摘要
提出了一种新的变步长最小均方(least-mean-square,LMS)自适应谐波检测算法,并将其应用于有源电力滤波器中。以一种时变步长迭代方法取代传统的定步长迭代法,旨在提高算法在求取均方差最小值的过程中方向估值的精度,从而达到提高算法收敛速度的目的。文中详细推导了2阶步长迭代公式,并导出了扩展的N阶步长迭代公式的递归表达式。通过仿真将所提出的新算法与传统的定步长算法进行了比较,证明新算法的收敛特性明显优于传统算法。
A novel adaptive harmonic detection algorithm, in which the least-mean-square （LMS） algorithm with variable step size is used, is proposed and applied to active power filter （APF）. In the proposed algorithm the traditional fixed step size iteration is substituted by time-varying step size iteration to improve the accuracy of directional estimation while the minimum mean square deviation is solved to achieve the purpose of increasing the convergence speed of the proposed algorithm. In this paper a second order step size iteration formula is derived in detail and the recursive expression for the expanded N-order step size iteration formula is deduced. By means of simulation, the proposed algorithm is compared with traditional fixed step size algorithm, thus it is proved that the convergence speed of the proposed algorithm is evidently superior to traditional algorithm.
出处
《电网技术》
EI
CSCD
北大核心
2008年第13期40-44,共5页
Power System Technology
