Method, device and system for estimating the charge state of a battery

Patent metadata

Field	Value
Publication number	US-10310020-B2
Application number	US-201414912582-A
Country	US
Kind code	B2
Filing date	Sep 10, 2014
Priority date	Sep 11, 2013
Publication date	Jun 4, 2019
Grant date	Jun 4, 2019

Abstract

A method for estimating the charge of a battery comprises: acquiring at least one time series of measurements of voltage across the terminals of the battery, and at least one other time series of measurements of another physical parameter of the battery or of its environment; determining an operating regime of the battery; choosing a regression model from among a predefined set of such models; and estimating the charge of the battery by applying the regression model to the time series of voltage measurements and to at least one other time series of measurements. A device for estimating the charge of a battery and a device for training regression models of the charge of a battery, adapted for the implementation of the method are provided. A system for estimating the charge of a battery comprising a device for estimating the charge and a device for training regression models is provided.

First claim

The invention claimed is:

1. A method for estimating a state of charge of a battery integrated into an energy-consuming device, the method comprising:
acquiring, by a voltage sensor integrated in a state of charge estimation device also integrated into said energy-consuming device, at least one time series of measurements of voltage across terminals of said battery;
acquiring, by a second sensor integrated in the state of charge estimation device, at least one time series of measurements of a second physical parameter of said battery or of an environment of said battery, wherein the second physical parameter is chosen from among: a current provided or absorbed by the battery, an internal temperature of the battery, an ambient or internal temperature, a mean of values of said at least one time series of voltage measurements, a charging or discharging status of said battery, a charging or discharging rate, an impedance of the battery, and a measurement of a state of health of the battery;
determining, by a processor integrated in the state of charge estimation device, as a function of said measurements, an operating regime of said battery;
choosing, by said processor, as a function of said operating regime, a non-linear regression model from among a predefined set of models; and
estimating the state of charge of said battery by applying said non-linear regression model to said time series of voltage measurements and to said at least one time series of measurements of said second physical parameter,
wherein at least one of the steps is implemented in real-time by the state of charge estimation device.

2. The method claim 1 , wherein said second physical parameter is a current provided or absorbed by the battery, and determining the operating regime is on the basis of at least one voltage value and one current value of said time series of measurements.

3. The method of claim 2 , wherein determining said operating regime of said battery is on the basis of a mean value of voltage across its terminals and of a mean value of current absorbed or provided by the battery by means of a correspondence table or function.

4. The method of claim 2 , wherein choosing the non-linear regression model comprises choosing a model as a function of said second physical parameter or a state of health of said battery.

5. The method of claim 1 , further comprising low-pass filtering or smoothing of said at least one said time series of measurements prior to estimating the state of charge of said battery.

6. The method of claim 1 , wherein said predefined set of models comprises at least one of a kernel regression model, a support vector regression model, and a relevance vector machine.

7. The method of claim 1 , further comprising:
determining whether said time series of measurements correspond to one or more parameters of the battery or of its environment not taken into account during construction of the non-linear regression models of said set; and
storing said one or more time series of measurements in a database.

8. The method of claim 7 , wherein storing said one or more time series of measurements comprises determining an item of information relating to a state of charge of said battery and associating said item of information with at least of said time series of measurements, the method further comprising:
excluding from said database at least one time series of measurements on the basis of a comparison between the state of charge item of information which is associated therewith and an item of information regarding state of charge of the battery determined subsequently.

9. The method of claim 1 , further comprising constructing the non-linear regression models of said set by training on the basis at least of the time series of measurements of voltage across the terminals of said battery and of at least one of the series of measurements of the second parameter of said battery or of its environment, and of corresponding reference values of the state of charge of said battery.

10. The method of claim 9 , further comprising:
determining whether said time series of measurements correspond to one or more parameters of the battery or of its environment not taken into account during the construction of the non-linear regression models of said set;
storing said one or more time series of measurements in a database; and
reconstructing by training the said or at least one said non-linear regression model, or constructing by training a new non-linear regression model of said set, by taking account of the one or more time series stored in said database while storing.

11. The method of claim 10 , wherein constructing the non-linear models and/or reconstructing comprises low-pass filtering or smoothing and sub-sampling of said time series of measurements.

12. The method of claim 10 , wherein constructing the non-linear regression models comprises calculating in non-real time coulometric estimators of the charge of said battery on the basis of a time series of measurements of current, and using said coulometric estimators as reference values of the state of charge of said battery for the construction or reconstruction by training of said non-linear regression models.

13. The method of claim 1 , further comprising:
constructing the non-linear regression models of said set by training on the basis at least of a plurality of time series of measurements of voltage across the terminals of said battery and of at least one other time series of measurements of the second physical parameter of said battery or of its environment, and of corresponding reference values of the state of charge of said battery, wherein the constructing is implemented in non-real time by a regression training device not integrated into said energy-consuming device.

14. A device for estimating state of charge of a battery, the device and the battery being integrated into an energy-consuming device, the device comprising:
at least one voltage sensor configured to measure a voltage across terminals of said battery;
at least second sensor configured to measure a second physical parameter of said battery or of its environment, wherein the second physical parameter is chosen from among: a current provided or absorbed by the battery, an internal temperature of the battery, an ambient or internal temperature, a mean of values of said at least one time series of voltage measurements, a charging or discharging status of said battery, a charging or discharging rate, an impedance of the battery, and a measurement of a state of health of the battery;
a memory configured to store a set of non-linear regression models of the state of charge of said battery; and
a processor programmed to implement a method for estimating the state of charge of the battery comprising: acquiring at least one time series of measurements of voltage across the terminals of said battery and as at least one other time series of measurements of the second physical parameter of said battery or of its environment;
determining, as a function of said measurements, an operating regime of said battery;
choosing, as a function of said operating regime, a non-linear regression model from among a predefined set of such models; and
estimating the state of charge of said battery by applying said non-linear regression model to said time series of voltage measurements and to said or to at least one said other time series of measurements;

wherein said method is executed with said sensors and said memory.

15. A system for estimating the state of charge of a battery comprising the state of charge estimation device of clai

Assignees

• Commissariat Energie Atomique

Inventors

• Biletska Krystyna
• Suard Frédéric
• Montaru Maxime

Classifications

• B60L2240/549

  Current · CPC title

• G01R31/3842

  combining voltage and current measurements · CPC title

• G01R31/387

  Determining ampere-hour charge capacity or SoC · CPC title

• B60L2240/547

  Voltage · CPC title

• G01R31/367 Primary

  Software therefor, e.g. for battery testing using modelling or look-up tables · CPC title