Application of Deep Learning in Fault Diagnosis of Rolling Bearings Under Variable Operating Conditions

Rolling bearing supports the rotating shaft and the parts on the shaft, and fault diagnosis of bearings driven by big data is an important means to ensure the safe operation of equipment. In response to the increasing complexity of vibration signals in rolling bearings under variable operating conditions and the entry of mechanical fault diagnosis into the "big data era", the fault diagnosis method based on signal processing is inadequate. This paper applies deep learning to fault diagnosis of rolling bearings under variable operating conditions, constructs a deep convolutional neural network model with small convolutional kernels, applies experimental data of rolling bearings under variable operating conditions to construct variable domain signal datasets, and trains network models with the same structure. Three types of datasets were used to train deep learning network models with the same structure, and the trained models were tested on the test dataset. The results showed that the small convolutional kernel convolutional neural network trained directly on the vibration signal dataset had a fault diagnosis accuracy of up to 100%. This network has good stability and achieves end-to-end fault diagnosis of variable working condition rolling bearings in the era of big data, it overcomes the drawbacks of traditional machine learning, which relies on feature engineering and has low efficiency and accuracy.