摘要
为了提高轴承故障识别精度,同时减少甚至消除参数选择的影响,基于轴承振动加速度信号,提出了一种基于复合多尺度Bubble熵的滚动轴承故障诊断方法。基于轨道车辆轴箱轴承故障响应信号具有较强的非线性和非平稳特征,采用小波包变换频域能量特征重构,对重构信号提取复合多尺度Bubble熵作为故障特征,输入中等高斯支持向量机完成模型训练和故障模式识别。通过基于全尺寸单轮对-轴箱轴承滚动试验台试验的故障轴承数据集以及美国凯斯西储大学公用轴承数据集验证了所提方法的有效性。该方法在不经过参数调节过程的情况下可以达到较高的分类精度,在两种轴承数据集中的正常轴承与故障轴承之间的识别率均为100%,其中在公用轴承数据集中4种故障分类中总识别率为99.83%,在单轮对-轴承滚动试验台数据集中总识别率为93.75%,均高于同类轴承故障诊断算法。实验结果表明,该方法能够有效地提取轴承故障特征,为轨道车辆轴箱轴承状态监测与故障诊断提供了新的解决方案。
To improve the accuracy of bearing fault detection and reduce or eliminate the influence of parameter selection,a fault diagnosis method for rolling bearings based on composite multiscale bubble entropy is proposed using the vibration acceleration signal of the bearing.Since the response signals of axle-box bearings in rail vehicles exhibit strong nonlinear and non-stationary characteristics,wavelet packet transform frequency domain energy feature reconstruction is adopted.The composite multiscale Bubble entropy is extracted from the reconstructed signal as the fault feature,which is then input into a medium gaussian support vector machine for model training and fault mode recognition.The effectiveness of this method is verified using a dataset of faulty bearings from full-scale single-wheel-on-axle-box bearing rolling test rigs and a dataset of public bearings from Case Western Reserve University in the United States.This method achieves high classification accuracy without undergoing a parameter tuning process.The identification rates between normal and faulty bearings in both datasets are 100%,with a total identification rate of 99.83%for the four fault classifications in the public bearing dataset and 93.75%for the single-wheel-on-axle-box bearing rolling test rig dataset,which are higher than those of similar bearing fault diagnosis algorithms.Experimental results demonstrate that this method can effectively extract bearing fault features,providing a new solution for the monitoring and fault diagnosis of axle-box bearings in rail vehicles.
作者
邱小杰
樊麟华
陆正刚
Qiu Xiaojie;Fan Linhua;Lu Zhenggang(Insititute of Rail Transit,Tongji University,Shanghai 200092,China)
出处
《机电工程技术》
2024年第5期196-202,共7页
Mechanical & Electrical Engineering Technology
