Granted Patent B2

US 12,561,972 · App. 18/311,061 · Granted Feb 24, 2026

Inventors: Su Liu (Austin, TX); Martin G. Keen (Cary, NC); Jeremy R. Fox (Georgetown, TX); Hamid Majdabadi (Ottawa, CA)

Assignee: International Business Machines Corporation

G06V20/13G16Y20/10G16Y40/10G16Y40/20

Abstract

Claims (58)

1 . A computer implemented method for monitoring a selected type of emission, the computer implemented method comprising: training, by a number of processor units, a machine learning model using historical satellite spectral data correlated with known emission levels for the selected type of emission, wherein the machine learning model is configured to identify spectral signatures indicative of potentially undesired emission levels;

analyzing, by the number of processor units, current satellite data from a planetary region using the trained machine learning model to identify a set of areas on a planet with spectral signatures matching the potentially undesired emission level for the selected type of emission;

identifying, by the number of processor units, infrastructure in the set of areas from spectral signatures in the satellite data;

selecting, by the number of processor units, a number of existing Internet of Things sensors associated with the infrastructure to monitor for the selected type of emission in the set of areas in real time, wherein the selecting comprises choosing from sensors already deployed in the infrastructure;

analyzing, by the number of processor units, emissions data generated by the number of Internet of Things sensors from monitoring for the selected type of emission to form an analysis; and

performing, by the number of processor units, a set of actions using the analysis.

2 . The computer implemented method of claim 1 further comprising:

generating, by the number of processor units, a visual display for at least one of the set of actions or the analysis in a graphical user interface on a display system and wherein the analysis comprises at least one of: a level of emissions for the selected type of emission in a number of locations in the set of areas, an identification of a set of sources for the selected type of emission in the set of areas, a spread of the selected type of emission, or a trend for the selected type of emission over time.

3 . The computer implemented method of claim 1 further comprising:

determining, by the number of processor units, a spread of emissions for the selected type of emission using the emissions data generated by the Internet of Things sensors; and

selecting, by the number of processor units, new Internet of Things sensors in new locations to monitor for the selected type of emission based on the spread of the selected type of emission, wherein the new Internet of Things sensors generate new emissions data used in the analysis.

4 . The computer implemented method of claim 1 , wherein identifying, by the number of processor units, the set of areas further comprises:

identifying, by a number of processor units, the set of areas on the planet with the potentially undesired emission level for the selected type of emission using the satellite data, the machine learning model, and at least one of weather patterns, land-use patterns, or emissions levels of the selected type of emission.

5 . The computer implemented method of claim 4 , wherein identifying, by the number of processor units, the set of areas further comprises:

identifying, by the number of processor units, a set of locations in infrastructure in the set of areas for the potentially undesired emission level for the selected type of emission using satellite data and a machine learning model trained to identify the potentially undesired emission level for the selected type of emission.

6 . The computer implemented method of claim 5 , wherein selecting, by the number of processor units, the number of Internet of Things sensors further comprises:

selecting, by the number of processor units, the number of Internet of Things sensors in the set of locations in the infrastructure to monitor for the selected type of emission.

7 . The computer implemented method of claim 1 , wherein the set of actions is selected from at least one of generating an alert, recommending an action to reduce the selected type of emission, collecting additional emissions data in the set of areas, collecting the additional emissions data in infrastructure in the set of areas, indicating a presence of a leak of the selected type of emission, or adjusting operation of equipment identified as generating the potentially undesired emission level for the selected type of emission.

8 . The computer implemented method of claim 1 , wherein the satellite data is selected from at least one of a satellite image, spectral data, or concentration data.

9 . The computer implemented method of claim 1 , wherein the selected type of emission is selected from a group comprising a greenhouse gas, methane, carbon dioxide, nitrous oxide, a fluorinated gas, a particulate matter, a volatile organic compound, nitrous oxide, sulfur dioxide, and ozone.

