论文部分内容阅读
针对机械智能监测和故障诊断中故障样本缺乏的问题,提出了一种支持向量数据描述和用双谱提取故障特征值相结合的机械故障诊断方法。该方法可以只利用正常状态数据样本来建立单值分类器,判别机器的运行状态。高阶谱能有效地抑制噪声,对不同类型的故障,高阶谱存在明显差异。采用双谱对角切片对原始数据信号进行特征提取,将特征值作为SVDD的输入参数进行分类。运用该方法在滚动轴承的故障诊断中。
Aiming at the lack of fault samples in mechanical intelligent monitoring and fault diagnosis, a mechanical fault diagnosis method based on support vector data description and bispectrum extraction of fault eigenvalues is proposed. The method can use only the normal state data samples to establish a single value classifier to determine the running status of the machine. Higher order spectra can effectively suppress the noise, and there are obvious differences for different types of faults and higher order spectra. The bispectrum diagonal slices were used to extract the features of the original data signals, and the eigenvalues were classified as the input parameters of SVDD. Using this method in the fault diagnosis of rolling bearing.