Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Dynamic fault detection and diagnosis for alkaline water electrolyzer with variational Bayesian Sparse principal component analysis

Qi Zhang, Weihua Xu, Lei Xie, Hongye Su

TL;DR

The study tackles fault detection and diagnosis in alkaline water electrolyzers under noisy industrial data. It develops variational Bayesian sparse PCA (VBSPCA) with Gaussian and Laplace priors, tying sparsity to $L_2$ and $L_1$ regularization, and couples it with a sparse vector autoregression to capture dynamic latent-variable relationships. The approach yields probabilistic loading matrices, automatic sparsification via ARD-type gamma priors, and dynamic fault indicators based on $T_2$ and $SPE$, with KDE-based thresholding. Industrial data from a large-scale AWE plant show effective fault detection and diagnosis, demonstrating robustness to noise and improved interpretability.

Abstract

Electrolytic hydrogen production serves as not only a vital source of green hydrogen but also a key strategy for addressing renewable energy consumption challenges. For the safe production of hydrogen through alkaline water electrolyzer (AWE), dependable process monitoring technology is essential. However, random noise can easily contaminate the AWE process data collected in industrial settings, presenting new challenges for monitoring methods. In this study, we develop the variational Bayesian sparse principal component analysis (VBSPCA) method for process monitoring. VBSPCA methods based on Gaussian prior and Laplace prior are derived to obtain the sparsity of the projection matrix, which corresponds to $\ell_2$ regularization and $\ell_1$ regularization, respectively. The correlation of dynamic latent variables is then analyzed by sparse autoregression and fault variables are diagnosed by fault reconstruction. The effectiveness of the method is verified by an industrial hydrogen production process, and the test results demonstrated that both Gaussian prior and Laplace prior based VBSPCA can effectively detect and diagnose critical faults in AWEs.

Dynamic fault detection and diagnosis for alkaline water electrolyzer with variational Bayesian Sparse principal component analysis

TL;DR

The study tackles fault detection and diagnosis in alkaline water electrolyzers under noisy industrial data. It develops variational Bayesian sparse PCA (VBSPCA) with Gaussian and Laplace priors, tying sparsity to and regularization, and couples it with a sparse vector autoregression to capture dynamic latent-variable relationships. The approach yields probabilistic loading matrices, automatic sparsification via ARD-type gamma priors, and dynamic fault indicators based on and , with KDE-based thresholding. Industrial data from a large-scale AWE plant show effective fault detection and diagnosis, demonstrating robustness to noise and improved interpretability.

Abstract

Electrolytic hydrogen production serves as not only a vital source of green hydrogen but also a key strategy for addressing renewable energy consumption challenges. For the safe production of hydrogen through alkaline water electrolyzer (AWE), dependable process monitoring technology is essential. However, random noise can easily contaminate the AWE process data collected in industrial settings, presenting new challenges for monitoring methods. In this study, we develop the variational Bayesian sparse principal component analysis (VBSPCA) method for process monitoring. VBSPCA methods based on Gaussian prior and Laplace prior are derived to obtain the sparsity of the projection matrix, which corresponds to regularization and regularization, respectively. The correlation of dynamic latent variables is then analyzed by sparse autoregression and fault variables are diagnosed by fault reconstruction. The effectiveness of the method is verified by an industrial hydrogen production process, and the test results demonstrated that both Gaussian prior and Laplace prior based VBSPCA can effectively detect and diagnose critical faults in AWEs.
Paper Structure (16 sections, 58 equations, 9 figures, 3 tables)

This paper contains 16 sections, 58 equations, 9 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Related work
  3. PCA and Probabilistic PCA
  4. Sparsity analysis
  5. Variational Bayesian Sparse PCA
  6. Gaussian prior-based VBSPCA
  7. Laplace prior-based VBSPCA
  8. Variational Bayesian inference
  9. A review of variational inference
  10. Variational inference for Gaussian prior-based VBSPCA
  11. Variational inference for VBSPCA with Laplace prior
  12. Dynamic Process Monitoring
  13. Fault detection
  14. Fault diagnosis
  15. Industrial AWE hydrogen production studies
  16. ...and 1 more sections

Figures (9)

  • Figure 1: Flow chart of alkaline electrolysis system.
  • Figure 2: Dynamic latent variables.
  • Figure 3: Autocorrelation coefficients of dynamic latent variables.
  • Figure 4: Detection result of fault 2.
  • Figure 5: Diagnosis result of fault 2.
  • ...and 4 more figures

Theorems & Definitions (10)

  • Remark 1
  • Remark 2
  • Remark 3
  • Remark 4
  • Remark 5
  • Remark 6
  • Remark 7
  • Remark 8
  • Remark 9
  • Remark 10