论文部分内容阅读
机器振动信号包含有关机器状态的大量信息,可用来识别机器状态。W.Gersch[1]和屈梁生[2]用Kallback-Leibler信息数处理振动信号以识别机器状态,取得了较好效果。但是,对一些实际情况,尤其是对于恒转速机器,Gersch方法未能充分利用已知信息。本文用KL数在频域识别恒转速机器状态。由例行检测和试验,可以得到机器各状态的样本功率谱,从其中任取一个作为基准谱。在识别过程中,求出状态样本和待识别样本功率谱与基准谱的差,称为“差谱”。差谱中只包含各状态中互不相同的成分,提高了分析信号的信噪比。用KL数度量各状态差谱间的距离,并用聚类分析方法识别状态,提高了识别精度。经初步实验验证,效果良好。
The machine vibration signal contains a wealth of information about the machine’s status and can be used to identify the machine’s status. W. Gersch [1] and Qu Liangsheng [2] used Kallback-Leibler information to process the vibration signal to identify the machine state and achieved good results. However, for some practical purposes, especially for constant speed machines, the Gersch approach fails to take full advantage of known information. In this paper, the number of KL identification in the frequency domain constant speed machine state. By routine testing and testing, you can get the machine state of the sample power spectrum, from which any one as a reference spectrum. In the process of identification, the difference between the power spectrum of the sample to be identified and the reference spectrum is found, which is called “difference spectrum”. The difference spectrum contains only the different components in each state, which improves the signal-to-noise ratio of the analyzed signal. The distance between each state difference spectrum is measured by KL number, and the status is identified by cluster analysis to improve the recognition accuracy. After preliminary experimental verification, the effect is good.