System and method for estimating long term characteristics of battery

• US 8,412,658 B2
• Filed: 09/12/2008
• Issued: 04/02/2013
• Est. Priority Date: 09/13/2007
• Status: Active Grant

• Current Assignee: LG Energy Solution Limited (LG Corporation)
• Original Assignee: LG Chem Limited (LG Corporation)
• Inventors: Song, Hyun-Kon, Cho, Jeong-Ju, Choo, Yeon-Uk, Son, Mi-Young, Lee, Ho-Chun
• Primary Examiner(s): Chaki, Kakali
• Assistant Examiner(s): Tran, Mai T
• Application Number:
  US12/678,094
• Publication Number:
  US 20100312733A1
• Time in Patent Office: 1,663 Days
• Field of Search: 706/25, 320/132, 324/426, 702/63, 702/64, 702/193
• US Class Current: 706/25
• 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
• CPC Class Codes: H01M 10/48 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 Class Codes: H01M 10/482 for several batteries or cells simultaneously or sequentially
• CPC Class Codes: Y02E 60/10 Energy storage using batteries

Abstract

A system includes a learning data input unit for receiving initial and long term characteristic learning data of a battery to be a learning object; a measurement data input unit for receiving initial characteristic measurement data of a battery to be an object for long term characteristic estimation; an artificial neural network operation unit for converting the learning data into first and second data structures, allowing an artificial neural network to learn the learning data based on each data structure, converting the measurement data into first and second data structures, and individually applying the learned artificial neural network corresponding to each data structure to calculate and output long term characteristic estimation data based on each data structure; and a long term characteristic evaluation unit for calculating an error of the estimation data of each data structure and determining reliability of the estimation data depending on error.

Claims

1. A system for estimating long term characteristics of a battery, comprising:

a learning data input unit for receiving initial characteristic learning data, obtained at a cycle conducted in an opening part of a repeated cycle charging/discharging process with respect to a battery to be a learning object and long term characteristic learning data, obtained at a cycle conducted in a latter part of the cycle charging/discharging process;

a measurement data input unit for receiving initial characteristic measurement data, obtained at a cycle conducted in an opening part of the cycle charging/discharging process with respect to a battery to be an object for estimation of long term characteristics;

an artificial neural network operation unit for converting the initial characteristic learning data and the long term characteristic learning data into first and second data structures, allowing an artificial neural network to learn the initial characteristic learning data and the long term characteristic learning data based on each data structure, converting the input initial characteristic measurement data into first and second data structures, and individually applying the learned artificial neural network corresponding to each data structure to calculate and output long term characteristic estimation data based on each data structure; and

a long term characteristic evaluation unit for calculating an error of the output long term characteristic estimation data of each data structure and determining reliability of the long term characteristic estimation data depending on the error.

2. The system according to claim 1,

wherein the learned artificial neural network based on each data structure has at least one neuron layer arranged in series, and

wherein the neuron layer converts an input vector into an output vector such that a bias vector and a weight matrix calculated by the learning of the artificial neural network are reflected on the input vector, the input vector on which the bias vector and the weight matrix are reflected is processed by a neuron transfer function, and then result of the neuron transfer function is output as an output vector.

3. The system according to claim 2,

wherein, in the serial arrangement of the neuron layer, the first neuron layer has an input vector composed of initial characteristic measurement data.

4. The system according to claim 1,

wherein the data relating to initial characteristics includes a charging characteristic variation data of a battery, measured in a battery activating process;

or a charging characteristic variation data, a discharging characteristic variation data, a thickness variation data, or an open circuit voltage variation data of a battery, obtained by measurement of initial cycle characteristics, and

wherein the data relating to long term characteristics includes a charging characteristic variation data, a discharging characteristic variation data, a thickness variation data, or an open circuit voltage variation data of a battery at predetermined long term cycles.

5. The system according to claim 1, further comprising an initial characteristic measurement sensor for measuring a charging characteristic of a battery put into an activating process and then outputting the measured charging characteristic as an initial characteristic measurement data,

wherein the measurement data input unit receives the initial characteristic measurement data from the initial characteristic measurement sensor.

