Remaining Bearing Life Prediction for Generative Adversarial Networks Based on Data Alignment

This article proposes a method for predicting the remaining life of bearings based on Data Alignment Generative Adversarial Network (DA-GAN). GAN is used to learn the distribution of health status data and establish health indicators to determine the starting point of prediction. To address the differences in data distribution between machines, a data alignment strategy is introduced to extract invariant features, and data alignment is achieved through adversarial training in the learning subspace of healthy and degraded data, using generalized features to predict lifespan. Meanwhile, the effectiveness of this method was demonstrated through comparative experimental results.