基于CNN-DE-SVM的滚动轴承故障诊断研究

Research on Fault Diagnosis of Rolling Bearing Based on CNN-DE-SVM

滚动轴承是各类旋转机械中最为重要的元件之一,若滚动轴承在机械设备运行过程中发生故障又无法及时判断出故障,所造成的连锁反应会对整条生产线产生影响,从而给企业造成经济损失。为了及时判断出滚动轴承所发生的故障,提出一种基于卷积神经网络(CNN)和差分进化算法(DE)优化支持向量机(SVM)的故障诊断模型(CNN-DE-SVM),针对滚动轴承的典型故障开展研究。结果表明,CNN-DE-SVM模型拥有较高的特征提取性能与故障诊断精度。

The rolling bearing is the most important component in all kinds of rotating machinery,and is widely used in modern industrial equipments.If the rolling bearing fails in the process of mechanical equipment operation,and the failure can not be timely judged,the chain reaction caused will affect the whole production line,thus cause huge economic losses to the enterprise.In order to detect rolling bearing faults in time,convolution neural network(CNN)and differential evolution(DE)algorithm are proposed in this paper.An optimized support vector machines(SVM)fault diagnosis model(CNN-DE-SVM)is presented,and is used to study typical rolling bearing faults.The research results show that the CNN-DE-SVM model has high feature extraction performance and fault diagnosis accuracy,and can be extended and applied to the measured data of rolling bearings in enterprise production.