A deep neural network based on kernel function and auto-encoder for bearing fault diagnosis

Intelligent diagnosis technology has been widely applied in the classification of rotating machinery condition. The determination of the effective features from raw vibration data and the provision of accurate fault diagnosis results are important tasks for a bearing fault diagnosis. In this paper, a deep neural network based on Gaussian radial basis kernel function and auto-encoder (AE) is proposed. Firstly, a new AE is constructed with a kernel function to enhance the feature learning capability. Subsequently, a stacked AE network is developed with one new AE and T AEs. Finally, an error back-propagation algorithm is used to fine-tune the model parameters and then the deep characteristics are used to train the intelligent diagnosis model. Experimental data for an aircraft-engine inter-shaft bearing are used to verify the method. The results show that the proposed method has a better feature extraction capability, requires fewer iterations, and has a higher accuracy than standard methods using a stacked AE.