声发射与混合维深度融合的滚动轴承智能诊断

Intelligent Diagnosis of Rolling Bearings of Acoustic Emission and Hybrid Dimension Deep Fusion

针对滚动轴承早期及复合故障难以准确诊断的问题,提出一种基于声发射与混合维深度特征融合的滚动轴承早期故障智能诊断模型,该模型可自适应捕获滚动轴承早期故障特征并自行诊断.首先,将滚动轴承早期原始声发射信号经连续小波变换转化为二维时频图.接着,分别将上述一维、二维数据输入以卷积神经网络(CNN)构建的1D-CNN与2D-CNN智能诊断模型框架,并提出采用基于特征金字塔网络的深度融合算法融合模型的低层与高层特征,同时以全局平均池化层代替全连接层抑制模型过拟合现象.试验结果表明,提出的方法具有更高的准确率、稳定性与鲁棒性.

As early and compound faults of rolling bearings are difficult to diagnose accurately, this paper proposes an intelligent diagnosis model for early faults of rolling bearings based on acoustic emission and the hybrid dimension deep feature fusion, which can adaptively capture and diagnose early fault features of rolling bearings. Firstly, the original acoustic emission signals of rolling bearings are transformed into two-dimensional time-frequency diagrams by continuous wavelet transform. Then, the above mentioned one-dimensional and two-dimensional data are input into the 1D-CNN and 2D-CNN intelligent diagnosis model frameworks constructed by Convolutional Neural Network(CNN);a deep fusion algorithm based on a feature pyramid network is proposed to fuse the low-level and high-level features of the model;and the global average pooling layer is used to replace the full connection layer to suppress the over-fitting phenomenon of the model. The experimental results show that the proposed method has higher accuracy, stability, and robustness.