An Efficient Machine Learning Approach for Degradation Forecasting in AEM Water Electrolysis

Marco Veneriano; Ani Gjergji; Sebastiano Bellani; Andrea Riva; Vito Paolo Pastore; Matteo Santacesaria

cs.LGSep 29, 2026

An Efficient Machine Learning Approach for Degradation Forecasting in AEM Water Electrolysis

Authors: Marco Veneriano, Ani Gjergji, Sebastiano Bellani, Andrea Riva, Vito Paolo Pastore, Matteo Santacesaria

Organizations: MaLGa Center Universit`a degli Studi di Genova Genoa, Italy · Antares Electrolysis S.r.l. Genoa, Italy

Abstract

This study provides a data-driven analysis of a novel dataset of single-cell Anion Exchange Membrane water electrolyzers (AEMWE), operated under constant current load across multiple heterogeneous experimental campaigns. We train and evaluate a range of machine learning models with different complexity, including linear baselines, LSTMs and CNNs, to perform medium-term forecasting of the cell voltage degradation curve. The models are assessed within a rigorous training and evaluation framework specifically designed for heterogeneous industrial data.

Figures & tables

Fig. 1: Representative raw voltage profile from a single experiment.

Horizon	Metric	Linear	Ridge	RF	Shallow NN	LSTM	1D-CNN
24h	Norm	$2.708$	$2.567$	$2.960\pm 0.015$	$\mathbf{1.816}\pm\mathbf{0.015}$	$1.977\pm 0.205$	$1.845\pm 0.158$
24h	cV	$2.308$	$1.909$	$2.362\pm 0.007$	$\mathbf{1.000}\pm\mathbf{0.005}$	$1.063\pm 0.146$	$1.284\pm 0.221$
18h	Norm	$2.386$	$2.127$	$2.251\pm 0.023$	$1.733\pm 0.019$	$\mathbf{1.730}\pm\mathbf{0.035}$	$1.756\pm 0.059$
18h	cV	$2.001$	$1.710$	$1.772\pm 0.024$	$1.150\pm 0.020$	$\mathbf{1.091}\pm\mathbf{0.059}$	$1.359\pm 0.074$
12h	Norm	$2.064$	$1.614$	$1.727\pm 0.025$	$\mathbf{1.471}\pm\mathbf{0.004}$	$1.562\pm 0.041$	$1.543\pm 0.007$
12h	cV	$1.868$	$1.393$	$1.483\pm 0.014$	$1.180\pm 0.010$	$\mathbf{1.157}\pm\mathbf{0.039}$	$1.278\pm 0.013$

TABLE I: Global performance summary: Comparison of models across time horizons (MAE).

Explore similar work

Performance Monitoring of Proton Exchange Membrane Water Electrolyzer by Transformers-Based Machine Learning Model
May 18, 2026Bingqing Chen, Ivan Batalov, Qiu Chen +2Predictive Maintenance Time Series Transformers
Bosch Research & Technology Center
BatteryMFormer: Multi-level Learning for Battery Degradation Trajectory Forecasting
May 26, 2026Ruifeng Tan, Jintao Dong, Weixiang Hong +3Multivariate Time Series Forecasting RUL Estimation
School of Computer Science and Engineering, Central South University · Material Genome Institute, Shanghai University
Causal Anomaly Detection for Lithium-Ion Battery Degradation
May 17, 2026Dieter W. Heermann, Hagen HeermannPredictive Maintenance Time-Series Anomaly Detection
Institute for Theoretical Physics, Heidelberg University, Philosophenweg 19, D-69120 Heidelberg, Germany · Intilion GmbH, Wollmarktstraße 115c, D-33098 Paderborn, Germany

May 18, 2026, cs.LGJ/K move · Enter open · S save

Performance Monitoring of Proton Exchange Membrane Water Electrolyzer by Transformers-Based Machine Learning Model

Bingqing Chen, Ivan Batalov, Qiu Chen, Weiqi Ji+1

Bosch Research & Technology Center

Predictive Maintenance Time Series Transformers

Open paper AskPDF arXiv

Green hydrogen plays an essential role in decarbonization, with capacity projected to scale to 560 GW by 2030 (vs. 1.39 GW in 2023) in net-zero settings. Proton exchange membrane (PEM) electrolysis is one of the most promising technology routes to green hydrogen production, and real-time system health monitoring of PEM electrolyzers is essential for their scalable deployment. In lab settings, performance degradation can be characterized through electrochemical testing protocols by periodic pauses of normal operation. Such interruption is not practical for full-scale stack deployments, limiting system operators' ability to make real-time assessments of state-of-health (SoH). We present a machine learning (ML) framework that performs virtual electrochemical characterization during normal operation. The method uses an encoder-decoder transformer, conditioned on operational data, to reconstruct characterization outputs, focusing here on polarization curves. Inspired by patch-based sequence tokenization, we segment the inputs into patches and encode them to form meaningful tokens, which substantially improves learning efficiency. Across four longitudinal runs, lasting up to 478 hours on different test cells and loading cycles, the model accurately reconstructed polarization curves and achieved 10x reduction in mean squared error (MSE) compared to a vanilla transformer. This proof-of-concept demonstrates that ML models can enable continuous performance monitoring for PEM electrolyzers and that the encoder captures meaningful latent representations of SoH, opening up opportunities to derive interpretable indicators in future work.

Abstract

Explore similar work

Performance Monitoring of Proton Exchange Membrane Water Electrolyzer by Transformers-Based Machine Learning Model

BatteryMFormer: Multi-level Learning for Battery Degradation Trajectory Forecasting

Causal Anomaly Detection for Lithium-Ion Battery Degradation