System and method for determining a battery condition

1. A system for detecting a battery anomaly in a battery comprising:

at least one of:

a current sensor configured to measure a current dataset associated with the battery;

a voltage sensor configured to measure a voltage dataset associated with the battery; or

a temperature sensor configured to measure a temperature dataset associated with the battery;

a model operable to predict a local property dataset of the battery using the at least one of the current dataset, the voltage dataset, or the temperature dataset;

a classifier configured to classify whether the battery is operating in an anomalous condition based on the estimated local properties of the battery, wherein the classifier is operable in:

a steady state mode wherein the classifier is operable to classify the anomalous condition as a thermal anomaly or an electrical short using a local property dataset derived from steady state readings of the at least one of current, voltage, or temperature; and

a jump change mode wherein the classifier is operable to classify the anomalous condition as a battery cell disconnection or a loss of thermal connection using a local property dataset jump changes in the at least one of current, voltage, or temperature, wherein the jump changes are at least about 0.1 A/s to classify the battery cell disconnection.

2. The system of claim 1 , wherein the local property dataset comprises at least one of a capacity, a resistance, a local temperature, a battery variable that depends on a battery temperature, a current, or a state of charge.

3. The system of claim 2 , wherein the modeler is further configured to estimate the local property dataset of the battery based on a battery state determined using a state estimator processing the at least one of the current dataset, voltage dataset, or the temperature dataset.

4. The system of claim 2 , wherein:

the capacity is modelled as a function of the state of charge;

the resistance is modelled as a function of the state of charge, the local temperature, and a polarization of a battery cell of the battery;

the local temperature is modelled as a function of the temperature dataset, a historic temperature dataset, an ambient temperature, and a heat flux; and

wherein the state of charge is modelled as a function of at least one of a history of use of the battery, the current dataset, the voltage dataset, or the temperature dataset.

5. The system of claim 1 , wherein the classifier operates in the steady state mode or the jump change mode responsive to an inertial moment unit reading associated with the battery.

6. The system of claim 1 , further comprising a second classifier configured to determine whether the battery is operating in the anomalous state, wherein the classifier comprises a statistical test comprising at least one of a z-test, a student's t-test, or an F-test, wherein the second classifier comprises at least one of a Kalman filter or an auto regressive integrated average model.

7. The system of claim 1 , wherein the classifier is configured to classify whether the battery is operating in the anomalous condition during operation of the battery.

8. The system of claim 1 , wherein the classifier is configured to classify whether the battery is operating in the anomalous condition based on at most 100 estimated local parameters.

9. A method for determining whether a battery is operating in an anomalous condition comprising:

measuring a timeseries of sensor data comprising at least one of current, voltage, or temperature;

estimating a timeseries of a local property of the battery using the timeseries of sensor data comprising:

detecting a series of current jumps in the timeseries of sensor data, wherein each current jump of the series of current jumps exceeds a threshold current jump amplitude, wherein the threshold current jump amplitude is about 0.1 A/s; and

calculating a resistance of the battery at each of the current jumps of the series of current jumps; and

classifying whether the battery is operating in the anomalous condition using a statistical test of the timeseries of the local property of the battery;

wherein the anomalous condition comprises a cell disconnection within the battery.

10. The method of claim 9 , wherein the statistical test comprises at least one of a student t-test, a z-test, or an F-test.

11. The method of claim 9 , wherein the timeseries of the local property comprises at most about 100 local property values.

12. The method of claim 9 , wherein estimating the timeseries of the local property comprises estimating at least one of a battery cell capacity, a battery cell temperature, a battery resistance, a variable that depends on a temperature of the battery, the current, or a state of charge of the battery.

13. The method of claim 9 , further comprising estimating a battery state of the battery using a state estimator, wherein estimating the timeseries of the local property of the battery further uses the battery state.

14. The method of claim 12 , wherein:

the battery cell capacity is modelled as a function of the state of charge;

the resistance is modelled as a function of the state of charge, the local temperature, and a polarization of a battery cell of the battery;

the battery cell temperature is modelled as a function of the temperature dataset, a historic temperature dataset, an ambient temperature, and a heat flux; and

wherein the state of charge is modelled as a function of at least one of a history of use of the battery, the current dataset, the voltage dataset, or the temperature dataset.

15. The method of claim 9 , wherein estimating the timeseries of the local property comprises:

detecting a series of temperature jumps in the timeseries of sensor data, wherein each temperature jump of the series of temperature jumps exceeds a threshold temperature jump amplitude; and

calculating a resistance or a local temperature of the battery or a battery cell thereof at each of the temperature jumps of the series of temperature jumps;

wherein the anomalous condition comprises a thermal disconnection within the battery.

16. The method of claim 9 , wherein estimating the timeseries of the local property comprises:

detecting a steady state current draw or the battery is at rest; and

calculating a resistance or a local temperature of the battery or a battery cell thereof;

wherein the anomalous condition comprises a thermal anomaly proximal the battery.

17. The method of claim 9 , wherein estimating the timeseries of the local property comprises:

detecting when the battery is at rest; and

calculating a state of charge or a current draw while the battery is at rest;

wherein the anomalous condition comprises a short circuit within the battery.

18. The method of claim 9 , further comprising: selecting one or more models from a model library; where each model of the model library is operable to estimate the timeseries of a local property; wherein each model of the model library is associated with either an anomalous condition or a normal condition; wherein a statistical probability is assigned to each model of the model library based on the timeseries of sensor data or the time series of the local property; wherein when the statistical probability of a model associated with a failure mode has the greatest statistical probability, the battery is classified as operating in the anomalous condition.

19. The method of claim 18 , further comprising, when the battery is classified as operating in the anomalous condition, determining a corrective action.