Lithium ion battery lifetime prediction method, discharge capacity retention rate prediction method, lifetime prediction program, discharge capacity retention rate prediction program, and information processing device

Patent metadata

Field	Value
Publication number	US-12540978-B2
Application number	US-202118248429-A
Country	US
Kind code	B2
Filing date	Oct 12, 2021
Priority date	Oct 15, 2020
Publication date	Feb 3, 2026
Grant date	Feb 3, 2026

Abstract

A lithium ion battery lifetime prediction method executes, by a computer, acquiring training data including cycle measurement data and lifetime data of a battery, learning a lifetime prediction model using the training data with respect to one or more cycle numbers at which a prediction is made, to acquire a set of learned lifetime prediction models corresponding to the cycle numbers at which the prediction is made, respectively, successively acquiring cycle measurement data for prediction of a battery that is a prediction target, up to the cycle numbers at which the prediction is made, respectively, and inputting the cycle measurement data for prediction acquired up to the cycle numbers at which the prediction is made, to the learned lifetime prediction models of the corresponding cycle numbers at which the prediction is made, and acquiring a probability distribution of a lifetime at the cycle numbers at which the prediction is made, respectively, as an output.

First claim

The invention claimed is:

1 . A computer-implemented lithium ion battery lifetime prediction method comprising:
acquiring training data including cycle measurement data and lifetime data of a battery; training a lifetime prediction model using the training data with respect to one or more cycle numbers at which a prediction is made, to acquire a set of trained lifetime prediction models corresponding to the cycle numbers at which the prediction is made, respectively; successively acquiring cycle measurement data for prediction of a battery that is a prediction target, up to the cycle numbers at which the prediction is made, respectively; and inputting the cycle measurement data for prediction acquired up to the cycle numbers at which the prediction is made, to the trained lifetime prediction models of the corresponding cycle numbers at which the prediction is made, to acquire a probability distribution of a lifetime at the cycle numbers at which the prediction is made, respectively, as an output, wherein the acquiring the training data performs a charge and discharge cycle test including a constant current charging process and a subsequent constant current discharge process with respect to each cell of the battery, to acquire the training data, including the cycle measurement data recorded with a voltage applied to each cell of the battery, a current flowing through each cell of the battery, and a current capacity, and the lifetime data.

2 . The computer-implemented lithium ion battery lifetime prediction method as claimed in claim 1 , wherein the lifetime prediction model includes
a data shaping part configured to shape the cycle measurement data into fixed length data, a feature extraction part configured to compresses the fixed length data into compressed data, a nonlinear conversion part configured to convert the compressed data into nonlinear feature data mapped in a high dimensional space, and a regression part configured to receive the nonlinear feature data as an input, and to output the probability distribution of the lifetime, that are coupled in this order.

3 . The computer-implemented lithium ion battery lifetime prediction method as claimed in claim 2 , wherein the feature extraction part uses a dimension reduction technique as a data compression technique.

4 . The computer-implemented lithium ion battery lifetime prediction method as claimed in claim 3 , wherein the dimension reduction technique is a principal component analysis.

5 . The computer-implemented lithium ion battery lifetime prediction method as claimed in claim 3 , wherein:
the data shaping part receives cycle measurement data An i for training, amounting to initial ni cycles extracted from the cycle measurement data of each cell of the battery, generates the fixed length data by arranging a value of each measured item of each cycle in a direction in which a number of columns increases, and arranges the generated fixed length data in a direction in which a number of rows increases for a number of cells of the battery, to acquire the fixed length data having the matrix format, where i is an integer greater than or equal to 1, and the feature extraction part compresses the fixed length data having the matrix format into the compressed data.

6 . The computer-implemented lithium ion battery lifetime prediction method as claimed in claim 1 , wherein the lifetime prediction model comprises a data shaping part that resamples current capacity data at voltage sampling points within a specific voltage range for each cycle, and produces data having a matrix format specifically tailored for principal component analysis and Gaussian process regression.

7 . A computer-implemented discharge capacity retention prediction method for a lithium ion battery, comprising:
acquiring training data including cycle measurement data of a battery and a discharge capacity retention rate in each cycle; learning a discharge capacity retention rate in each cycle using the training data, with respect to one or more cycle numbers at which a prediction is made, to acquire a set of trained discharge capacity retention rate prediction models corresponding to the cycle numbers at which the prediction is made, respectively; successively acquiring cycle measurement data for prediction of a battery that is a prediction target, up to the cycle numbers at which the prediction is made, respectively; and inputting the cycle measurement data for prediction acquired up to the cycle numbers at which the prediction is made, to the trained discharge capacity retention rate prediction models of the corresponding cycle numbers at which the prediction is made, to acquire a probability distribution of a discharge capacity retention rate at cycles from the cycle numbers at which the prediction is made to a lifetime, respectively, as an output, wherein the acquiring the training data performs a charge and discharge cycle test including a constant current charging process and a subsequent constant current discharge process with respect to each cell of the battery, to acquire the training data, including the cycle measurement data recorded with a voltage applied to each cell of the battery, a current flowing through each cell of the battery, and a current capacity, and the lifetime data.

8 . The computer-implemented discharge capacity retention prediction method for the lithium ion battery as claimed in claim 7 , wherein:
the learning the discharge capacity retention rate trains the lifetime prediction model, using cycle measurement data Ani for training, amounting to initial n i cycles extracted from the cycle measurement data of each cell of the battery of the training data, as an explanatory variable, and a discharge retention capacity rate B′ in each cycle of each cell, as a response variable, with respect to the one or more cycle numbers ni at which the prediction is made, to acquire a set of trained discharge capacity retention rate prediction models C′n i corresponding to the cycle numbers ni at which the prediction is made, respectively, where i is an integer greater than or equal to 1, the successively acquiring the cycle measurement data performs a charge and discharge cycle test including a constant current charging process and a subsequent constant current discharge process with respect to each cell of the battery that is the prediction target, for cycle numbers n x at which the prediction is made, respectively, to acquire the cycle measurement data for prediction recorded with a voltage applied to each cell of the battery that is the prediction target, a current flowing through each cell of the battery that is the prediction target, and a current capacity, where x is an integer greater than or equal to 1 , and the inputting the cycle measurement data inputs the cycle measurement data for prediction acquired up to the cycle numbers nx at which the prediction is made, to the trained discharge capacity retention rate prediction models C′n i of the corresponding cycle numbers n i at which the prediction is made, and acquires the probability distribution of the discharge capacity retention rate at the cycle numbers nx at which the prediction is made, respectively, as the output, where i=x.

9 . The computer-implemented discharge capacity retention prediction method for the lithium ion battery as claimed in claim 8 , wherein the lifetime prediction model includes:
a data shaping part that receives the cycle measurement data up to the cycle number ni at which the prediction is made, as the input, and outputs the fixed length data having the matrix format in which the number of rows is the number of cells, and the number of columns is the fixed length, a feature extraction part t

Assignees

• Resonac Corp

Inventors

• Kurauchi Yuji
• Minami Takuya
• Ito Yuji
• KOHNO Daisuke
• Inoue Hirofumi
• Okuno Yoshishige

Classifications

• G01R31/392

  Determining battery ageing or deterioration, e.g. state of health · CPC title

• G01R31/378

  specially adapted for the type of battery or accumulator · CPC title

• G01R31/367

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

• G01R31/3648 Primary

  comprising digital calculation means, e.g. for performing an algorithm · CPC title

• G06N20/10

  using kernel methods, e.g. support vector machines [SVM] · CPC title