一种改进变步长的自适应谐波检测算法

An improved variable step-size adaptive harmonic detecting algorithm

在有源电力滤波器(APF)谐波电流检测自适应滤波算法中,改进变步长自适应滤波系数调整方法。算法根据稳态谐波零均值的特点,利用系统历史积累误差均值估计代替单步误差更新步长因子,由滑动指数加权窗方式,提取出单纯的滤波器跟踪误差,步长更新的大小只取决于滤波器跟踪误差,不受谐波本身影响。通过步长因子与系统误差均值估计的类Sigmoid函数关系,动态控制类Sigmoid函数参数的变化,由此克服了类Sigmoid函数参数固定引起的稳态阶段误差偏大的缺陷。改进算法的复杂度有所增加,但能够保证较快的收敛速度、较高的检测精度和动态跟踪效果。通过对稳态和跃变的负载电流的仿真,结果证明了算法的有效性。

This paper presents an improved method that can adjust the factors of the adaptive variable step-size filter in the harmonic current detection （APF） adaptive filtering algorithm. In this approach, according to the fact that the steady-state harmonic has a zero mean, the mean estimation of the system historical accumulative errors is utilized instead of the single step error to update the step factor. The pure tracking error of the filter is extracted by using the sliding exponential window. Thus, the size of step factor merely depends on the filter tracking error instead of the harmonic itself. Through the generic Sigmoid functional relationship between the step factor and the system error mean estimate, the change of the parameters for the generic Sigmoid function is controlled dynamically so that the problem of large errors in the steady state caused by the fixed parameters can be solved. In spite of its increased complexity, the proposed method ensures faster convergence, higher detection accuracy and better dynamic tracking. The simulation results of the steady and step rectifier load currents indicate the validity of the improved algorithm. This work is supported by Natural Science Foundation of Jiangsu Province（No. BK2009202）.