# US 2024/0204269 A1 · Apparatus and method for diagnosing fault for lithium-air battery based power supply device

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2024/0204269 A1
- Document ID: US20240204269A1
- Status: Published application
- Kind: A1
- Application number: 18/332,216
- Filed: 2023-06-09
- Published: 2024-06-20
- Earliest claimed: 2022-12-16
- Assignees: Hyundai Motor Company; Kia Corporation; Seoul National University R&Db Foundation
- Applicants: Hyundai Motor Company; Kia Corporation; Seoul National University R&Db Foundation
- Inventors: Hoimin Kwon; Youngsuk Cho; Suhyun Kim; Kyounghan Ryu; Yeongho Lee; Jinyoung Park; Yooil Lee; Minsoo Kim
- Web page: https://patentgrep.com/patent/US20240204269A1

## CPC

- H01M 10/4285: {Testing apparatus}
- G01R 31/3644: {Constructional arrangements}
- G01R 31/367: Software therefor, e.g. for battery testing using modelling or look-up tables
- G01R 31/378: specially adapted for the type of battery or accumulator
- 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)
- H01M 12/08: composed of a half-cell of a fuel-cell type and a half-cell of the secondary-cell type
- Y02E 60/10: Energy storage using batteries
- H01M: PROCESSES OR MEANS, e.g. BATTERIES, FOR THE DIRECT CONVERSION OF CHEMICAL ENERGY INTO ELECTRICAL ENERGY
- G01R: MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES (indicating correct tuning of resonant circuits H03J3/12)
- Y02E: REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION

## Abstract

An apparatus for diagnosing a fault for a lithium-air battery based power supply device includes a data collection part configured to collect data by measuring operational state changes occurring in the power supply device components for each type of fault, a training part configured to train a neural network model using the collected data to classify the type of fault, and a fault diagnosis part configured to diagnose the type of fault by inputting residual changes of each component during operation of the power supply device as input values for the trained neural network model.

## Claims

### Claim 1

An apparatus for diagnosing a fault of a lithium-air battery based power supply device, the apparatus comprising:
a data collection part configured to perform a test operation on a lithium-air battery based power supply device which comprises an air supply part, the air supply part comprising an air pump for suctioning external air, a moisture adsorption part for absorbing moisture in the air supplied from the air pump, an oxygen concentration part for concentrating oxygen in the air supplied through the moisture adsorption part, and a battery part for receiving the concentrated oxygen from the oxygen concentration part at a time of discharge, and in response to a fault occurring in the air supply part, the moisture adsorption part, the oxygen concentration part, and the battery part, measure operational state changes of the air supply part, the moisture adsorption part, the oxygen concentration part, and the battery part according to the types of the fault, to collect data;
a training part configured to train a neural network model by repeating a process of inputting residual changes according to the types of the fault into a neural network model as input values and classifying the types of the fault based on output values of the neural network model which are measurements of the operational state changes corresponding to the input values; and
a fault diagnosis part configured to, during an actual operation of the lithium-air battery based power supply device, input residual changes of the air supply part, the moisture adsorption part, the oxygen concentration part, and the battery part as input values for the trained neural network model, and diagnose a type of a fault based on output values output in response to the input values of the neural network model.

### Claim 2 (depends on 1)

The apparatus of claim 1, wherein the data collection part comprises:
a first sensor part for detecting a fault of at least one solenoid valve provided in: (i) an air route connecting the air pump to the moisture absorption part, (ii) an air route connecting the moisture absorption part to the oxygen concentration part, (iii) an air route connecting the oxygen concentration part to a vacuum pump discharging air of the oxygen concentration part to an outside, or (iv) an air route connecting the oxygen concentration part to the battery part,
wherein the data collection part is configured to collect data obtained by measuring the operational state changes occurring in the air supply part, the moisture absorption part, the oxygen concentration part, and the battery part in response to a fault being detected by the first sensor part.

