Modeling a change in battery degradation

1. A computer-implemented method for controlling a battery, comprising:

training parameters for a battery capacity prediction model based usage of similar batteries and capacity information for the respective similar batteries, wherein the model characterizes a capacity decay rate;

predicting future battery capacity for a battery under control based on the battery capacity prediction model and a present value of the battery capacity; and

controlling one or more operational parameters of the battery under control based on the predicted future battery capacity.

2. The method of claim 1 , wherein a usage pattern comprises a timestamp, an electric current, a voltage, and a temperature.

3. The method of claim 2 , wherein the usage pattern further comprises a state of charge or difference in state of charge.

4. The method of claim 1 , wherein the battery capacity prediction model comprises an expression for a current battery capacity y at a time t:

y≡−at b(y,θ) +c

wherein a is a gradient of a capacity degradation curve, c is an initial capacity retention rate, and b (y, θ) is a function that depends on a current battery capacity y and a trained parameter θ.

5. The method of claim 4 , wherein the function b(y, θ) has functional form b=θ 0 +θ 1 y , where θ 0 and θ 1 are learned parameters.

6. The method of claim 4 , wherein training parameters comprises learning the parameter a with a trajectory regression.

7. The method of claim 1 , wherein the battery capacity prediction model comprises an expression for a current battery capacity y at a time t:

y t ≅−at b(y t−Δt, θ) +c

wherein a is a gradient of a capacity degradation curve, c is an initial capacity retention rate, Δt is a small elapsed time, and b(y t−Δt , θ) is a function that depends on a previous battery capacity y and a trained parameter θ.

8. The method of claim 1 , wherein the model accounts for an initial increase in battery capacity over time followed by a decline in battery capacity that is proportional to a square root of time elapsed.

9. The method of claim 1 , wherein controlling one or more operational parameters of the battery comprises setting a maximum or minimum terminal voltage in accordance with usage patterns and the prediction.

10. The method of claim 1 , wherein controlling one or more operational parameters of the battery comprises setting a charging speed of the battery in accordance with usage patterns and the prediction.

11. A computer readable storage medium comprising a computer readable program for controlling a battery, wherein the computer readable program when executed on a computer causes the computer to perform the steps of:

training parameters for a battery capacity prediction model based usage of similar batteries and capacity information for the respective similar batteries, wherein the model characterizes a capacity decay rate;

predicting future battery capacity for a battery under control based on the battery capacity prediction model and a present value of the battery capacity; and

controlling one or more operational parameters of the battery under control based on the predicted future battery capacity.

12. A battery controller, comprising:

a processor configured to train parameters for a battery capacity prediction model based on usage of similar batteries and capacity information for the respective similar batteries, wherein the model characterizes a capacity decay rate, and further configured to predict future battery capacity for a battery under control based on the battery capacity prediction model and a present value of the battery capacity; and

a battery interface configured to control one or more operational parameters of the battery under control based on the predicted future battery capacity.

13. The battery controller of claim 12 , wherein a usage pattern comprises a timestamp, an electric current, a voltage, and a temperature.

14. The battery controller of claim 13 , wherein the usage pattern further comprises a state of charge or difference in state of charge.

15. The battery controller of claim 12 , wherein the battery capacity prediction model comprises an expression for a current battery capacity y at a time t:

i y≡−at b(y,θ) +c

wherein a is a gradient of a capacity degradation curve, c is an initial capacity retention rate, and b(y, θ) is a function that depends on a previous battery capacity y t−Δt and a trained parameter θ.

16. The battery controller of claim 15 , wherein the function b(y, θ)has functional form b=θ 0 +θ 1 y , where θ 0 and θ 1 are learned parameters.

17. The battery controller of claim 12 , wherein the battery capacity prediction model comprises an expression for a current battery capacity y at a time t:

y t ≅−at b(y t−Δt, θ) +c

wherein a is a gradient of a capacity degradation curve, c is an initial capacity retention rate, Δt is a small elapsed time, and b(y t−Δt , θ) is a function that depends on a previous battery capacity y and a trained parameter θ.

18. The battery controller of claim 12 , wherein the model accounts for an initial increase in battery capacity over time followed by a decline in battery capacity that is proportional to a square root of time elapsed.

19. The battery controller of claim 12 , wherein the battery interface is configured to set a maximum or minimum terminal voltage in accordance with usage patterns and the prediction.

20. The battery controller of claim 12 , wherein the battery interface is configured to set a charging speed of the battery in accordance with usage patterns and the prediction.