IP Library Granted Patent US 12,614,270
Granted Patent B2
US 12,614,270 · App. 18/327,153 · Granted Apr 28, 2026

Machine learning system for natural gas leak detection

Inventors: Mehdi Korjani (Pasadena, CA); David A. Conley (Berthoud, CO); Mark Smith (Gilbert, AZ)
Assignee: Clean Connect AI, Inc.
G06T7/0004G06V20/46G06V20/52G06T2207/10016G06T2207/10048
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 12,614,270
App. No.
18/327,153
Granted
Apr 28, 2026
Kind
B2
Abstract

Various embodiments of the present technology relate to solutions for gas leak detection in natural gas extraction and storage environments. In some examples, a machine learning interface generates feature vectors based on video data that depicts a natural gas storage facility and feeds the feature vectors to a machine learning engine. The machine learning engine ingests the feature vectors and generates a machine learning output that indicates the presence of the gas leak in the natural gas storage facility. The machine learning interface receives the machine learning output that indicates the presence of a gas leak in the natural gas storage facility. The machine learning interface generates and transfers an alert based on the machine learning output.

Claims (51)

1 . A method of operating a leak detection system to detect gaseous leaks in a natural gas storage environment, the method comprising:

generating feature vectors based on video data that depicts a natural gas storage facility;

providing the feature vectors as input to a motion detection machine learning model, a background detection machine learning model, an object detection machine learning model, and a gas leak detection machine learning model, wherein:

the motion detection machine learning model indicates a presence or lack thereof of motion in the video data;

the background detection machine learning model indicates a presence or lack thereof of background environment in the video data;

the object detection machine learning model indicates a presence or lack thereof of natural gas extraction, storage, or transfer equipment in the video data;

the gas leak detection machine learning model indicates a presence or lack thereof of a gas leak;

providing a motion detection indication, a background detection indication, an object detection indication, and a gas leak detection indication to a non-linear function machine learning model, wherein the non-linear function machine learning model indicates a likelihood of a presence or lack thereof of the gas leak in the natural gas storage facility;

in response to an indication of the likelihood of the presence or lack thereof of the gas leak in the natural gas storage facility, generating and transferring data for rendering a user interface that comprises the indication.

2 . The method of claim 1 wherein:

the video data comprises infrared video data;

generating the feature vectors comprises generating numerical representations of the infrared video data; and

providing the feature vectors as the input to the motion detection machine learning model, the background detection machine learning model, the object detection machine learning model, and the gas leak detection machine learning model comprises providing the numerical representations of the infrared video data as the input to the motion detection machine learning model, the background detection machine learning model, the object detection machine learning model, and the gas leak detection machine learning model.

3 . The method of claim 1 wherein generating and transferring the data for rendering the user interface that comprises the indication comprises generating and transferring the data for rendering the user interface that comprises the likelihood of the presence or lack thereof of the gas leak, an estimated gas flow rate, and an equipment identification number.

4 . The method of claim 1 wherein generating the feature vectors based on the video data comprises generating the feature vectors based on the video data and sensor telemetry data for the natural gas storage facility.

5 . The method of claim 1 further comprising generating the video data that depicts the natural gas storage facility.

6 . A leak identification system configured to detect gaseous leaks in a natural gas extraction environment, the leak identification system comprising:

a machine learning interface configured to generate feature vectors based on video data that depicts a natural gas storage facility and provide the feature vectors as input to a motion detection machine learning model, a background detection machine learning model, an object detection machine learning model, and a gas leak detection machine learning model;

the motion detection machine learning model configured to ingest the feature vectors and generate a motion detection output that indicates the presence of motion in the video data;

the background detection machine learning model configured to ingest the feature vectors and generate a background detection output that indicates background portions of the video data;

the object detection machine learning model configured to ingest the feature vectors and generate an object detection output that indicates natural gas extraction, storage, or transfer equipment in the video data;

the gas leak detection machine learning model configured to ingest the feature vectors and generate a gas leak detection output that indicates portions of the video data that depict a gas leak;

a non-linear function machine learning model configured to ingest the motion detection output, the background detection output, the object detection output, and the gas leak detection output and utilize a non-linear function machine learning algorithm to generate a non-linear function output that indicates a likelihood of the presence of the gas leak in the natural gas storage facility; and

the machine learning interface configured to receive the non-linear function output that indicates the likelihood of the presence of the gas leak in the natural gas storage facility, generate data for rendering a user interface that comprises the indication, and transfer the data.

7 . The system of claim 6 wherein:

the video data comprises infrared video data; and

the machine learning interface is configured to generate numerical representations of the infrared video data and provide the numerical representations of the infrared video data as the input to the motion detection machine learning model, the background detection machine learning model, the object detection machine learning model, and the gas leak detection machine learning model.

8 . The system of claim 6 wherein the machine learning interface is configured to generate and transfer the data for rendering the user interface that comprises the likelihood of the presence or lack thereof of the gas leak, an estimated gas flow rate, and an equipment identification number.

