Model-based prognostics for batteries which estimates useful life and uses a probability density function

• US 8,332,342 B1
• Filed: 11/19/2009
• Issued: 12/11/2012
• Est. Priority Date: 11/19/2009
• Status: Active Grant

• Current Assignee: USA As Represented By The Administrator Of The NASA
• Original Assignee: The United States of America As Represented By The Secretary of Agriculture
• Inventors: Saha, Bhaskar, Goebel, Kai F.
• Primary Examiner(s): VINCENT, DAVID ROBERT
• Application Number:
  US12/622,407
• Time in Patent Office: 1,118 Days
• Field of Search: 706/12, 706/45, 706/62
• US Class Current: 706/45
• CPC Class Codes: G01R 31/367 Software therefor, e.g. for battery testing using modelling or look-up tables
• CPC Class Codes: G01R 31/392 Determining battery ageing or deterioration, e.g. state of health

Abstract

This invention develops a mathematical model to describe battery behavior during individual discharge cycles as well as over its cycle life. The basis for the form of the model has been linked to the internal processes of the battery and validated using experimental data. Effects of temperature and load current have also been incorporated into the model. Subsequently, the model has been used in a Particle Filtering framework to make predictions of remaining useful life for individual discharge cycles as well as for cycle life. The prediction performance was found to be satisfactory as measured by performance metrics customized for prognostics for a sample case. The work presented here provides initial steps towards a comprehensive health management solution for energy storage devices.

Claims

1. A method for estimating remaining useful life (RUL) of a battery during discharge of the battery, the method comprising providing a computer system that is programmed:

to provide or receive a quantitative empirical model with at least one associated model parameter for at least one electro-chemical process that is active during discharge of the battery;

to receive and use measured values provided by one or more sensors for at least one electro-chemical process that is active during discharge of the battery;

to receive and use training data comprising at least one of;

at least one operating, condition for the battery, at least one sensor measurement value for battery operation, and at least one ground truth attribute for battery discharge;

to compute and incorporate at least one numerical parameter value for the electro-chemical process that characterizes battery discharge behavior;

to identify at least one uncertainty in the quantitative model, including an uncertainty range for the at least one model parameter and an uncertainty range for the at least one measured sensor value;

to provide and incorporate at least one numerical value for at least one probability density function (pdf) corresponding to a distribution of the at least one uncertainty;

to provide at least one process model of at least one process component with at least one estimate of an value of a probability density function (pdf) for a distribution of at least one uncertainty in the at least one process model, to provide a characterization of battery discharge;

to provide or receive run-time data, including the at least one battery operating condition and at least one sensor measurement value; and

to apply the quantitative model of the battery in a particle filtering framework to estimate at least one battery discharge variable of interest, comprising at least one of state of charge (SOC) and terminal voltage of the battery, and to contemporaneously modify the at least one model parameter value used in the quantitative model.

2. The method of claim 1, wherein said computer is further programmed to choose said battery operating condition from a group of conditions comprising battery terminal voltage, battery load current, and battery temperature.

3. The method of claim 1, wherein said computer is further programmed to incorporate ambient temperature in said at least one operating condition.

4. The method of claim 1, further comprising decomposing said battery discharge behavior into a plurality of sub-processes of said battery, with at least one sub-process comprising at least one of mass-transfer, battery self-discharge, and reactant depletion, with corresponding model parameters and uncertainty distributions.

5. The method of claim 1, wherein said computer is further programmed so that, when a user of said computer program indicates that prediction of said battery discharge behavior is desired, said computer is further programmed:

to provide or receive at least one anticipated future use condition of said battery;

to provide or receive at least one said pdf that is associated with said at least one uncertainty distribution in said at least one sensor measurement value associated with operation of said battery;

to propagate said at least one battery discharge variable of interest to a selected time later than present time or to at least one predetermined threshold value for said battery discharge variable of interest being propagated; and

to estimate said RUL of said battery during discharge, said RUL comprising time remaining until said battery reaches an end-of-discharge (EOD) state.

6. The method of claim 5, further comprising decomposing said battery discharge behavior into a plurality of sub-processes of said battery, with at least one sub-process comprising at least one of mass-transfer, self-discharge;

and reactant depletion, with corresponding model parameters and uncertainty distributions.

