基于多尺度样本熵改进极限学习机的列车转向架轴承故障诊断

Fault Diagnosis of Train Bogie Bearing Based on Multi-scale Sample Entropy Improved Extreme Learning Machine

列车转向架轴承受到轨道激扰、运行速度不确定、轨道接缝的冲击振动和其他部件振动因素的影响,轴承振动信号呈现非线性和非平稳性,导致其故障特征难以被提取,故障诊断准确率低。针对上述问题,文章提出一种基于多尺度样本熵改进极限学习机的列车转向架轴承故障诊断方法。其首先利用多尺度样本熵提取故障特征,构成特征向量集;然后利用粒子群算法优化极限学习机,得到输入权值和隐含节点阈值;最后,将特征向量集划分为测试集和训练集,利用改进的极限学习机作为模式识别算法进行故障模式识别。实验结果表明,本文提出的方法能够有效地进行故障模式识别,识别准确率达到96%,适用于列车轴承故障诊断。

Train bogie bearing is affected by track excitation,uncertainty of running speed,impact vibration of track joints and vibration of other components.The vibration signal is nonlinear and non-stationary,which makes it difficult to extract fault characteristics and fault diagnosis accuracy is low.In view of above problems,a fault diagnosis method of train bogie bearing based on multi-scale sample entropy improved extreme learning machine is proposed.Firstly,fault features are extracted by using the multi-scale sample entropy to form a feature vector set.Then,the extreme learning machine is optimized by particle swarm optimization,and the input weights and hidden node thresholds are obtained.Finally,the feature vector set is divided into test set and training set,and the improved extreme learning machine is used as a pattern recognition algorithm for fault pattern recognition.Experimental results show that the proposed method can effectively carry out fault pattern recognition,and the recognition accuracy reaches 96%,which is suitable for train bearing fault diagnosis.