Research on Fault Diagnosis of Air Compressor Bearings Based on VMD BiLSTM Attention

The safe, efficient, and stable operation of the power workshop equipment in cigarette factories is the key to ensuring production efficiency. A fault diagnosis model based on variational mode decomposition, attention mechanism, and bidirectional long short-term memory network (VMD BiLSTM Attention) is proposed for fault feature extraction, diagnosis, prediction, and early warning of bearing equipment such as air compressors. The initial vibration signal is directly used as the model input, and the modal components are reconstructed through VMD decomposition to remove noise. Multiple bidirectional LSTM networks of different scales are trained to extract features from the initial signal and obtain multi-scale equipment fault features. By introducing attention mechanism and optimizing parameters based on different feature weights extracted, the expected features of the target are selected and retained, redundant features are removed, and effective fault features are accurately extracted. Different types and degrees of equipment fault features can be identified, and the predicted results of the equipment status are sent to the fault classifier for judgment, which determines whether to issue a fault warning. This article validates the effectiveness of this model in fault feature extraction and high-precision fault classification through training on device status data.