7. The method of claim 6, applied to a vehicle having partial or complete electric propulsion, where said future use conditions comprise at least one of desired destination, terrain information for a desired route, trajectory gradient information for the desired route, traffic information for the desired route, weather data, and expected temperature profile along the desired route.

8. The method of claim 7, further comprising using information on said EOD with at least one of said destination information, said terrain information, and said traffic information to provide at least one driving recommendation that may extend a time at which said EOD will occur, determined by performing a trade-off analysis of at least two load scenarios for said battery.

9. A method for estimating remaining useful life (RUL) of a battery over battery cycle life as the battery experiences a plurality of charge, discharge, and rest periods, the method comprising providing a computer system that is programmed:

to provide or receive a quantitative empirical model with at least one associated parameter to be determined from sensor measurements, wherein the basis for the form of the model is linked to at least one internal electro-chemical process of the battery that is active during at least one of charge, discharge and rest period of the battery;

to receive and use at least one of said measured values provided by one or more sensors for the at least one electro-chemical process that is active during the at least one of the charge, discharge and rest periods of the battery, to infer or estimate at least one numerical parameter value for the model;

to receive training data, comprising at least one of;

at least one operating condition for the battery, battery storage condition, at least one measured sensor value for battery operation, and at least one ground truth attribute for battery capacity;

to identify at least one uncertainty in the quantitative model, including an uncertainty range for at least one model parameter and an uncertainty range for at least one measured sensor value, and to initialize at least one probability density function (pdf) for a distribution of the at least one uncertainty;

to provide at least one process model of at least one process component with at least one estimate of the at least one uncertainty pdf to provide a characterization of battery ageing behavior;

to provide or receive run-time data, including the at least one battery operating condition, battery storage condition, and the at least one sensor measurement value for battery operation; and

to apply the quantitative model of the battery in a particle filtering framework to track or monitor at least one battery cycle life variable of interest, comprising at least one of state of life (SOL) and capacity of the battery, and to contemporaneously modify at least one model parameter value used in the quantitative model.

10. The method of claim 9, wherein said computer is further programmed to choose said at least one battery operating condition from at least one condition, comprising battery depth of discharge (DOD), battery charge current, and battery storage state of charge SOC.

11. The method of claim 10, wherein said battery storage includes at least first and second battery rest periods having different rest period lengths.

12. The method of claim 9, wherein said computer is further programmed to incorporate ambient temperature in said at least one operating condition.

13. The method of claim 9, wherein said battery ageing behavior is decomposed into a plurality of sub-processes of said battery, wherein at least one sub-process comprises at least one of capacity loss due to Coulombic efficiency factor and capacity recovery during rest, with corresponding model parameters and uncertainty distributions.

14. The method of claim 9, wherein said at least one battery cycle life variable of interest is based on a lumped parameter model, in which at least one component, comprising electrolyte resistance, Warburg resistance, charge transfer resistance, and dual-layer capacitance, is determined by electrochemical impedance spectroscopy (EIS) measurements.

15. The method of claim 9, wherein said computer is further programmed so that, when a user of said computer program indicates that prediction of battery life is desired, said computer is further programmed:

to provide or receive at least one anticipated future use condition of said battery, comprising at least one battery charge profile, at least one battery discharge profile, and at least one battery storage condition;

to provide or receive said at least one pdf that is associated with a distribution of said at least one uncertainty in said at least one sensor value associated with use of said battery;

to propagate said at least one battery cycle life variable of interest to a selected time later than present time or to at least one predetermined threshold value for said at least one battery cycle life variable of interest being propagated; and

to estimate said RUL of said battery, said RUL comprising a time remaining until said battery reaches an end-of-life (EOL) state.

16. The method of claim 15, further comprising decomposing battery ageing behavior into a plurality of sub-processes of said battery, with at least one of the sub-processes comprising at least one of capacity loss due to Coulombic efficiency factor, and capacity recovery during rest, with corresponding model parameters and uncertainty distributions.

