On June 19, 2024, Victoria O’Brien, an Electrical Engineer in the Sandia National Laboratories Energy Storage program,presented the results of the paper entitled “A Comparison of Online Model-Based Anomaly Detection Methods for a Lithium-Ion Battery Cell” at the 2024 International Symposium on Power Electronics, Electrical Drives, Automation, and Motion (SPEEDAM), in Ischia, Italy. The paper, authored by Victoria O’Brien and Rodrigo Trevizan, compared the ability of four anomaly detection methods to detect bias anomalies corrupting the voltage sensor of a battery cell. The study found a recursive, statistics-based cumulative sum algorithm to be the most effective anomaly detector due to its detection speed, accuracy, and ability to classify the bias of the anomaly as positive or negative.
Sensor anomaly detection is a critically important issue in battery energy storage system safe operation, as battery management systems equipped with estimation functionalities could be corrupted by sensor anomalies. Improper estimation has been found to cause battery degradation, equipment malfunctions, and thermal events. The research in this paper compared four methods that could be deployed in battery management systems to detect sensor anomalies, which is a critical first step in mitigating the damage from these hazardous events. The researchers found the Cumulative Sum Method to be the best option of the four methods studied to detect anomalies distorting voltage measurements on a Lithium-ion battery cell.
The 27th International SPEEDAM was a large, diverse conference with over 200 attendees originating from more than 30 countries. The paper, “A Comparison of Online Model-Based Anomaly Detection Methods for a Lithium-Ion Battery Cell,” was presented in the special session Advances in Grid Energy Storage Solutions.
This material is based upon work supported by the U.S. Department of Energy, Office of Electricity (OE), Energy Storage Division.