基于卷积双向长短期记忆网络的轴承故障尺寸估计

基于振动监测数据的航空发动机滚动轴承损伤大小识别，对于研究滚动轴承故障演化、故障预测和故障诊断具有重要意义。针对传统模型对先验知识依赖性高、特征提取不充分、故障尺寸训练类别有限等问题，提出了一种基于深度学习的滚动轴承损伤尺寸预计方法，能够对训练过程中未出现的中间尺寸进行准确识别。在经典模型的基础上，搭建了一种深度卷积网络与长短期记忆网络组合模型，该模型可对轴承振动信号的多维特征与时序特征进行充分提取，实现轴承故障的智能和高效诊断。最后，利用滚动轴承加速疲劳试验机，进行了多种转速与损伤尺寸下的滚动轴承故障试验，基于试验数据进行了方法的比较，结果表明，该组合网络的在正常和加噪的情况下预测精度分别达到99.94%和98.67%，较单独的深度卷积网络、长短期记忆网络及其他模型精度更高，比较结果充分表明了本文所提方法的优越性。

Bearing fault size estimation based on convolutional bidirectional long and short term memory networks

The damage size identification of aero-engine rolling bearing based on vibration monitoring data is of great significance to the study of rolling bearing fault evolution, prediction and diagnosis. In view of inherent restrictions in traditional identification models such as high dependence on prior knowledge, insufficient feature extraction and limited category of training fault sizes, a prediction method of rolling bearing damage size based on deep learning was proposed, which can accurately identify the middle sizes that did not appear in the training process. A combined model of deep convolutional long-short-term memory network was developed, which can sufficiently extract the multi-dimensional and time-series characteristics of bearing vibration signal, and realize the intelligent and efficient diagnosis of bearing fault. On the basis of theoretical analysis, the rolling bearing fault tests under various damage sizes and rotational velocities were carried out by using the accelerated fatigue testing machine for rolling bearings, and the traditional and novel methods were compared based on the test data. The results showed that the prediction accuracy of the combined network can reach 99.94% and 98.67%, respectively, under normal and noisy conditions, higher than the single deep convolution network, long-short-term memory network and other models. The comparison results amply demonstrate the superiority of the proposed method.