6. The system according to claim 1, further comprising a display for receiving the long term characteristic estimation data calculated based on each data structure from the artificial neural network operation unit to display the long term characteristic estimation data in a graphic-user interface through a display device.

7. The system according to claim 1,

wherein, in case the error is less than a criterion value, the long term characteristic evaluation unit determines any one of the long term characteristic estimation data calculated based on each data structure or average data of the long term characteristic estimation data as a long term characteristic estimation data, and then outputs the long term characteristic estimation data.

8. The system according to claim 7,

wherein the long term characteristic evaluation unit determines long term characteristic quality of the battery by comparing the determined long term characteristic estimation data with a criterion long term characteristic data.

9. The system according to claim 8,

wherein the long term characteristic evaluation unit outputs a long term characteristic quality determination result of the battery in a graphic-user interface through a display device.

10. A method for estimating long term characteristics of a battery, comprising:

(a) receiving initial characteristic learning data and long term characteristic learning data, respectively obtained at a cycle conducted in an opening part and at a cycle conducted in an opening part of a repeated cycle charging/discharging process with respect to a battery to be a learning object;

(b) converting the received initial characteristic learning data and the received long term characteristic learning data into first and second data structures and then individually allowing an artificial neural network to learn based on each data structure;

(c) receiving initial characteristic measurement data, obtained at a cycle conducted in an opening part of the cycle charging/discharging process with respect to a battery to be an object for estimation of long term characteristics;

(d) converting the received initial characteristic measurement data into first and second data structures, then applying the learned artificial neural network corresponding to each data structure thereto, and then calculating and outputting long term characteristic estimation data based on each data structure; and

(e) calculating an error of the output long term characteristic estimation data based on each data structure and then determining reliability of the long term characteristic estimation data depending on the error.

11. The method according to claim 10,

wherein the learned artificial neural network based on each data structure has at least one neuron layer arranged in series, and

wherein, in the step (d), the process of applying the learned artificial neural network based on each data structure includes:

(d1) converting the initial characteristic measurement data into an input vector;

(d2) inputting the converted input vector into a first neuron layer of the neuron layer arrangement;

(d3) each neuron layer of the neuron layer arrangement reflecting a bias vector and a weight matrix calculated by the learning of the artificial neural network on the input vector and then processing the input vector by a neuron transfer function such that the input vector is converted into an output vector and then output; and

(d4) a last neuron layer of the neuron layer arrangement outputting a long term characteristic estimation vector as an output vector.

12. The method according to claim 10,

wherein the data relating to initial characteristics includes a charging characteristic variation data of a battery, measured in a battery activating process;

or a charging characteristic variation data, a discharging characteristic variation data, a thickness variation data, or an open circuit voltage variation data of a battery, obtained by measurement of initial cycle characteristics, and

wherein the data relating to long term characteristics includes a charging characteristic variation data, a discharging characteristic variation data, a thickness variation data, or an open circuit voltage variation data of a battery at predetermined long term cycles.

13. The method according to claim 10,

wherein the initial characteristic learning data is a charging characteristic variation data of the battery to be a learning object, put into a battery activating process, and the long term characteristic learning data is a charging characteristic variation data, a discharging characteristic variation data, a thickness variation data, or an open circuit voltage variation data of the battery to be a learning object at predetermined long term cycles, and

wherein the step (c) includes:

putting the battery to be an object for long term characteristic estimation into the battery activating process;

measuring a charging characteristic variation from the battery; and

receiving the measured charging characteristic variation data as an initial characteristic measurement data.

14. The method according to claim 10, further comprising:

visually displaying the long term characteristic estimation data.

15. The method according to claim 10, further comprising:

determining any one of the long term characteristic estimation data calculated based on each data structure or average data of the long term characteristic estimation data as a long term characteristic estimation data;

comparing the determined long term characteristic estimation data with a criterion long term characteristic data to determine long term characteristic quality of the battery.

16. The method according to claim 15, further comprising:

visually displaying a long term characteristic quality determination result of the battery.