# US 10,769,770 B2 · Quality monitoring system and quality monitoring method for fuel cell manufacturing line and quality monitoring system for manufacturing line

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

- Number: US 10,769,770 B2
- Document ID: US10769770B2
- Status: Granted
- Kind: B2
- Application number: 15/972,853
- Filed: 2018-05-07
- Published: 2019-11-07
- Granted: 2020-09-08
- Earliest claimed: 2018-05-07
- Assignees: Cummins Enterprise LLC
- Applicants: Cummins Enterprise LLC
- Inventors: Honggang Wang; Todd Striker; Matthew J. Alinger
- Web page: https://patentgrep.com/patent/US10769770B2

## CPC

- G06T 7/0004: {Industrial image inspection}
- G06F 16/2379: {Updates performed during online database operations; commit processing}
- G06F 16/285: {Clustering or classification}
- H01M 8/10: Fuel cells with solid electrolytes
- H01M 8/12: operating at high temperature, e.g. with stabilised ZrO2 electrolyte
- H01M 8/2404: Processes or apparatus for grouping fuel cells
- H01M 8/2432: Grouping of unit cells of planar configuration
- H04N 7/18: Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast
- G06T 2207/10004: Still image; Photographic image
- G06T 2207/20081: Training; Learning
- G06T 2207/30164: Workpiece; Machine component
- H01M 8/0202: Collectors; Separators, e.g. bipolar separators; Interconnectors
- H01M 8/124: characterised by the process of manufacturing or by the material of the electrolyte
- Y02E 60/50: Fuel cells
- Y02P 70/50: Manufacturing or production processes characterised by the final manufactured product
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- H01M: PROCESSES OR MEANS, e.g. BATTERIES, FOR THE DIRECT CONVERSION OF CHEMICAL ENERGY INTO ELECTRICAL ENERGY
- H04N: PICTORIAL COMMUNICATION, e.g. TELEVISION
- Y02E: REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
- Y02P: CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS

## Abstract

Quality monitoring system and method for a fuel cell manufacturing line are disclosed. The system includes an image collection unit and a real-time quality control computer. The image collection unit is configured for generating a captured image of a surface of one fuel cell in the fuel cell manufacturing line. The computer is configured to receive the captured image and generate a set of feature vectors based on the captured image. The computer comprises a defect model repository comprising a defect detection model repository and a defect classification model repository, a defect detection module and a defect classification module. The defect detection module is configured to access the defect detection model repository and determine whether the fuel cell is defective based on the set of feature vectors and the defect detection model repository. The defect classification module is configured to access the defect classification model repository when the defect detection module determines the fuel cell is defective and determine a defect type of the defective fuel cell based on the set of feature vectors and the defect classification model repository.

## Claims

### Claim 1

A quality monitoring system for a fuel cell manufacturing line, comprising:
an image collection unit configured to generate a captured image of a surface of one fuel cell in the fuel cell manufacturing line;
a real-time quality control computer coupled to the image collection unit and configured to receive the captured image and generate a set of feature vectors based on the captured image, wherein the real-time quality control computer comprises:
a defect model repository comprising a defect detection model repository and a defect classification model repository;
a defect detection module configured to access the defect detection model repository and determine whether the fuel cell is defective based on the set of feature vectors and the defect detection model repository; and
a defect classification module configured to access the defect classification model repository when the defect detection module determines the fuel cell is defective and determine a defect type of the defective fuel cell based on the set of feature vectors of the defective fuel cell and the defect classification model repository.

### Claim 2 (depends on 1)

The quality monitoring system of claim 1, wherein the defect detection model repository has one or more models to determine whether the fuel cell is defective; and the defect classification model repository has one or more models to determine one or more defect types presented in the fuel cell manufacturing line.

