改进的自适应变步长LMS算法-电测与仪表.DOC

改进变步长LMS算法在软起动谐波检测中的应用 赵景，李华 (兰州交通大学自动化与电气工程学院 兰州730070) 摘要：针对现有自适应变步长最小均方(LMS)谐波电流检测算法在低信噪比环境中易受干扰影响，提出一种改进自适应变步长LMS算法。该算法将误差信号与前一工频周期误差信号的差值作为反馈量，结合箕舌线函数构造出随动的动态因子，将此动态因子也作为调节权值的动量因子，利用自相干估计误差控制步长。该算法折中考虑收敛速度和稳态精度，并有效降低噪声信号的干扰。通过对电机软起动器工作过程中产生的周期重复性谐波进行检测和分析，证明了该谐波电流检测算法的可行性。 关键词：有源电力滤波器；谐波电流检测；动态因子；箕舌线函数；软起动器 中图法分类号：TM711； 文献标识码：A 文章编号：1001-1390（2014）00-0000-00 Application of Improved Variable Step Size LMS Algorithm to the Soft Starter Harmonic Detection ZHAO Jing, LI Hua (School of Automation and Electrical Engineering, Lanzhou Jiaotong University, Lanzhou 730070, China) Abstract: In view that the existing adaptive variable step size least mean square (LMS) harmonic current detection algorithm is easy to be disturbed in the low signal-to-noise environment, an improved adaptive variable step size LMS algorithm was proposed. In this algorithm, the difference between the error signal and that from the previous power cycle was used as the feedback, and its combination with versiera function constructed servo dynamic factors. The dynamic factors were regarded as the momentum factors adjusting weight. The coherence estimation error was used to control the step size. The algorithm compromised the convergence rate and stability accuracy, and it also reduced effectively the interference of noise signals. Through the detection and analysis of the periodically repeated harmonic in the working process of the motor soft starter, , it was verified that the new detection algorithm was feasible. Key word: active power filter, harmonic current detection, dynamic factor, versiera function, soft starter 0 引 言 直接起动异步电机会产生过大的冲击电流[1]，由三相六支反并联晶闸管组成的软起动器通过控制晶闸管触发角，调节电机起动电压、转速，限制起动电流，减小机组系统传动磨损，已得到广泛应用[2]。但是晶闸管软起动器作为电力电子器件长足进步的代表产品，它所引起的高次谐波对电网造成了严重污染，对电力系统中设备的安全运行造成威胁。目前解决这种谐波污染问题的重要途径是采用有源滤波器(Active Power Filter , APF)装置[34]。APF性能的重要指标之一，是能否准确、实时检测负载谐波信号。自适应谐波检测法因其检测算法对电网参数依赖性小，计算相对简单，具有良好的鲁棒性，近年来受到广泛重视。 自适应谐波检测随研究的深入，方法不断改进完善。如今变步长LMS算法早已取代传统定步长算法。如，算法的步长变化是基于Sigmoid