【摘要】 以机械碰摩信号为例 ,采用具有“时频局部化”性能和“方向极化”性能的二维小波变换 ,成功地分辨出故障信号与正常信号 ,避免了由于多信号转化成单信号带来的信息冗余以及能量的转化 ,同时能方便地分辨出故障信号的相位角 ,为解决机械故障诊断领域多信号、多特征的分析问题提出新的思路更多还原

【Abstract】 Mechanical fault diagnosis is always a key question that is focused by many researchers. And the theory of wavelet transformations has been widely used to draw the fault characteristics. But there is no a good method to analyze several sets of sampling data simultaneously, and they always have to be changed into a single set or dealt respectively. The theory of two dimensional (2d) wavelet transformations, which has been widely used to process images question, can deal with multi dimensional data simultaneously, and has the ability to show the detailed characteristics of the signals in time frequency region and in a certain direction. In this paper, taking a set of mechanical rub impact signals as an example, two dimensional wavelet transformations is applied to analyze a new multi dimensional matrix formed with this set of signals. To sure the ability to determine the phase of signals, a new set of data is sampled after rotating the rum pact mass a determined angle and its 2d wavelet analyzed result is compared with the frontal one. Thus the result shows that it can distinguish the fault signals from several sets of sampling data and confirm the phase of fault signals without data transformation. And it can avoid the redundancy and can be thought as a new method to deal with some multi signals or multi characteristics questions. It can be also thought as a promising trial for further application of two dimensional wavelet transformation to fault diagnosis.更多还原