### Claim 3 (depends on 2)

The quality monitoring system of claim 2, wherein the fuel cell manufacturing line comprises a substrate assembly phase, an anode deposition phase, an electrolyte deposition phase, a cathode deposition phase, a glass deposition phase and a fuel cell stacking phase, the image collection unit being located before at least one phase of cathode deposition or fuel cell stacking.

### Claim 4 (depends on 3)

The quality monitoring system of claim 3, wherein the anode deposition phase and the electrolyte deposition phase use thermal spray technique.

### Claim 5 (depends on 3)

The quality monitoring system of claim 3, wherein the substrate assembly phase, the anode deposition phase, the electrolyte deposition phase, the cathode deposition phase, and the glass deposition phase occur in series to result in a finished fuel cell which constructs only one component for adding operating voltage of a fuel cell stack.

### Claim 6 (depends on 5)

The quality monitoring system of claim 5, wherein the image collection unit is located after the glass deposition phase and before the fuel cell stacking phase.

### Claim 7 (depends on 3)

The quality monitoring system of claim 3, wherein the fuel cell manufacturing line further comprises a cathode interconnect coating phase, and the defect classification model repository has at least one of a first defect model repository or a second defect model repository, the first defect model repository having one or more models to determine one or more defect types presented on a coating deposited at the phases of substrate assembly, electrolyte deposition, or combinations thereof, and the second defect model repository having one or more models to determine one or more defect types presented at the phases of cathode interconnect coating, cathode deposition, glass deposition or combinations thereof.

### Claim 8 (depends on 1)

The quality monitoring system of claim 1, wherein the set of feature vectors generated comprises at least one of the following sets:
a first set of feature vectors associated with one or more of part size, datum positions, flow field alignment, anode coating defects, electrolyte coating defects, color anomalies indicating contamination and uniformity of cathode interconnect coating; and
a second set of feature vectors associated with one or more of alignment of cathode, alignment of glass seal in reference to part perimeter and manifolds, uniformity of glass seal coverage and uniformity of cathode.

### Claim 9 (depends on 1)

The quality monitoring system of claim 1, further comprising:
a removing unit configured to remove a defective fuel cell from the fuel cell manufacturing line.

### Claim 10 (depends on 9)

The quality monitoring system of claim 9, wherein the real-time quality control computer further comprises:
a confidence determination module configured to determine a confidence value of the defect model repository which are indicative of reliability of the defect model repository, wherein when the confidence value of the defect model repository is above a threshold, the removing unit removes the defective fuel cell from the fuel cell manufacturing line.

### Claim 11 (depends on 10)

The quality monitoring system of claim 10, wherein when the confidence value of the defect model repository drops below the threshold, the defect model repository is updated with the set of feature vectors.

### Claim 12 (depends on 1)

The quality monitoring system of claim 1, wherein the real-time quality control computer is further configured to learn interactively with a field operator via an interactive interface.

### Claim 13 (depends on 12)

The quality monitoring system of claim 12, further comprising a fuel cell database, wherein when the number of defective fuel cells accumulated in the fuel cell database and confirmed by the field operator reaches a limit, the defect model repository is updated with sets of feature vectors generated from images of the defective fuel cells.

### Claim 14 (depends on 1)

The quality monitoring system of claim 1, further comprising:
a labeling unit configured to label the fuel cell with a quality detection result of the fuel cell.

### Claim 15

A quality monitoring method for a fuel cell manufacturing line, comprising:
generating, by a first image collection unit, a captured first image of a surface of one fuel cell in one phase of the fuel cell manufacturing line;
generating a set of feature vectors based on the captured first image; and
accessing a defect model repository and generating a quality detection result of the fuel cell based on the set of feature vectors and the defect model repository, comprising:
accessing a defect detection model repository of the defect model repository;
determining whether the fuel cell is defective based on the set of feature vectors and the defect detection model repository;
when the fuel cell is determined to be defective, accessing a defect classification model repository of the defect model repository; and
determining a defect type of the defective fuel cell based on the set of feature vectors of the defective fuel cell and the defect classification model repository.

