US 11,989,631 B1
System and method for using artificial intelligence to detect lithium plating
Anil Ozturk, Istanbul (TR); Mustafa Burak Gunel, Istanbul (TR); Muharrem Ugur Yavas, Istanbul (TR); and Can Kurtulus, Istanbul (TR)
Assigned to ROM Technologies, Inc., Brookfield, CT (US)
Filed by Eatron Technologies Limited, Warwick (GB)
Filed on Mar. 24, 2023, as Appl. No. 18/189,660.
Application 18/189,660 is a continuation of application No. 18/184,305, filed on Mar. 15, 2023, granted, now 11,845,357.
Claims priority of provisional application 63/482,353, filed on Jan. 31, 2023.
This patent is subject to a terminal disclaimer.
Int. Cl. G06N 20/00 (2019.01); B60L 58/16 (2019.01); G01R 31/367 (2019.01); G01R 31/396 (2019.01); G06Q 10/30 (2023.01); H01M 10/44 (2006.01); H02J 7/00 (2006.01)
CPC G06N 20/00 (2019.01) [B60L 58/16 (2019.02); G01R 31/367 (2019.01); G01R 31/396 (2019.01); G06Q 10/30 (2013.01); H01M 10/441 (2013.01); H02J 7/0047 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A computer-implemented method comprising:
while a battery pack is charging, receiving, from one or more sensors, one or more measurements associated with the battery pack, wherein the battery pack comprises one or more cells;
separating the one or more measurements into separate profiles for the one or more cells, wherein the separate profiles include data pertaining to current, voltage, temperature, or some combination thereof;
identifying, using the separate profiles, one or more features;
generating a training dataset by reducing the one or more features based on a mean-comparison technique, a minority scaling technique, or both;
generating a trained machine learning model using the training dataset comprising the one or more reduced features as labeled input and one or more true lithium plating occurrence statuses as labeled output; and
predicting, using the trained machine learning model, an occurrence of lithium plating by inputting subsequently received data into the trained machine learning model.