### Claim 3 (depends on 2)

The apparatus of claim 2, wherein the data collection part further comprises:
a second sensor part for detecting a fault of at least one of: (i) the air pump, (ii) a first adsorption dehumidifier and a second adsorption dehumidifier included in the moisture absorption part, (iii) a heating part for individually heating the first adsorption dehumidifier and the second adsorption dehumidifier, (iv) a first oxygen concentrator and a second oxygen concentrator included in the oxygen concentration part, (v) the vacuum pump, (vi) a lithium-air battery included in the battery part, and (vii) a pressure regulator for regulating an internal pressure of the battery part,
wherein the data collection part is configured to collect data obtained by measuring the operational state changes occurring in the air supply part, the moisture absorption part, the oxygen concentration part, and the battery part in response to a fault being detected by the second sensor.

### Claim 4 (depends on 1)

The apparatus of claim 1, wherein the training part represents the residual changes using normalized data values between 0 and 1.

### Claim 5 (depends on 1)

The apparatus of claim 1, wherein the training part represents the output values of the neural network model as a unique pattern of matrix values according to the type of the fault.

### Claim 6 (depends on 1)

The apparatus of claim 1, wherein the fault diagnosis part adjusts a threshold value used for determining an occurrence of a fault based on the output values of the neural network model such that a minor fault having an insignificant impact on the operation of the lithium-air battery based power supply device is treated as normal.

### Claim 7

A method for diagnosing a fault of a lithium-air battery based power supply device, the method comprising:
performing a test operation on a lithium-air battery based power supply device which comprises an air supply part comprising an air pump for suctioning external air, a moisture adsorption part for absorbing moisture in the air supplied from the air pump, an oxygen concentration part for concentrating oxygen in the air supplied through the moisture adsorption part, and a battery part for receiving the concentrated oxygen from the oxygen concentration part at a time of discharge;
collecting data, in response to a fault occurring in the air supply part, the moisture adsorption part, the oxygen concentration part, and the battery part, by measuring operational state changes of the air supply part, the moisture adsorption part, the oxygen concentration part, and the battery part according to the types of the fault;
training a neural network model by repeating a process of inputting residual changes according to the types of the fault into a neural network model as input values and classifying the types of the fault based on output values of the neural network model which are measurements of the operational state changes corresponding to the input values;
inputting residual changes of the air supply part, the moisture adsorption part, the oxygen concentration part, and the battery part, during an actual operation of the lithium-air battery based power supply device, as input values for the trained neural network model during an actual operation of the lithium-air battery based power supply device; and
diagnosing a type of a fault based on output values output in response to the input values of the neural network model.

### Claim 8 (depends on 7)

The method of claim 7, wherein the collecting data comprises:
detecting, by a first sensor, a fault of at least one of solenoid valves respectively provided in an air route connecting the air pump to the moisture absorption part, an air route connecting the moisture absorption part to the oxygen concentration part, an air route connecting the oxygen concentration part to a vacuum pump discharging air of the oxygen concentration part to an outside, and an air route connecting the oxygen concentration part to the battery part; and
measuring the operational state changes occurring in the air supply part, the moisture absorption part, the oxygen concentration part, and the battery part in response to the fault detected by the first sensor.

### Claim 9 (depends on 8)

The method of claim 8, wherein the collecting data further comprises:
detecting, by a second sensor, a fault of at least one of the air pump, a first adsorption dehumidifier and a second adsorption dehumidifier included in the moisture absorption part, a heating part for individually heating the first adsorption dehumidifier and the second adsorption dehumidifier, a first oxygen concentrator and a second oxygen concentrator included in the oxygen concentration part, the vacuum pump, a lithium-air battery included in the battery part, and a pressure regulator for regulating an internal pressure of the battery part; and
measuring the operational state changes occurring in the air supply part, the moisture absorption part, the oxygen concentration part, and the battery part in response to the fault detected by the second sensor.

### Claim 10 (depends on 7)

The method of claim 7, wherein the training the neural network model comprises:
representing the residual changes using normalized data values between 0 and 1.

### Claim 11 (depends on 7)

The method of claim 7, wherein the training the neural network model comprises:
representing the output values of the neural network model as a unique pattern of matrix values according to the type of the fault.

### Claim 12 (depends on 7)

The method of claim 7, wherein the diagnosing the type of fault comprises:
adjusting a threshold value used for determining an occurrence of a fault based on the output values of the neural network model such that a minor fault having an insignificant impact on the operation of the lithium-air battery based power supply device is treated as normal.
