基于CNN-Transformer模型的自动扶梯轴承故障诊断方法研究

Research on escalator bearing fault diagnosis method based on CNN-Transformer model

针对自动扶梯滚动轴承故障诊断模型诊断识别率不高的问题,提出一种CNN与Transformer相结合的滚动轴承故障诊断方法。该方法采用CNN注意力和Transformer时域注意力融合机制,从信号的通道和时域两个维度提取最有利于识别的特征信息,降低信道和噪声等影响。根据特征信息的重要程度进行权重分配,以提升故障识别率。基于滚动轴承数据集,将所提算法模型和CNN、RNN、SVM算法进行对比。结果表明,该方法识别轴承故障的准确度更高,实现了对轴承故障的准确分类。

In allusion to the problem of low diagnostic recognition rate of escalator rolling bearing fault diagnosis model,a method of rolling bearing fault diagnosis combining CNN and Transformer is proposed.In the method,the fusion mechanism of CNN attention and Transformer time-domain attention is used to extract the feature information most conducive to recognition from both channel and time-domain dimensions of the signal,reduce the influences of channel and noise,etc.The weights are assigned according to the importance of the feature information to improve the fault recognition rate.Based on the rolling bearing dataset,the proposed algorithm model is compared with CNN,RNN,and SVM algorithms,and the results show that the method can identify bearing faults with higher accuracy,realizing the accurate classification of bearing faults.