Method and apparatus for estimating state of battery

Patent metadata

Field	Value
Publication number	US-10928456-B2
Application number	US-201815864405-A
Country	US
Kind code	B2
Filing date	Jan 8, 2018
Priority date	Aug 17, 2017
Publication date	Feb 23, 2021
Grant date	Feb 23, 2021

Abstract

Disclosed is a battery state estimation method and apparatus, the method includes extracting data from target intervals in sensing data of a battery, generating feature vectors of the data extracted from each of the target intervals, applying a weight to each of the generated feature vectors, merging the feature vectors to which the weight is applied, and determining state information of the battery based on the merging.

First claim

What is claimed is:

1. A method of estimating a state of a battery, the method comprising:
extracting data from target intervals in sensing data of a battery;
generating feature vectors of the data extracted from each of the target intervals;
applying first weights to the generated feature vectors;
merging the feature vectors to which the first weights are applied;
determining state information of the battery based on the merging, and
updating the state information in response to an occurrence of an update event of the state information,
wherein the updating comprises:
setting an additional target interval in the sensing data and extracting data from the additional target interval;
generating an additional feature vector indicating a feature vector of the data extracted from the additional target interval; and
updating the state information based on applying second weights to the generated additional feature vector and a portion of the Generated feature vectors.

2. The method of claim 1 , wherein the generating of the feature vectors comprises:
sampling the extracted data from the each of the target intervals; and
encoding the sampled data to generate the feature vectors.

3. The method of claim 1 , wherein the applying comprises:
calculating the first weights based on the generated feature vectors and a previous state information of the battery; and
applying the calculated first weights to the generated feature vectors.

4. The method of claim 1 , wherein the generating of the additional feature vector comprising generating the additional feature vector by encoding the data extracted from the additional target interval.

5. The method of claim 1 , further comprising:
randomly setting each of the target intervals in the sensing data.

6. The method of claim 1 , wherein lengths of the target intervals are different from one another.

7. The method of claim 1 , wherein the update event corresponds to any one of a user input or a time exceeding an update time period.

8. The method of claim 1 , wherein a greatest first weight of the first weights is applied to a feature vector associated with a target interval having a most stable pattern change from among the target intervals.

9. A non-transitory computer-readable storage medium storing instructions that, when executed by a processor, cause the processor to perform the method of claim 1 .

10. An apparatus for estimating a state of a battery, the apparatus comprising:
a controller configured to extract data from target intervals in sensing data of a battery, to generate feature vectors of the data extracted from each of the target intervals, to apply first weights to the generated feature vectors, to merge the feature vectors to which the first weights are applied, to determine state information of the battery based on the merged feature vectors, and to update the state information in response to an occurrence of an update event of the state information,
wherein, for updating the state information, the controller is further configured to set an additional target interval in the sensing data and extracting data from the additional target interval to generate an additional feature vector indicating a feature vector of the data extracted from the additional target interval, and update the state information based on applying second weights to the generated additional feature vector and a portion of the generated feature vectors.

11. The apparatus of claim 10 , wherein the controller is further configured to sample the extracted data from the each of the target intervals, to encode the sampled data, and to generate the feature vectors.

12. The apparatus of claim 10 , wherein the controller is further configured to calculate the first weights based on the generated feature vectors and a previous state information of the battery, and to apply the first calculated weights to the generated feature vectors.

13. The apparatus of claim 10 , wherein the controller is further configured to et generate the additional feature vector by encoding the data extracted from the additional target interval.

14. The apparatus of claim 10 , wherein the controller is further configured to randomly set each of the target intervals in the sensing data.

15. The apparatus of claim 10 , wherein lengths of the target intervals are different from one another.

16. The apparatus of claim 10 , wherein a greatest first weight of the first weights is applied to a feature vector associated with a target interval having a most stable pattern change from among the target intervals.

17. An apparatus for estimating a state of a battery, the apparatus comprising:
a controller configured to extract data from target intervals in sensing data of a battery, and to determine state information of the battery based on the extracted data and a state estimation model,
wherein the state estimation model comprises:
a first layer configured to generate feature vectors of the data extracted from each of the target intervals;
a second layer configured to apply first weights to the generated feature vectors and to merge the feature vectors to which the first weights are applied; and
a third layer configured to determine the state information of the battery based on the merged feature vectors, and
wherein, in response to an occurrence of an update event of the state information:
the controller is further configured to set an additional target interval in the sensing data and extract data from the additional target interval,
the first layer is further configured to generate an additional feature vector indicating a feature vector of the data extracted from the additional target interval,
the second layer is further configured to apply second weights to the generated additional feature vector and a portion of the generated feature vectors, and
the third layer is further configured to u date the state information based on applying the second weights to the generated additional feature vector and the portion of the generated feature vectors.

18. The apparatus of claim 17 , wherein the first layer is further configured to recognize a pattern change of each piece of the extracted data.

19. The apparatus of claim 17 , wherein the second layer is further configured to calculate the first weights based on the generated feature vectors and previous state information of the battery.

20. The apparatus of claim 17 , wherein the third layer is further configured to determine the state information by performing regression on the merged feature vectors.

21. A vehicle comprising:
a battery module;
sensors configured to sense data of the battery module; and
a battery state estimation apparatus implemented on a processor, the battery state estimation apparatus comprising:
an extractor configured to receive the sensed data, to set target intervals in the sensed data, and to extract data from each of the target intervals,
an encoder configured to generate feature vectors corresponding to each target interval based on encoding the extracted data from the each target interval, respectively,
a vector merger configured to apply first weights to the generated feature vectors, and to merge the weighted feature vectors, and
an estimator configured to determine state information of the battery module based on the merged feature vectors,

wherein, in response to an occurrence of an u date event of the state information:
the extractor is further configured to set an additional target interval in th

Assignees

• Samsung Electronics Co Ltd

Inventors

• You Kaeweon
• Park Sangdo
• Sung Young Hun

Classifications

• G06N3/045

  Combinations of networks · CPC title

• G06N3/044

  Recurrent networks, e.g. Hopfield networks · CPC title

• G06N3/0442

  characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU] · CPC title

• G06N3/09

  Supervised learning · CPC title

• B60L3/0046 Primary

  relating to electric energy storage systems, e.g. batteries or capacitors · CPC title