Systems and methods for determining indica for remaining useful life of a battery
View Patent ↗Systems, methods, and other embodiments described herein relate to determining salient attributes of batteries for characterizing battery value. In one embodiment, a method includes acquiring usage information about batteries that indicates aspects about prior use of the batteries. The method includes analyzing, using a predictor, the usage information to generate battery indicators about at least remaining useful life (RUL) for separate ones of the batteries. The method includes generating correlation values that specify an extent of correlation between the usage information and the RUL. The method includes providing a correlating subset of the usage information according to the correlation values to identify which parameters of the usage information specify the RUL for the batteries.
1 . A correlation system for determining salient attributes of batteries for characterizing battery value, comprising:
one or more processors;
a memory communicably coupled to the one or more processors and storing:
a battery module including instructions that, when executed by the one or more processors, cause the one or more processors to:
acquire usage information about the batteries that indicates aspects about prior use of the batteries;
analyze, using a predictor, the usage information to generate battery indicators about at least remaining useful life (RUL) for separate ones of the batteries;
generate correlation values that specify an extent of correlation between the usage information and the RUL; and
provide a correlating subset of the usage information according to the correlation values to identify which parameters of the usage information specify the RUL for the batteries without using a model, wherein the instructions to provide the correlating subset include instructions to adapt a battery management system (BMS) of a vehicle using the correlating subset to facilitate efficient determination of the RUL and adapt use of a battery by the BMS.
2 . The correlation system of claim 1 , wherein the batteries are of a same defined model that share physical characteristics and operating characteristics, and
wherein the instructions to generate the correlation values include instructions to select one or more of the parameters from the usage information as the correlating subset according to a ranking of the parameters based, at least in part, on the correlation values.
3 . The correlation system of claim 1 , wherein the battery module includes instructions to analyze the usage information using the predictor including instructions to infer the RUL using the predictor with the usage information.
4 . The correlation system of claim 3 , wherein the battery module includes instructions to analyze the usage information including instructions to infer one or more missing indicators for partially complete portions of the usage information for a subset of the batteries using the predictor.
5 . The correlation system of claim 3 , wherein the predictor is a deep neural network (DNN) that is trained to infer the RUL and one or more missing indicators using supervised learning, and
wherein the battery module includes instructions to acquire the usage information including instructions to electronically retrieve the usage information from at least the battery management system (BMS).
6 . The correlation system of claim 1 , wherein the battery module includes instructions to generate the correlation values including instructions to calculate correlation coefficients for the parameters of the usage information using statistical functions including at least one of a linear correlation function, a nonlinear correlation function, and a monotonic correlation function.
7 . The correlation system of claim 1 , wherein providing the correlating subset includes modifying an electronic indicator to specify the RUL according to the correlating subset.
8 . The correlation system of claim 1 , wherein the batteries are automotive batteries for electric vehicles.
9 . A non-transitory computer-readable medium for determining salient attributes of batteries for characterizing battery value and including instructions that, when executed by one or more processors, cause the one or more processors to:
acquire usage information about the batteries that indicates aspects about prior use of the batteries;
analyze, using a predictor, the usage information to generate battery indicators about at least remaining useful life (RUL) for separate ones of the batteries;
generate correlation values that specify an extent of correlation between the usage information and the RUL; and
provide a correlating subset of the usage information according to the correlation values to identify which parameters of the usage information specify the RUL for the batteries without using a model, wherein the instructions to provide the correlating subset include instructions to adapt a battery management system (BMS) of a vehicle using the correlating subset to facilitate efficient determination of the RUL and adapt use of a battery by the BMS.
10 . The non-transitory computer-readable medium of claim 9 , wherein the batteries are of a same defined model that share physical characteristics and operating characteristics, and
wherein the instructions to generate the correlation values include instructions to select one or more of the parameters from the usage information as the correlating subset according to a ranking of the parameters based, at least in part, on the correlation values.
11 . The non-transitory computer-readable medium of claim 9 , wherein the instructions to analyze the usage information using the predictor include instructions to infer the RUL using the predictor with the usage information.
12 . The non-transitory computer-readable medium of claim 11 , wherein the instructions to analyze the usage information include instructions to infer one or more missing indicators for partially complete portions of the usage information for a subset of the batteries using the predictor.
13 . The non-transitory computer-readable medium of claim 11 , wherein the predictor is a deep neural network (DNN) that is trained to infer the RUL and one or more missing indicators using supervised learning, and
wherein the instructions to acquire the usage information include instructions to electronically retrieve the usage information from at least the battery management system (BMS).
14 . A method, comprising:
acquiring usage information about batteries that indicates aspects about prior use of the batteries;
analyzing, using a predictor, the usage information to generate battery indicators about at least remaining useful life (RUL) for separate ones of the batteries;
generating correlation values that specify an extent of correlation between the usage information and the RUL; and
providing a correlating subset of the usage information according to the correlation values to identify which parameters of the usage information specify the RUL for the batteries without using a model, wherein providing the correlating subset includes adapting a battery management system (BMS) of a vehicle using the correlating subset to facilitate efficient determination of the RUL and adapt use of a battery by the BMS.
15 . The method of claim 14 , wherein the batteries are of a same defined model that share physical characteristics and operating characteristics, and
wherein generating the correlation values includes using the correlation values to select one or more of the parameters from the usage information as the correlating subset according to a ranking of the parameters based, at least in part, on the correlation values.
16 . The method of claim 14 , wherein analyzing the usage information using the predictor includes inferring the RUL using the predictor with the usage information, and
wherein the correlating subset of the usage information defines a subset of the parameters that indicate a direct correspondence with the RUL.
17 . The method of claim 16 , wherein analyzing the usage information includes inferring one or more missing indicators for partially complete portions of the usage information for a subset of the batteries using the predictor.
18 . The method of claim 16 , wherein the predictor is a deep neural network (DNN) that is trained to infer the RUL and one or more missing indicators using supervised learning, and
wherein acquiring the usage information includes electronically retrieving the usage information from at least the battery management system (BMS).
19 . The method of claim 14 , wherein generating the correlation values includes calculating correlation coefficients for the usage information using statistical functions including at least one of a linear correlation function, a nonlinear correlation function, and a monotonic correlation function.
20 . The method of claim 14 , wherein providing the correlating subset includes modifying an electronic indicator to specify the RUL according to the correlating subset; and
wherein the batteries are automotive batteries for electric vehicles.