System for predicting lifetime of battery

Patent metadata

Field	Value
Publication number	US-9316699-B2
Application number	US-201213532148-A
Country	US
Kind code	B2
Filing date	Jun 25, 2012
Priority date	Apr 5, 2012
Publication date	Apr 19, 2016
Grant date	Apr 19, 2016

Abstract

A system for predicting a lifetime of a battery cell, including a learning data input unit, the learning data input unit being configured to receive at least one learning measurement factor and at least one learning factor, a target data input unit, the target data input unit being configured to receive at least one target factor, a machine learning unit, the machine learning unit being coupled to the learning data input unit, the machine learning unit assigning weights to respective ones of the learning factors input to the learning data input unit, and a lifetime prediction unit, the lifetime prediction unit being coupled to the target data input unit and the machine learning unit, the lifetime prediction unit using the weights assigned by the machine learning unit to predict one or more characteristics indicative of the lifetime of the target battery cell.

First claim

What is claimed is:

1. A system for predicting a lifetime of a battery cell, the system comprising:
a learning data input unit, the learning data input unit being configured to receive at least one learning measurement factor and at least one learning factor, the learning measurement factor and the learning factor associated with a learning battery cell that was previously manufactured;
a target data input unit, the target data input unit being configured to receive at least one target factor, the target factor associated with a target battery cell;
a machine learning unit, the machine learning unit being coupled to the learning data input unit, the machine learning unit to generate a prediction function by assigning weights to respective ones of the learning factors input to the learning data input unit based on the learning measurement factor and the learning factor from the learning data input unit; and
a lifetime prediction unit, the lifetime prediction unit being coupled to the target data input unit and the machine learning unit, the lifetime prediction unit to predict a lifetime of the target battery by using the target factor in association with the target battery and the prediction function generated by the machine learning unit before manufacturing the target battery cell, wherein:
the learning data input unit is configured to receive at least one learning process factor, the learning process factor being indicative of a process parameter used during manufacture of the learning battery cell.

2. The system as claimed in claim 1 , wherein the learning measurement factor is obtained by actual measurement of a characteristic value of the learning battery cell.

3. The system as claimed in claim 2 , wherein the learning measurement factor is selected from the group of a change in a capacity of the learning battery cell depending on a number of cycles and a change in a thickness of the learning battery cell depending on a number of cycles.

4. The system as claimed in claim 3 , wherein one cycle consists of one charge, one discharge, and one idle time, the idle time being a time between the charge and the discharge or a time between the charge and/or discharge and a next charge and/or discharge.

5. The system as claimed in claim 1 , wherein the at least one learning factor is selected from the group of a learning design factor, the learning process factor, and a learning formation factor.

6. The system as claimed in claim 5 , wherein the learning data input unit is configured to receive at least one learning design factor, the learning design factor being indicative of a design parameter of the learning battery cell.

7. The system as claimed in claim 6 , wherein the learning design factor is selected from the group of a capacity of the learning battery cell, an energy density of the learning battery cell, a thickness of the learning battery cell, a length of the learning battery cell, a width of the learning battery cell, a current density of the learning battery cell, a slurry concentration of the learning battery cell, an electrode thickness of the learning battery cell, a loading level of the learning battery cell, a form factor of the learning battery cell, a width of a separator of the learning battery cell, a thickness of the separator of the learning battery cell, a kind of the separator of the learning battery cell, a presence or absence of separator coating on the separator of the learning battery cell, a number of windings of an electrode plate of the learning battery cell, a number of windings of the separator of the learning battery cell, an adhesion between an electrode plate and the separator of the learning battery cell, a type of electrolyte used, an electrolyte composition of the learning battery cell, an electrolyte amount of the learning battery cell, a kind of additive of the learning battery cell, an amount of additive of the learning battery cell, a discharge rate (C-rate) of the learning battery cell, a porosity of the learning battery cell, a thickness of a current collector of the learning battery cell, a strength of the current collector of the learning battery cell, a thickness of a pouch of the learning battery cell, a physical property value of an active material of the learning battery cell, and a physical property value of a binder material of the learning battery cell.

8. The system as claimed in claim 6 , wherein the learning design factor is known prior to manufacture of the learning battery cell.

9. The system as claimed in claim 1 , wherein the learning process factor is selected from the group of a winding tension of a component of the learning battery cell, a degassing and folding condition of the learning battery cell, and a tab welding method of the learning battery cell, the component being a separator or an electrode plate.

10. The system as claimed in claim 5 , wherein the learning data input unit is configured to receive at least one learning formation factor, the learning formation factor being indicative of a formation parameter of the learning battery cell following the assembly of the learning battery cell.

11. The system as claimed in claim 10 , wherein the learning formation factor is selected from the group of a temperature, a time, a charge and/or discharge current, a voltage, a cutoff condition, and a pressure associated with one or more selected from the group of aging, charging and/or discharging, and degassing and resealing the learning battery cell.

12. The system as claimed in claim 1 , wherein the at least one target factor is selected from the group of a target design factor, a target process factor, and a target formation factor.

13. The system as claimed in claim 1 , wherein the lifetime prediction unit is configured to predict a characteristic selected from the group of a change in a capacity of the target battery cell depending on a number of cycles and a change in a thickness of the target battery cell depending on a number of cycles.

14. The system as claimed in claim 1 , wherein:
the machine learning unit includes a number of machine learning subunits, and
the lifetime prediction unit includes a number of lifetime prediction subunits corresponding to the number of machine learning subunits.

15. The system as claimed in claim 14 , wherein each machine learning subunit assigns a weight based on a corresponding learning factor, the learning factor being selected from the group of a learning design factor, a learning process factor, and a learning formation factor.

16. The system as claimed in claim 15 , wherein each lifetime prediction subunit performs a prediction function using a weight assigned by a corresponding machine learning subunit.

17. The system as claimed in claim 1 , further comprising a lifetime indication unit, the lifetime indication unit being coupled to the lifetime prediction unit, the lifetime indication unit indicating a predicted lifetime of the target battery cell based on the one or more predicted characteristics.

18. A method of predicting a lifetime of a target battery cell, the method comprising:
establishing a prediction function, establishing the prediction function including:
receiving a first factor of a previously-manufactured learning battery cell and receiving a second factor of the learning battery cell, the first factor being determined from an historical measurement of a characteristic of the learning battery cell, the second factor corresponding to the manufacture of the learning battery cell and being selected from the group of a design factor of the learning batte

Assignees

• Baek Hoyul
• Shin Jeongsoon
• Park Cheol Ho
• Chai Heekwan
• Ko Younghwa
• Suh Seung Bum
• Cheong Kyeong-Beom
• Park Nokap
• Samsung Sdi Co Ltd

Inventors

• Baek Hoyul
• Shin Jeongsoon
• Park Cheol Ho
• Chai Heekwan
• Ko Younghwa
• Suh Seung Bum
• Cheong Kyeong-Beom
• Park Nokap

Classifications

• H01M10/48

  Accumulators combined with arrangements for measuring, testing or indicating the condition of cells, e.g. the level or density of the electrolyte (constructional details of current conducting connections for detecting conditions inside cells or batteries, e.g. details of voltage sensing terminals, H01M50/569) · CPC title

• G01R31/3651

  Physics · mapped topic

• G01R31/3679 Primary

  Physics · mapped topic

• G06N5/00

  Computing arrangements using knowledge-based models · CPC title

• G01R31/36 Primary

  Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC] · CPC title