US 12,742,751 B2Granted

Ultrasonic non-destructive testing method and system for lithium precipitation in lithium-ion battery

AssigneesBeihang University, Shandong Nonmetallic Materials Institute

InventorsJing Rao, Yan Sun, Jian Duan, Lijun Xu

Life of this patent

Estimating the term…

Loading patent…

Abstract

An ultrasonic non-destructive testing method and system for detecting lithium precipitation in lithium-ion batteries are provided. The method includes the following steps: acquiring full matrix capture data of the lithium-ion battery using an ultrasonic phased array transducer; inputting the measured ultrasonic data into the initial fully convolutional network (FCN), which outputs a predicted ultrasonic velocity model; using the predicted model as the initial input for an acoustic wave equation that outputs reconstructed ultrasonic data; constructing a loss function using the deviation between the measured and reconstructed ultrasonic data, and iterating until the loss function converges. This process completes the training of the unsupervised FCN. The trained FCN then outputs an accurate reconstructed ultrasonic velocity model that represents the lithium precipitation in the battery.

Claims5

Dependent claims

An ultrasonic non-destructive testing (NDT) method for detecting lithium precipitation in a lithium-ion battery, comprising:

acquiring measured full matrix capture (FMC) data of the lithium-ion battery utilizing an ultrasonic phased array transducer;
inputting the measured FMC data into an initial fully convolutional network (FCN) to output a predicted ultrasonic velocity model;
inputting the predicted ultrasonic velocity model as an initial model into an acoustic wave equation, the acoustic wave equation outputs reconstructed ultrasonic data;
constructing a loss function based on a discrepancy between measured ultrasonic data and the reconstructed ultrasonic data, and performing iterations until the loss function converges, wherein training of an unsupervised FCN is completed to obtain a trained FCN;
and
the trained FCN outputting an accurate predicted ultrasonic velocity model, wherein the accurate predicted ultrasonic velocity model is indicative of the lithium precipitation in the lithium-ion battery,
wherein the loss function comprises a pixel-wise loss and a perceptual loss,
wherein the pixel-wise loss (Lpw) is expressed as:
L
pw
(
d
,
d
˜
)
=
λ
1
⁢
L
1
(
d
,
d
˜
)
+
λ
2
⁢
L
2
(
d
,
d
˜
)
,
wherein λ1 and λ2 denote two hyperparameters, L1 and L2 represent L1 norm and L2 norm, respectively, d represents the measured ultrasonic data, and {tilde over (d)} represents the reconstructed ultrasonic data,
wherein the perceptual loss Lpl is given by:
L
p
⁢
l
(
d
,
d
~
)
=
λ
3
⁢
L
1
(
∅
⁡
(
d
)
,
∅
⁡
(
d
˜
)
)
+
λ
4
⁢
L
2
(
∅
⁡
(
d
)
,
∅
⁡
(
d
˜
)
)
,
where λ3 and λ4 denote two hyperparameters and Ø represents a network architecture parameter of the FCN.

92 more elements

Dependencies are inferred from the claim text.

Facts

Application: 18/530,260Filed 6 Dec 2023
Priority: CN 202310797031.330 Jun 2023
Published: US 2025/0003928 A12 Jan 2025
Granted: 22 Sep 2026
Term ends: Not estimatedThe source records don’t support an estimate.

Classification10 codes

G01N 29/4418
By comparison: with a model, e.g. best-fit, regression analysis
G01N 29/07
Analysing solids: by measuring propagation velocity or propagation time of acoustic waves
G01N 29/11
Analysing solids: by measuring attenuation of acoustic waves
G01N 29/4427
By comparison: with stored values, e.g. threshold values