基于非线性数学形态变换的概念设计了形态非抽样小波变换算法,通过构造信号分解算子和结构元素,经过多尺度形态小波分解既能够平滑噪声又提取了信号中的故障特征成分。分别对模拟信号和实验数据进行分析处理,结果均表明该方法对信号冲击特征的提取是有效的。最后通过与包络解调分析方法的对比,说明了形态非抽样小波变换对滚动轴承故障特征的提取效果更明显。由于形态非抽样小波变换算法只涉及加减和取极大、极小运算,运算简单,执行高效,非常适于滚动轴承故障的在线监测和诊断。
Based on mathematical morphology theory, the morphological undecimated wavelet transform (MUWT) algorithm was presented according to the shape feature of the signals. The multi-scale MUWT operation can not only smooth the background noises but also extract the characteristic components by constructing the signal decomposition operators and the structuring elements. This method was used to analyze the simulated data and measured signals from the bearing test rig. The results reveal that it is effective to the impulse characteristics extraction. Comparied with the normal enveloping demodulation method, the MUWT operation is more simple and effective for defect diagnosis in the roller bearing. The MUWT algorithm includes addition, subtraction, maximum and minimum operations, and does not involve multiplication and division, the signal information is not lost in the decomposition procedure. It is suitable for the on-line faults monitoring and diagnosis of roller bearing.