变步长自适应滤波算法的谐波检测

为提高自适应滤波算法的收敛速度,并降低其稳态误差,建立了LMS算法理论最优步长值与误差信号和输入信号之间的关系,提出了一种新的变步长LMS自适应谐波检测算法。该算法的优点是:根据误差信号的平方时间均值估计来调节步长因子,克服了以往算法在自适应稳态阶段步长调整过程中的不足。即使待检测信号的信噪比较低,检测过程也具有较快的动态响应速度和保持较小的稳态失调噪声。计算机仿真表明,该算法具有更好的收敛精度。

Harmonic detection based on variable step size adaptive algorithm

Aimed at increasing the convergent speed of adaptive filtering algorithm and reducing its steady-state error, this paper describes the development of the relation among theoretical optimal value of step for LMS algorithm, error signal and input signal and proposes a novel variable step-size LMS algo- rithm for harmonic detection. This algorithm presents following advantages： performing step size adjust- ment using the estimation of square of the time-average of error signal makes it possible to eliminate the shortcomings due to the adjustment of step size using conventional adaptive steady-state process; besides, the resultant detection demonstrate a faster dynamic response and less steady-state offset noise despite lower SNR of the signals to be detected. Some computer simulation shows that this algorithm has a better convergent precision.