9 . The system of claim 6 wherein the machine learning interface is configured to generate the feature vectors based on the video data and sensor telemetry data for the natural gas storage facility.

10 . The system of claim 6 further comprising an imaging system configured to generate the video data that depicts the natural gas storage facility.

11 . A non-transitory computer-readable medium stored thereon instructions to detect gaseous leaks in a natural gas extraction environment, that, in response to execution, cause a system comprising a processor to perform operations, the operations comprising:

generating feature vectors based on video data that depicts a natural gas storage facility;

providing the feature vectors as input to a motion detection machine learning model, a background detection machine learning model, an object detection machine learning model, and a gas leak detection machine learning model, wherein:

the motion detection machine learning model indicates a presence or lack thereof of motion in the video data;

the background detection machine learning model indicates a presence or lack thereof of background environment in the video data;

the object detection machine learning model indicates a presence or lack thereof of natural gas extraction, storage, or transfer equipment in the video data;

the gas leak detection machine learning model indicates a presence or lack thereof of a gas leak;

providing a motion detection indication, a background detection indication, an object detection indication, and a gas leak detection indication to a non-linear function machine learning model, wherein the non-linear function machine learning model indicates a likelihood of the presence or lack thereof of the gas leak in the natural gas storage facility; and

in response to an indication of the likelihood of the presence or lack thereof of the gas leak in the natural gas storage facility, generating and transferring data for rendering a user interface that comprises the indication.

12 . The non-transitory computer readable medium claim 11 wherein:

the video data comprises infrared video data;

generating the feature vectors comprises generating numerical representations of the infrared video data; and

providing the feature vectors as the input to the motion detection machine learning model, the background detection machine learning model, the object detection machine learning model, and the gas leak detection machine learning model comprises providing the numerical representations of the infrared video data as the input to the motion detection machine learning model, the background detection machine learning model, the object detection machine learning model, and the gas leak detection machine learning model.

13 . The non-transitory computer readable medium claim 11 wherein generating and transferring the data for rendering the user interface comprises generating and transferring the data for rendering the user interface that comprises the likelihood of the presence or lack thereof of the gas leak, an estimated gas flow rate, and an equipment identification number.

14 . The non-transitory computer readable medium claim 11 wherein generating the feature vectors based on the video data comprises generating the feature vectors based on the video data and sensor telemetry data for the natural gas storage facility.

15 . The method of claim 1 further comprising estimating a flow rate of the gas leak.

16 . The method of claim 1 further comprising rendering the user interface to display the indication based on the data.

17 . The system of claim 6 wherein the gas leak detection machine learning model is further configured to estimate a flow rate of the gas leak.

18 . The system of claim 6 further comprising a user computer configured to render the user interface to display the indication based on the data.

19 . The non-transitory computer readable medium claim 11 wherein the operations further comprise estimating a flow rate of the gas leak.

20 . The non-transitory computer readable medium claim 11 wherein the operations further comprise rendering the user interface to display the indication based on the data.

Assignments (4)
SECURITY INTEREST Recorded May 8, 2026
From: CLEAN CONNECT AI, INC.
To: LAGO EVERGREEN CREDIT
Reel/Frame 074605/0753 →
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Apr 16, 2024
From: CC HOLDCO, INC.
To: CLEAN CONNECT AI, INC.
Reel/Frame 067116/0457 →
CHANGE OF NAME Recorded Oct 31, 2023
From: CLEAN CONNECT, INC.
To: CC HOLDCO, INC.
Reel/Frame 065413/0269 →
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Jun 1, 2023
From: KORJANI, MEHDI; CONLEY, DAVID A.; SMITH, MARK
To: CLEAN CONNECT, INC.
Reel/Frame 063819/0798 →
Continuity (2)
Provisional Application 63391908 · Jul 25, 2022
Related Publication 20240029232A1 · Jan 25, 2024
References Cited (14)
US 10657443B2 · Araujo et al. · 2020 [cited by applicant]
US 10726558B2 · Ruda et al. · 2020 [cited by applicant]
US 11609151B1 · Collins et al. · 2023 [cited by applicant]
US 20120206595A1 · Alphenaar · 2012 [cited by examiner]
US 20160284075A1 · Phan et al. · 2016 [cited by applicant]
US 20190180464A1 · Kraft · 2019 [cited by examiner]
US 20190354772A1 · Tasli et al. · 2019 [cited by applicant]
US 20200169734A1 · Fainstain · 2020 [cited by examiner]
US 20210216852A1 · Reece et al. · 2021 [cited by applicant]
US 20210232741A1 · Ogiso et al. · 2021 [cited by applicant]
US 20230063604A1 · Al Daif · 2023 [cited by examiner]
US 20230105173A1 · Spinelli · 2023 [cited by examiner]
US 20230177726A1 · Spears · 2023 [cited by examiner]
US 20230186445A1 · DeLuca · 2023 [cited by examiner]