Remaining Life Assessment of Rolling Bearing Based on Graph Neural Network

This chapter mainly focuses on the application of graph neural networks in intelligent maintenance. As one of the key components of machinery and equipment, bearings are highly susceptible to failure, and the safety and reliability of their operating conditions will directly affect the performance of the whole system. Therefore, it is important to carry out the remaining useful life assessment of rolling bearings. In recent years, deep learning-based methods have become popular in the field of intelligent maintenance for their superiority in fault characterization and extraction, due to the fact that degraded information of machines can be fully mined for fault information by deep learning. However, these methods still have the following drawbacks: (1) Due to the strong nonlinearity of bearing vibration signals, it is difficult for intelligent maintenance methods based on traditional feature extraction and deep learning to mine and characterize the structured information of the signals. (2) It is difficult for the existing deep learning methods to model data in non-Euclidean spaces. In order to solve the above problems, this chapter proposes a remaining useful life assessment method based on time–frequency recursive graph neural network. First, the time–frequency recursive graph of the vibration signal is calculated to obtain the periodic characteristics of the signal in the time–frequency space. Then the graph sample and aggregate network is constructed to further extract the nonlinear features to achieve the remaining life assessment of the bearing. The results were validated by full-life experimental data analysis and showed that the proposed method has certain superiority.