State-of-health models for lithium-silicon batteries
Systems and methods are provided for state-of-health models for lithium-silicon batteries. State-of-health (SOH) of a lithium-ion cell may be assessed, with the assessing including calculating the state-of-health (SOH) using an enhanced state-of-health (SOH) model, with the enhanced state-of-health (SOH) model using input data other than data provided directly by the lithium-ion cell. The input data includes at least data acquired during operation of the lithium-ion cell and/or data acquired during manufacturing and initialization of the lithium-ion cell or electrodes of the lithium-ion cell. The lithium-ion cell may be a silicon-dominant cell including a silicon-dominant anode with silicon >50% of active material of the anode, and the enhanced state-of-health (SOH) model may be configured based on one or more characteristics unique to silicon-dominant cells.
1. A method for managing a lithium-ion cell, the method comprising:
assessing state-of-health (SOH) of the lithium-ion cell, wherein:
the assessing comprises calculating the state-of-health (SOH) using an enhanced state-of-health (SOH) model, and
the enhanced state-of-health (SOH) model uses input data, wherein at least a portion of the input data is obtained from sources other than the lithium-ion cell, and wherein the input data comprises data acquired during operation of the lithium-ion cell and/or data acquired during manufacturing and initialization of the lithium-ion cell or electrodes of the lithium-ion cell; and
training the enhanced state-of-health (SOH) model using training data that is adaptively configured based on the lithium-ion cell.
2. The method of claim 1 , wherein the lithium-ion cell comprises a silicon-dominant cell comprising a silicon-dominant anode with silicon >50% of active material of the anode, and wherein the enhanced state-of-health (SOH) model is configured based on one or more characteristics unique to silicon-dominant cells.
3. The method of claim 1 , wherein the enhanced state-of-health (SOH) model is configured for use regardless of an operation mode of the lithium-ion cell.
4. The method of claim 1 , wherein the input data comprise time series history data relating to one or more parameters or metrics associates with the lithium-ion cell, the one or more parameters or metrics comprising one or more of voltage, current, and temperature.
5. The method of claim 1 , further comprising obtaining or generating at least a portion of the input data based on a predefined measurement sequence, including but not limited to: current pulse, open circuit voltage measurement, reference performance test, hybrid pulse power characterization test, and direct current internal resistance measurement.
6. The method of claim 5 , wherein the predefined measurement sequence comprises current pulse, open circuit voltage measurement, reference performance test, hybrid pulse power characterization test, and direct current internal resistance (DCIR) measurement.
7. The method of claim 1 , further comprising training the enhanced state-of-health (SOH) model using one or more machine learning (ML) algorithms.
8. The method of claim 7 , wherein the one or more machine learning (ML) algorithms comprise one or more of linear regression, logistic regression, nonlinear regression, decision tree ensemble methods, neural networks, recurrent neural networks, long-short-term memory networks, Gaussian process algorithms, Bayesian algorithms, support vector regression, K-nearest-neighbors regression, and graph neutral networks.
9. The method of claim 1 , further comprising generating the training data by operating one or more cells using randomized voltage, current, and temperature profiles.
10. The method of claim 1 , further comprising generating the training data by operating one or more cells using a standardized drive-cycle test, or a similar test that replicates operation of a cell in an actual use case.
11. A system comprising:
a lithium-ion cell; and
one or more circuits for managing operation of the lithium-ion cell, wherein the one or more circuits are configured to:
assess state-of-health (SOH) of the lithium-ion cell, wherein:
the assessing comprises calculating the state-of-health (SOH) using an enhanced state-of-health (SOH) model, and
the enhanced state-of-health (SOH) model uses input data, wherein at least a portion of the input data is obtained from sources other than the lithium-ion cell, and wherein the input data comprises data acquired during operation of the lithium-ion cell and/or data acquired during manufacturing and initialization of the lithium-ion cell or electrodes of the lithium-ion cell; and
train the enhanced state-of-health (SOH) model using training data that is adaptively configured based on the lithium-ion cell.
12. The system of claim 11 , wherein the lithium-ion cell comprises a silicon-dominant cell comprising a silicon-dominant anode with silicon >50% of active material of the anode, and wherein the enhanced state-of-health (SOH) model is configured based on one or more characteristics unique to silicon-dominant cells.
13. The system of claim 11 , wherein the enhanced state-of-health (SOH) model is configured for use regardless of an operation mode of the lithium-ion cell.
14. The system of claim 11 , wherein the input data comprises time series history data relating to one or more parameters or metrics associates with the lithium-ion cell, the one or more parameters or metrics comprising one or more of voltage, current, and temperature.
15. The system of claim 11 , wherein the one or more circuits are configured to obtain or generate at least a portion of the input data based on a predefined measurement sequence, including but not limited to: current pulse, open circuit voltage measurement, reference performance test, hybrid pulse power characterization test, and direct current internal resistance measurement.
16. The system of claim 15 , wherein the predefined measurement sequence comprises current pulse, open circuit voltage measurement, reference performance test, hybrid pulse power characterization test, and direct current internal resistance (DCIR) measurement.
17. The system of claim 11 , wherein the one or more circuits are configured to train the enhanced state-of-health (SOH) model using one or more machine learning (ML) algorithms.
18. The system of claim 17 , wherein the one or more machine learning (ML) algorithms comprise one or more of linear regression, logistic regression, nonlinear regression, decision tree ensemble methods, neural networks, recurrent neural networks, long-short-term memory networks, Gaussian process algorithms, Bayesian algorithms, support vector regression, K-nearest-neighbors regression, and graph neutral networks.
19. The system of claim 11 , wherein the one or more circuits are configured to generate the training data by operating one or more cells using randomized voltage, current, and temperature profiles.
20. The system of claim 11 , wherein the one or more circuits are configured to generate the training data by operating one or more cells using a standardized drive-cycle test, or a similar test that replicates operation of a cell in an actual use case.