Original title: Estimating the Health and State of Charge of Each Cell in a Second-Life Battery System from Field Data
Martin Cornejo, Julian Meyer-Schwickerath, Juan Victor Sandalinas, Andreas Jossen · September 2026
Old batteries get a second life in energy storage systems, but we often do not know their health or charge. Standard tracking models need a base voltage curve that is missing for used cells. Researchers created a new framework to solve this. It learns the missing curve directly from everyday field operation data using a statistical tool called Gaussian process regressiona machine learning method that predicts a smooth curve from scattered measurements, also giving a sense of uncertainty. At the same time, it tracks the status of an electrical model for each cell. When tested on a real system, the method spotted uneven health, a built-in charge imbalance, and two faulty cells. Looking only at whole modules rather than individual cells misses weak spots, dangerously overestimating health by up to 31% and remaining charge by up to 23%.
Imagine judging a team's strength by averaging everyone's weight. A strong member and a weak one balance out, but the result hides that the weak one cannot carry their share. In this paper the battery pack is the team, each cell is a member, and the lumped model is the average weight. The framework measures each cell individually, revealing the weak ones that limit the whole pack.
Second-life batteries are a cheap way to store energy, but their cells wear out at different rates. Using a single average model for the whole pack can overestimate how much energy is left, risking sudden failures. This work shows how to track each cell from everyday usage data, catch weak cells early, and use the pack more safely.
The abstract tests the framework on only one second-life battery system, so its performance on other chemistries or pack sizes is unknown. It does not detail the reference measurement used for validation, nor does it discuss the computational cost of applying Gaussian process regression to each of many cells.
Ask anything about this paper. Answers come from its abstract, and say so when the abstract doesn't cover your question.
Estimating the Health and State of Charge of Each Cell in a Second-Life Battery System from Field Data — Martin Cornejo, Julian Meyer-Schwickerath, Juan Victor Sandalinas et al.
Abstract licence: CC0
Explanation generated by Synthesis from the abstract — it is not the original paper. This is a preprint, so it hasn't necessarily been peer reviewed.