TU Darmstadt contributes to international study on physically informed machine learning for battery diagnostics

Researchers from the Control and Cyber-Physical Systems Laboratory at TU Darmstadt — Dr. Joachim Schaeffer and Prof. Dr.-Ing. Rolf Findeisen — have contributed to a study published in Energy & Environmental Science, one of the leading international journals in energy research: https://doi.org/10.1039/d5ee06439b

The work was carried out in an international collaboration spanning the Massachusetts Institute of Technology (MIT), Stanford University, and TU Darmstadt. It is based on a longstanding scientific collaboration between the Braatz and Bazant research groups at MIT and the Control and Cyber-Physical Systems Laboratory at TU Darmstadt.

The study addresses a central challenge for sustainable electrification: how to reliably assess internal battery degradation and predict lifetime without relying on time-consuming laboratory procedures.

The research team developed a modelling framework that combines physical battery simulation with physically informed and interpretable machine learning methods. This approach enables faster and more reliable identification of internal aging mechanisms and improves early prediction of battery lifetime — an important step toward safer electric mobility and more efficient renewable energy storage.

TU Darmstadt’s contribution focused on structured residual modelling with machine learning, systematically enhancing mechanistic simulation models while preserving physical interpretability and reliability. For his work in this area, Dr. Joachim Schaeffer has previously received an MIT Open Data Award.

The study highlights the role of simulation-based, data-driven methods and machine learning at the interface of materials science and modelling, and control engineering — a key research strength within TU Darmstadt and the Rhine-Main Universities (RMU) alliance.

Beyond battery diagnostics, the work reflects a broader control-theoretic perspective: combining mechanistic models with structured, physics-informed machine learning enables reliable decision-making in safety-critical physical systems. By explicitly preserving physical constraints and interpretability, the approach contributes to the development of safe learning-based monitoring and control strategies — an important direction in the research agenda of the Control and Cyber-Physical Systems Laboratory.

Abstract

Reliable battery health diagnosis and cycle life prediction remain a central challenge for energy storage systems. This work first provides a systematic analysis of key factors for battery health diagnosis, highlighting previously overlooked yet critical elements that affect health assessments. Building on these insights, a rate-adaptive transformation model converts high C-rate features into low C-rate equivalents, enabling rapid diagnostics of battery aging modes without time-consuming testing using a low C-rate. To address fitting inaccuracies caused by aging, blended materials, and kinetic effects, an interpretable residual learning model corrects voltage mismatches, which also enables low C-rate fitting by using high C-rate data. Leveraging mechanistic-informed features, early cycle life prediction achieves mean errors of less than 70 cycles using data from fewer than 30 equivalent full cycles across complex and unseen aging conditions. This interpretable and generalizable framework bridges electrochemical understanding with practical diagnosis and offers a fast and reliable path toward mechanism-informed battery prognostics.

DOI: https://doi.org/10.1039/d5ee06439b