Early Detection Makes Batteries Safer

TU Darmstadt and MIT Develop Lithium-Ion Battery Monitoring Methods Using Machine Learning.

Lithium-ion batteries hold major significance, particularly due to their use in electric vehicles and stationary energy storage systems. For safe operation of these batteries, state monitoring and early fault detection are critical. Researchers from TU Darmstadt and Massachusetts Institute of Technology (MIT) have developed new methods for battery analysis and monitoring systems using physics-informed machine learning approaches.

Joachim Schaeffer, Eric Lenz and Prof. Rolf Findeisen from the Institute for Automation Technology and Mechatronics at TU Darmstadt, working together with teams led by Prof. Richard Braatz and Prof. Martin Bazant at MIT, have developed a method that combines physical approaches with machine learning to detect time-dependent and operational changes in battery cells. This method, using recursive Gaussian processes, efficiently processes large quantities of data and enables real-time monitoring, allowing continuous observation of battery systems.

In this research, the researchers utilized a unique data set: an anonymized data set obtained from 28 battery systems containing more than 133 million data rows in total. While this data set provides in-depth information about the operation of battery cells, the findings enable better understanding of battery aging processes and fault conditions. These findings can also enhance safety by enabling continuous monitoring of battery systems in the future.

The study was published in the journal Cell Reports Physical Science with its results. The study demonstrates how a failure in a single cell can affect the entire system and creates an important reference for safe battery use. The open-access data set created during the project won Joachim Schaeffer, doctoral student, the MIT Open Data Award. Schaeffer was selected to receive one of ten awards chosen from over 70 applications for this award.

These methods aim to enhance the reliability of future energy storage solutions as a major step forward in battery monitoring and safe use.

Source: https://www.chemeurope.com/en/news/1184813/early-detection-makes-batteries-safer.html