### Claim 16 (depends on 15)

The quality monitoring method of claim 15, wherein the quality monitoring method is used for the fuel cell manufacturing line comprising a substrate assembly phase, an anode deposition phase, an electrolyte deposition phase, a cathode deposition phase, a glass deposition phase and a fuel cell stacking phase.

### Claim 17 (depends on 16)

The quality monitoring method of claim 16, wherein the first image collection unit is positioned before at least one phase of cathode deposition or fuel cell stacking.

### Claim 18 (depends on 16)

The quality monitoring method of claim 16, wherein a finished fuel cell which constructs only one component for adding operating voltage of a fuel cell stack is sequentially manufactured through the substrate assembly phase, the anode deposition phase, the electrolyte deposition phase, the cathode deposition phase, and the glass deposition phase occurring in series.

### Claim 19 (depends on 18)

The quality monitoring method of claim 18, wherein the first image collection unit is located after the glass deposition phase and before the fuel cell stacking phase and the quality monitoring method is configured for monitoring quality of the finished fuel cell.

### Claim 20 (depends on 15)

The quality monitoring method of claim 15, further comprising:
when the fuel cell is determined from the captured first image to be good, returning the fuel cell to the fuel cell manufacturing line;
generating, by a second image collection unit, a captured second image of a surface of the fuel cell in another phase of the fuel cell manufacturing line;
generating a second set of feature vectors based on the captured second image; and
accessing the defect model repository and generating a second quality detection result of the fuel cell based on the second set of feature vectors and the defect model repository.

### Claim 21 (depends on 20)

The quality monitoring method of claim 20, wherein the fuel cell manufacturing line comprises a substrate assembly phase, an anode deposition phase, an electrolyte deposition phase, a cathode deposition phase, a glass deposition phase and a fuel cell stacking phase occurring in series, and wherein the first image collection unit is located before cathode deposition phase, and the second image collection unit is located after the glass deposition phase and before the fuel cell stacking phase.

### Claim 22 (depends on 15)

The quality monitoring method of claim 15, further comprising:
when the quality detection result of the fuel cell indicates that the fuel cell is defective, removing the defective fuel cell from the fuel cell manufacturing line.

### Claim 23 (depends on 22)

The quality monitoring method of claim 22, further comprising:
determining whether a confidence value of the defect model repository is above a threshold, wherein when the confidence value of the defect model repository is above the threshold, the defective fuel cell is removed from the fuel cell manufacturing line.

### Claim 24 (depends on 23)

The quality monitoring method of claim 23, further comprising:
updating the defect model repository with the set of feature vectors when the confidence value of the defect model repository drops below the threshold.

### Claim 25 (depends on 23)

The quality monitoring method of claim 23, further comprising:
when the confidence value of the defect model repository drops below the threshold, notifying a field operator and confirming quality of the fuel cell by the field operator.

### Claim 26 (depends on 25)

The quality monitoring method of claim 25, further comprising:
updating the defect model repository with sets of feature vectors generated from images of the defective fuel cells when the number of accumulated defective fuel cells confirmed by the field operator reaches a limit.

### Claim 27

A quality monitoring system for a manufacturing line, comprising:
an image collection unit configured to generate a captured image of a surface of a manufactured item in the manufacturing line;
a real-time quality control computer coupled to the image collection unit and configured to receive the captured image and generate a set of feature vectors based on the captured image, wherein the real-time quality control computer comprises:
a defect model repository comprising a defect detection model repository and a defect classification model repository;
a defect detection module configured to access the defect detection model repository and determine whether the manufactured item is defective based on the set of feature vectors and the defect detection model repository; and
a defect classification module configured to access the defect classification model repository when the defect detection module determines the manufactured item is defective and determine a defect type of the manufactured item based on the set of feature vectors of the manufactured item and the defect classification model repository.