Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Utilizing Autoregressive Networks for Full Lifecycle Data Generation of Rolling Bearings for RUL Prediction

Junliang Wang, Qinghua Zhang, Guanhua Zhu, Guoxi Sun

TL;DR

This paper tackles the data scarcity problem in full lifecycle RUL prediction for rolling bearings by introducing CVGAN, a CVAE-GAN–based generator that conditions on historical vibration data and health indicators. An autoregressive generation strategy is proposed to produce continuous, full-lifecycle vibration signals, validated on the PHM 2012 dataset using $MMD$ and $FID$ as primary metrics. Results show CVGAN outperforms several baselines in both autoregressive and non-autoregressive modes, and training predictive models with CVGAN-generated full lifecycles significantly boosts RUL forecasting across multiple models. The work demonstrates the practical value of high-quality, conditionally generated data for PHM and highlights opportunities to extend to higher-dimensional signals and more efficient architectures.

Abstract

The prediction of rolling bearing lifespan is of significant importance in industrial production. However, the scarcity of high-quality, full lifecycle data has been a major constraint in achieving precise predictions. To address this challenge, this paper introduces the CVGAN model, a novel framework capable of generating one-dimensional vibration signals in both horizontal and vertical directions, conditioned on historical vibration data and remaining useful life. In addition, we propose an autoregressive generation method that can iteratively utilize previously generated vibration information to guide the generation of current signals. The effectiveness of the CVGAN model is validated through experiments conducted on the PHM 2012 dataset. Our findings demonstrate that the CVGAN model, in terms of both MMD and FID metrics, outperforms many advanced methods in both autoregressive and non-autoregressive generation modes. Notably, training using the full lifecycle data generated by the CVGAN model significantly improves the performance of the predictive model. This result highlights the effectiveness of the data generated by CVGans in enhancing the predictive power of these models.

Utilizing Autoregressive Networks for Full Lifecycle Data Generation of Rolling Bearings for RUL Prediction

TL;DR

This paper tackles the data scarcity problem in full lifecycle RUL prediction for rolling bearings by introducing CVGAN, a CVAE-GAN–based generator that conditions on historical vibration data and health indicators. An autoregressive generation strategy is proposed to produce continuous, full-lifecycle vibration signals, validated on the PHM 2012 dataset using and as primary metrics. Results show CVGAN outperforms several baselines in both autoregressive and non-autoregressive modes, and training predictive models with CVGAN-generated full lifecycles significantly boosts RUL forecasting across multiple models. The work demonstrates the practical value of high-quality, conditionally generated data for PHM and highlights opportunities to extend to higher-dimensional signals and more efficient architectures.

Abstract

The prediction of rolling bearing lifespan is of significant importance in industrial production. However, the scarcity of high-quality, full lifecycle data has been a major constraint in achieving precise predictions. To address this challenge, this paper introduces the CVGAN model, a novel framework capable of generating one-dimensional vibration signals in both horizontal and vertical directions, conditioned on historical vibration data and remaining useful life. In addition, we propose an autoregressive generation method that can iteratively utilize previously generated vibration information to guide the generation of current signals. The effectiveness of the CVGAN model is validated through experiments conducted on the PHM 2012 dataset. Our findings demonstrate that the CVGAN model, in terms of both MMD and FID metrics, outperforms many advanced methods in both autoregressive and non-autoregressive generation modes. Notably, training using the full lifecycle data generated by the CVGAN model significantly improves the performance of the predictive model. This result highlights the effectiveness of the data generated by CVGans in enhancing the predictive power of these models.
Paper Structure (17 sections, 7 equations, 15 figures, 10 tables)

This paper contains 17 sections, 7 equations, 15 figures, 10 tables.

Table of Contents

  1. Introduction
  2. Preliminary
  3. GAN
  4. CVGAN
  5. Network Structure
  6. Autoregressive Generation
  7. Experiments Overview
  8. Dataset Introduction
  9. Dataset Build
  10. Parameter Setting
  11. Experiments Details
  12. Selection of Metrics for Generation Method
  13. Network Structure Selection
  14. Ablation Study
  15. Comparsion with other Models
  16. ...and 2 more sections

Figures (15)

  • Figure 1: A flowchart of the proposed method.
  • Figure 2: The basic structure of CVAE-GAN
  • Figure 3: The specific network structure of VAE
  • Figure 4: The proposed autoregressive generation method.
  • Figure 5: Overview of PRO-NOSTIA
  • ...and 10 more figures