The prediction of the remaining useful life (RUL) of rolling bearings is a pivotal issue in 1
industrial production. A crucial approach to tackling this issue involves transforming vibration 2
signals into health indicators (HI) to aid model training. This paper presents an end-to-end HI 3
construction method, vector quantised variational autoencoder (VQ-VAE), which addresses the need 4
for dimensionality reduction of latent variables in traditional unsupervised learning methods such 5
as autoencoder. Moreover, concerning the inadequacy of traditional statistical metrics in reflecting 6
curve fluctuations accurately, two novel statistical metrics, mean absolute distance (MAD) and 7
mean variance (MV), are introduced. These metrics accurately depict the fluctuation patterns in the 8
curves, thereby indicating the model’s accuracy in discerning similar features. On the PMH2012 9
dataset, methods employing VQ-VAE for label construction achieved lower values for MAD and MV. 10
Furthermore, the ASTCN prediction model trained with VQ-VAE labels demonstrated commendable 11
performance, attaining the lowest values for MAD and MV.