基于格拉姆角场和PSO-CNN的滚动轴承故障诊断方法

Rolling bearing fault diagnosis method based on Gramian angle field and PSO-CNN

针对卷积神经网络的结构对滚动轴承故障诊断精度有较大影响的问题,提出一种基于格拉姆角场和粒子群优化卷积神经网络结构的故障诊断方法。采用格拉姆角场对一维轴承振动数据重构,保留原始数据信息的同时包含了时间相关性;采用粒子群优化算法对编码后的卷积神经网络结构迭代寻优。利用西储大学的轴承数据集进行试验验证,试验结果表明,该方法可自适应生成网络结构,平均诊断精度为99%,相对于其他主流卷积神经网络结构可以获得更好的故障诊断精度。

The structure of convolutional neural network has a great influence on the fault diagnosis accuracy of rolling bearings,a fault diagnosis method based on Gramian angular field and particle swarm optimization of convolutional neural network structure is proposed.The one-dimensional bearing vibration data is reconstructed using the Gramian angular field,retaining the original data information while including time correlation;the particle swarm optimization algorithm is used to iteratively optimize the encoded convolutional neural network structure.The method is experimentally validated using thebearing dataset from Case Western Reserve University.Experimental results show that the proposed method can generate network structure adaptively,and the average diagnostic accuracy is 99%,which can obtain better fault diagnosis accuracy than other mainstream convolutional neural network structures.