基于包络分析的汽轮机叶片振动监测信号处理方法  

Signal Processing Method for Vibration Monitoring of Turbine Blades Based on Envelope Analysis

作　　者：刘志德[1] 程学亮 魏红阳[1] 翟彦恺 LIU Zhi-de;CHENG Xue-liang;WEI Hong-yang;ZHAI Yan-kai(Harbin Turbine Company Limited,Harbin 150046,China)

机构地区：[1]哈尔滨汽轮机厂有限责任公司,哈尔滨150046

出　　处：《汽轮机技术》2024年第2期117-119,共3页Turbine Technology

摘　　要：以汽轮机末级叶片为研究对象,使用末级叶片径向围带上方的传感器采集叶片振动信号,针对信号波动较大,难以确定每支叶片通过传感器位置时刻的问题,采用时域信号的处理方法来确定触发时刻。对信号使用两层Hampel滤波器来消除信号的野点;对处理完的数据进行重采样,消除小范围的波动量;使用平滑滤波来处理数据,使得信号在局部具有单调性;求取信号的包络线,将包络线乘以系数作为信号的触发门限。相对于频域滤波方法,该方法较好地保持了信号的形状,能够精确地识别叶片通过传感器的时刻,对于波动大的信号尤其适用。This paper focuses on the vibration monitoring of the last-stage blades in steam turbines.It utilizes the vibration signals collected from sensors placed above the radial band of the last-stage blades.In order to address the issue of significant signal fluctuations and the challenge of determining the precise moments when each blade passes through the sensor position,a time-domain signal processing method is employed to determine the triggering moments.The method consists of the following steps.Two layers of Hampel filters are used to eliminate outliers in the signals.The processed data is resampled to remove small-scale fluctuations.Smoothing filters are applied to the data to enhance its local monotonicity.The envelope of the signal is calculated,and a coefficient is multiplied with the envelope to serve as the triggering threshold for the signal.Compared to frequency-domain filtering methods,this approach better preserves the shape of the signal and accurately identifies the moments when the blades pass through the sensor position.It is particularly suitable for signals with significant fluctuations.

关 键 词：叶片振动 时域 包络 滤波 

分 类 号：TK262[动力工程及工程热物理—动力机械及工程]