17. The method of claim 15, wherein said at least one battery cycle life variable of interest is based on a lumped parameter model, in which at least one component, comprising electrolyte resistance, Warburg resistance, charge transfer resistance, and dual-layer capacitance, is determined by at least one EIS measurement.

18. The method of claim 15, applied to a vehicle with partial or complete electric propulsion, where said future use conditions comprise at least one of vehicle driving profiles, vehicle storage conditions and vehicle maintenance schedules.

19. The method of claim 18, wherein information on climate and terrain and said EO state are received or provided to formulate at least one of storage recommendation and maintenance recommendation that may extend a time at which said EOL state will occur, by performing a trade-off analysis of at least two use scenarios for said battery.

20. A system for estimating remaining useful life (RUL) of a battery during discharge of the battery, embodying a program on instructions executable by a computer, wherein the computer system is programmed:

to provide or receive a quantitative empirical model with at least one associated model parameter for at least one electro-chemical process that is active during discharge of the battery;

to receive and use measured values provided by one or more sensors for at least one electro-chemical process that is active during discharge of the battery;

to receive and use training data comprising at least one of;

at least one operating condition for the battery, at least one sensor measurement value for battery operation, and at least one ground truth attribute for battery discharge;

to compute and incorporate at least one numerical parameter value for the electro-chemical process that characterizes battery discharge behavior;

to identify at least one uncertainty in the quantitative model, including an uncertainty range for at least one model parameter and an uncertainty range for at least one measured sensor value;

to provide and incorporate at least one numerical value for at least one probability density function (pdf) corresponding to a distribution of the at least one uncertainty;

to provide at least one process model of at least one process component with at least one estimate of a value of a probability density function (pdf) for a distribution of at least one uncertainty in the at least one process model, to provide a characterization of battery discharge;

to provide or receive run-time data, including the at least one battery operating condition and the at least one sensor measurement value; and

to apply the quantitative model of the battery in a particle filtering framework to provide an estimate of at least one battery discharge variable of interest, comprising state of charge (SOC) and terminal voltage of the battery, and to contemporaneously modify at least one model parameter value used in the quantitative model.

21. The system of claim 20, wherein said computer system is further programmed to choose said battery operating condition from a group of conditions comprising battery terminal voltage, battery load current, and battery temperature.

22. The system of claim 20, wherein said computer system is further programmed to incorporate ambient temperature in at least one of said operating conditions.

23. The system of claim 20, wherein said computer system is further programmed to decompose said battery discharge behavior into a plurality of sub-processes of said battery, with at least one sub-process comprising at least one of mass-transfer, battery self-discharge, and reactant depletion, with corresponding model parameters and uncertainty distributions.

24. The system of claim 20, wherein when a user of said computer system indicates that prediction of said battery discharge behavior is desired, said computer system is further programmed:

to provide or receive at least one anticipated future use condition of said battery;

to provide or receive at least one said pdf that is associated with said at least one uncertainty distribution in said at least one sensor measurement value associated with operation of said battery;

to propagate said at least one battery discharge variable of interest to a selected time later than present time or to at least one predetermined threshold value for said battery discharge variable of interest being propagated; and

to estimate said RUL of said battery during discharge, said RUL comprising time remaining until said battery reaches an end-of-discharge (EOD) state.

25. The system of claim 24, wherein said computer system is further programmed to decompose said battery discharge behavior into a plurality of sub-processes of said battery, with at least one sub-process comprising at least one of mass-transfer, self-discharge, and reactant depletion, with corresponding model parameters and uncertainty distributions.

26. The system of claim 25, wherein said computer system is applied to a vehicle having partial or complete electric propulsion, where said future use conditions comprise at least one of desired destination, terrain information for a desired route, trajectory gradient information for the desired route, traffic information for the desired route, weather data, and expected temperature profile along the desired route.

27. The system of claim 26, wherein said computer system is further programmed to use information on said EOD with at least one of said destination information, said terrain information, and said traffic information to provide at least one driving recommendation that may extend a time at which said EOD will occur, determined by performing a trade-off analysis of at least two load scenarios for said battery.

