基于EMD多尺度威布尔分布与HMM的轴承性能退化评估方法

轴承作为旋转机械中的重要部件,对其性能退化状态进行准确评估是开展预测性维护的重要前提。针对现有性能退化指标在鲁棒性和敏感性上的不足,提出一种基于多尺度威布尔分布与隐马尔可夫模型(Hidden Markov model,HMM)的滚动轴承性能退化评估方法。首先,采用经验模态分解(empirical mode decomposition,EMD)对轴承振动信号进行多尺度分解,将轴承振动数据分解到不同尺度的本征模态分量(intrinsic mode function,IMF)中;然后,通过峭度指标选取故障特征信息明显的IMF分量,并对各个IMF分量进行滑动窗口威布尔分布拟合,提取多尺度威布尔形状参数作为性能退化特征;最后,将轴承正常状态下退化特征参数输入隐马尔可夫模型(Hidden Markov model,HMM)进行训练,建立性能退化评估模型,从而实现轴承性能退化评估。试验结果表明,该评估方法可以有效反映轴承的性能退化趋势,与其他相关方法相比,该方法能够及时识别到轴承早期故障,并且具有较强的稳定性。

Evaluation method of bearing performance degradation based on EMD multi-scale Weibull distribution and HMM

Bearings are important parts of rotating machinery.Accurate evaluation of their performance degradation state is an important premise of predictive maintenance.Here,aiming at shortages of robustness and sensitivity of existing performance degradation indexes,a rolling bearing performance degradation evaluation method based on multi-scale Weibull distribution and HMM was proposed.Firstly,the multi-scale decomposition of bearing vibration signal was performed with the empirical mode decomposition(EMD),and bearing vibration data were decomposed into intrinsic mode function(IMF)components of different scales.Then,IMF components with obvious fault feature information were selected using kurtosis index,and each of them was fitted with sliding window Weibull distribution,and multi-scale Weibull shape parameters were extracted as performance degradation features.Finally,bearing degradation feature parameters under normal state were input into hidden Markov model(HMM)for training,and the performance degradation evaluation model was established to realize bearing performance degradation evaluation.The test results showed that the proposed evaluation method can effectively reflect bearing performance degradation trend;compared with other related methods,the proposed method can identify bearing early faults in time,and have stronger stability.