形态滤波方法在振动信号降噪中的应用

Application of morphology filtering method in vibration signal de-noising

针对在旋转机械振动信号采集过程中引入的噪声污染和基线漂移等问题,采用形态开-闭和闭-开组合运算对染噪的振动信号进行处理.采用形态滤波方法对振动信号进行降噪处理,无须考虑振动信号的频谱特征.在对信号进行基线漂移校正时,所取直线结构元素的长度为采样信号周期长度的一半即可,但是为了得到更好的效果,可以将结构元素取得长一些.为消除信号中混入的尖峰脉冲干扰,结构元素的长度应该远小于待滤波函数的长度,并大于干扰脉冲的宽度.通过在仿真信号中加入不同的噪声干扰检验形态滤波方法的降噪能力,并将形态滤波方法用于现场信号的降噪处理,处理后的信号较好地保持了原信号的特征,说明该方法具有良好的滤波降噪效果,而且算法简单,便于工程现场的使用.

Aimed at the noise interference and baseline drift introduced during the rotating machinery vibration signal acquisition process, the morphology filtering method was used with the combination of morphological open-closing and close-opening operations. There is no need to consider the vibration spectrum characteristics when using this method to de-noise the corrupted signals. In the processing of baseline drift, the length of structure element would be half of sample data＇s period. In order to get better effectiveness, the length could be prolonged. While in the processing of pulse noises, the length of structure element must be bigger than that of pulse noises, but it could not be longer than that of original signal. The de-noising capacity of the morphology filtering method was proved by adding different kinds of noises into the simulated signal. The results of this method applied to process practical data showed that the processed signal could keep the main characteristic of the original signal. This filtering method is efficient and convenient for field application.