To overcome the deficiency of sudden noise interference and huge computation contained in a-coustic emission signals for condition monitoring of sliding bearings, a morphological filter was used to de-noise the signals. Real-time morphology filtering was carried out with optimized filter for acoustic emission signals of sliding bearing obtained during field testing of a 310 MW turbine-generator set, based on which time-domain characteristic parameters were real-timely calculated for No. 4 sliding bearing in the process of speed increasing, such as the root mean square value VRMS, peak value Vc and kurtosis factor Fk, etc. Denoising effect of the optimized morphological filter was finally compared with that of wavelet filtering. Results show that the optimized morphological filter with reasonably selected structure elements can well denoise acoustic emission signals, retain their original characteristics, and produce better filtering effect than wavelet filter. The real-time characteristic parameters obtained by morphological filtering help to diagnose lubrication faults of sliding bearings rapidly and accurately, which therefore may serve as a reference for actual condition monitoring of sliding bearing with acoustic emission signals.%针对滑动轴承状态监测时声发射信号的噪声干扰严重、突发性强和信号处理量大的特点,采用形态滤波对声发射信号进行降噪处理.对形态滤波器进行了优化设计,针对某310 MW汽轮发电机组滑动轴承现场试验获得的声发射信号进行了实时形态滤波,在此基础上实时计算了4号滑动轴承升速过程中声发射信号的时域特征参数(均方根值VRMS、峰值Vc、峭度因子Fk),并与小波滤波法的计算值进行了对比.结果表明:经优选合适的结构元素设计的形态滤波器能更好地滤除滑动轴承声发射信号的噪声、保留原始信号特征,滤波效果优于小波滤波方法；经形态滤波后的实时特征参数能快速准确地诊断滑动轴承润滑故障,在滑动轴承声发射状态监测中具有很好的工程应用价值.
展开▼
