基于改进多尺度排列熵的列车轴箱轴承诊断方法研究

多尺度排列熵作为非线性方法,被广泛应用于时间序列复杂性和随机性的评估之中。由于粗粒化过程中的缺陷会导致熵值精度低、稳定性差,提出了改进多尺度排列熵。通过仿真信号与传统多尺度排列熵方法比较发现,在不同尺度下改进多尺度排列熵方法估计的熵值结果更加稳定,且误差减小。结合马氏距离特征选择与遗传算法优化的支持矢量机模式识别算法,提出了一种智能化的轴承故障诊断方法。通过列车轴箱轴承实验数据进行验证,结果表明该方法可准确识别出不同类型的故障轴承。

A Study on Fault Diagnosis Method for Train Axle Box Bearing Based on Modified Multiscale Permutation Entropy

Multiscale permutation entropy(MPE)is a well-known non-linear method widely used to evaluate the complexity and stochasticity of time series.However,the drawbacks of the coarse-graining process may lead to an imprecise and unstable entropy value.To overcome these problems,the modified multiscale permutation entropy(MMPE)was proposed.Compared with MPE method using synthetic signals,the MMPE obtained more reliable entropy values and had notably smaller standard deviation in each scale.In addition,combined with Mahalanobis Distance for feature selection and support vector machine-genetic algorithm,an intelligent rolling bearing multi-fault diagnosis method was proposed.The experimental results of train axle box bearings show the proposed method is an effective tool to identify different categories of defects in rolling element bearings.