基于双层相空间相似度的滚动轴承故障模式与故障程度的综合辨识

摘要

A comprehensive method for rolling bearing fault patterns and fault degree recognition based on two-layer algorithm structure of phase space similarity analysis was presented in this paper.In the first layer of the algorithm,the data were processed by the phase space reconstruction (PSR)to get a phase space which was equivalence in the topological sense.Then a sliding window was employed to chop the data segments and the normalized cross correlation function (NCC)was employed to execute similarity analysis,realizing the classification of bearing fault patterns.In the second layer,a SVR structure was trained by phase space similarity (PSS)that was obtained in different fault degree. The SVR structure was then used to recognize the fault degree.The results of experimental signal analysis show that the proposed method can effectively recognize comprehensive bearing fault pattern and fault degree.Compared with traditional methods,it shows an improvement in accuracy of recognition.%提出了一种基于双层相空间相似度分析算法结构，应用于滚动轴承故障类型和故障程度的综合辨识。该算法第一层结构中，对测试数据和样本数据进行相空间重构（PSR），得到在拓扑意义下等价的相空间，然后使用滑动窗截取数据段，采用归一化互相关函数（NCC）进行相空间相似度分析，实现轴承故障类型的分类；在第二层结构中，以已知不同故障程度数据之间的相空间相似度（PSS）为特征训练 SVR 结构，实现对故障程度的跟踪。实验信号分析结果表明，该方法能有效对轴承故障类型和故障程度进行综合辨识。与传统方法的对比表明该方法在准确性上有了一定的提高。