selecting of forgetting factor was proposed. In the last, this algorithm was applied to system identification successfully. The theoretical analysis and computer simulation results illustrate this algorithm leads to faster convergence rate and lower final misalignment.
出处
《系统仿真学报》
CAS
CSCD
北大核心
2009年第4期984-987,共4页
Journal of System Simulation
关键词
自适应滤波
仿射投影算法
变步长
系统辨识
adaptive filtering
affine projection algorithm
variable step size
system identification
