延伸奇异值分解包及其在高速列车轮对轴承故障诊断中的应用

Extended SVD packet and its application in wheelset bearing faultdiagnosis of high-speed train

结合多分辨奇异值分解包的分解结构和对滚动轴承故障信号的Hankel矩阵的奇异值分布特性研究,提出了延伸奇异值分解包。该算法的核心包括矩阵递推构造和矩阵重构。以分量信号能量为指标,提出了有效分量信号的筛选准则,并基于该准则,进一步提出了延伸奇异值分解包的快速算法。仿真结果表明,延伸奇异值分解包对信号中共振频带分量信号具有很好的分解能力,方法具有强鲁棒性,同时极大地改善了奇异值分解包中出现的模态混叠。应用高速列车轮对轴承试验数据对该方法进行试验验证,结果表明,该方法能有效分离高速列车轮对轴承复合故障信号的不同共振频带信号,对筛选的有效分量信号进行包络分析,可有效提取不同类型的故障特征频率及其谐波,对共振频带的聚集性和故障的表征力相比奇异值分解包均有显著提高。

Combining decomposed structure of the multi-resolution SVD package and singular value distribution characteristics of a rolling bearing fault signal’s Hankel matrix,the extended singular value decomposition(SVD)package was proposed.The core of this method included matrix recurrence construction and matrix reconstruction.Component signal energy was taken as an index to propose the screening criterion of effective component signals.Then based on the proposed criterion,a fast algorithm for the extended SVD packet was further proposed.The simulation results showed that the extended SVD packet has good decomposition ability for resonance frequency band components in a signal;the method has a strong robustness and greatly improves modal aliasing appearing in the SVD packet.The test data for high-speed train’s wheelset bearing were used to verify the proposed method.The results showed that this method can effectively separate different resonant frequency band signals in high speed train wheelset bearing compound fault signals,and perform envelope analysis for screened effective component signals to effectively extract different types fault feature frequencies and their harmonics;compared to the SVD package,the proposed method makes resonance bands’aggregation property and faults’characterizing ability be significantly improved.