28. A system for estimating remaining useful life (RUL) of a battery over battery cycle life as the battery experiences a plurality of charge, discharge, and rest periods, embodying a program on instructions executable by a computer, wherein the computer system is programmed:

to provide or receive a quantitative empirical model with at least one associated parameter to be determined from sensor measurements, wherein the basis for the form of the model is linked to at least one internal electro-chemical process of the battery that is active during at least one of charge, discharge and rest period of the battery;

to receive and use at least one measured value provided by one or more sensors for the at least one electro-chemical process that is active during the at least one of the charge, discharge and rest periods of the battery, to infer or estimate at least one numerical parameter value for the model;

to receive training data, comprising at least one of;

at least one operating condition for the battery, battery storage condition, at least one measured sensor value for battery operation, and at least one ground truth attribute for battery capacity;

to identify at least one uncertainty in the quantitative model, including an uncertainty range for at least one model parameter and an uncertainty range for at least one measured sensor value, and to initialize at least one probability density function (pdf) for a distribution of the at least one uncertainty;

to provide at least one process model of at least one process component with at least one estimate of the at least one uncertainty pdf to provide a characterization of battery ageing behavior;

to provide or receive run-time data, including at least one of battery operating condition, battery storage condition, and the at least one sensor measurement value for battery operation; and

to apply the quantitative model of the battery in a particle filtering framework to provide an estimate of at least one battery cycle life variable of interest, comprising state of life (SOL) and capacity of the battery, and to contemporaneously modify at least one model parameter value used in the quantitative model.

29. The system of claim 28, wherein said computer system is further programmed to choose said battery operating condition from at least one condition, comprising battery depth of discharge (DOD), battery charge current, and battery storage state of charge SOC.

30. The system of claim 29, wherein said battery storage includes at least first and second battery rest periods having different rest period lengths.

31. The system of claim 28, wherein said computer system is further programmed to incorporate ambient temperature in said at least one operating condition.

32. The system of claim 28, wherein said computer system is further programmed to decompose said battery ageing behavior into a plurality of sub-processes of said battery, wherein at least one sub-process comprises at least one of capacity loss due to Coulombic efficiency factor and capacity recovery during rest, with corresponding model parameters and uncertainty distributions.

33. The system of claim 28, wherein said computer system is further programmed so that said at least one battery cycle life variable of interest is based on a lumped parameter model, in which at least one component, comprising electrolyte resistance, Warburg resistance, charge transfer resistance, and dual-layer capacitance, is determined by electrochemical impedance spectroscopy (EIS) measurements.

34. The system of claim 28, wherein when a user of said computer system indicates that prediction of battery life is desired, said computer system is further programmed:

to provide or receive at least one anticipated future use condition of said battery, drawn from a group of conditions comprising at least one battery charge profile, at least one battery discharge profile, and at least one battery storage condition;

to provide or receive said at least one pdf that is associated with a distribution of said at least one uncertainty in said at least one sensor value associated with use of said battery;

to propagate said at least one variable of interest to a selected time later than present time or to at least one predetermined threshold value for said at least one battery cycle life variable of interest being propagated; and

to estimate said RUL of said battery, said RUL comprising a time remaining until said battery reaches an end-of-life (EOL) state.

35. The system of claim 34, wherein said computer system is further programmed to decompose said battery ageing behavior into a plurality of sub-processes of said battery, with at least one of the sub-processes comprising at least one of capacity loss due to Coulombic efficiency factor, and capacity recovery during rest, with corresponding model parameters and uncertainty distributions.

36. The system of claim 34, wherein said computer system is further programmed so that said at least one battery cycle life variable of interest is based on a lumped parameter model, in which at least one component, comprising electrolyte resistance, Warburg resistance, charge transfer resistance, and dual-layer capacitance, is determined by at least one EIS measurement.

37. The system of claim 34, wherein said computer system is applied to a vehicle with partial or complete electric propulsion, where said future use conditions comprise at least one of vehicle driving profiles, vehicle storage conditions and vehicle maintenance schedules.

38. The system of claim 37, wherein said computer system is further programmed so that information on climate and terrain and said EOL state are received or provided to formulate at least one of storage recommendation and maintenance recommendation that may extend a time at which said EOL state will occur by performing a trade-off analysis of at least two use scenarios for said battery.