Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
The authors benchmark a spectrum of time-series architectures—spanning linear baselines, CNNs, and LSTMs—for medium-term cell voltage degradation forecasting using a novel, multi-campaign Anion Exchange Membrane water electrolyzer dataset. Accurate degradation modeling under non-stationary industrial conditions is essential for extending the operational lifetime and commercial viability of green hydrogen systems. The resulting evaluation framework systematically reveals the performance and efficiency trade-offs of deep temporal models when subjected to real-world cross-campaign distribution shifts.
Standard temporal architectures can reliably forecast electrochemical cell degradation across disparate operational campaigns, proving that purely data-driven sequence models can overcome noisy, real-world physical shifts without complex hand-crafted priors.
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.