10 . The computer implemented method of claim 1 wherein the selected type of emission comprises methane and the method further comprises:

identifying the set of areas on the planet comprises analyzing historical satellite data of the set of areas over a predetermined time period to identify temporal emission patterns;

identifying the infrastructure comprises identifying at least one of oil and gas pipelines, natural gas storage facilities, or landfill sites in the set of areas; and

selecting the number of Internet of Things sensors comprises selecting sensors having methane detection sensitivity of at least 1 part per million and deploying the sensors at intervals determined based on detected wind patterns in the set of areas.

11 . The computer implemented method of claim 1 , wherein training the machine learning model further comprises:

creating a training dataset comprising historical satellite spectral data and historical emission data with correlated timestamps; and

training the machine learning model to identify patterns correlating spectral signatures in the historical satellite spectral data with known emission levels in the historical emission data, wherein the trained machine learning model predicts potentially undesired emission levels in areas where spectral signatures match the identified patterns but emission levels are not directly observable in the current satellite data.

12 . A computer system comprising:

a number of processor units, wherein the number of processor units executes program instructions to:

train a machine learning model using historical satellite spectral data correlated with known emission levels for the selected type of emission, wherein the machine learning model is configured to identify spectral signatures indicative of potentially undesired emission levels;

analyze current satellite data from a planetary region using the trained machine learning model to identify a set of areas on a planet with spectral signatures matching the potentially undesired emission level for a selected type of emission;

identify infrastructure in the set of areas from spectral signatures in the satellite data;

select a number of existing Internet of Things sensors associated with the infrastructure to monitor for the selected type of emission in the set of areas in real time, wherein the selecting comprises choosing from sensors already deployed in the infrastructure;

analyze emissions data generated by the number of Internet of Things sensors from monitoring for the selected type of emission to form an analysis; and

perform a set of actions using the analysis.

13 . The computer system of claim 12 , wherein the number of processor units further executes program instructions to:

generate a visual display for at least one of the set of actions or the analysis in a graphical user interface on a display system and wherein the analysis comprises at least one of: a level of emissions for the selected type of emission in a number of locations in the set of areas, an identification of a set of sources for the selected type of emission in the set of areas, a spread of the selected type of emission, or a trend for the selected type of emission over time; and performing, by the number of processor units, a set of actions using the analysis.

14 . The computer system of claim 12 , wherein the number of processor units further executes program instructions to:

determine a spread of emissions for the selected type of emission using the emissions data generated by the Internet of Things sensors; and

select new Internet of Things sensors in new locations to monitor for the selected type of emission based on the spread of the selected type of emission, wherein the new Internet of Things sensors generate new emissions data used in the analysis.

15 . The computer system of claim 14 , wherein in identifying the set of areas, the number of processor units further executes program instructions to:

identify the set of areas on the planet with the potentially undesired emission level for the selected type of emission using the satellite data, the machine learning model, and at least one of weather patterns, land-use patterns, or emissions levels of the selected type of emission.

16 . The computer system of claim 12 , wherein in identifying, by the number of processor units, the set of areas, the number of processor units further executes program instructions to:

identify a set of locations in infrastructure in the set of areas for the potentially undesired emission level for the selected type of emission using satellite data and a machine learning model trained to identify the potentially undesired emission level for the selected type of emission.

17 . The computer system of claim 16 , in selecting the number of Internet of Things sensors, the number of processor units further executes program instructions to:

select the number of Internet of Things sensors in the set of locations in the infrastructure to monitor for the selected type of emission.

18 . The computer system of claim 12 , wherein training the machine learning model further comprises:

creating a training dataset comprising historical satellite spectral data and historical emission data with correlated timestamps; and

training the machine learning model to identify patterns correlating spectral signatures in the historical satellite spectral data with known emission levels in the historical emission data, wherein the trained machine learning model predicts potentially undesired emission levels in areas where spectral signatures match the identified patterns but emission levels are not directly observable in the current satellite data.

19 . A computer program product for monitoring a selected type of emission, the computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by a computer system to cause the computer system to:

train a machine learning model using historical satellite spectral data correlated with known emission levels for the selected type of emission, wherein the machine learning model is configured to identify spectral signatures indicative of potentially undesired emission levels;

analyze current satellite data from a planetary region using the trained machine learning model to identify a set of areas on a planet with spectral signatures matching the potentially undesired emission level for the selected type of emission;

identify infrastructure in the set of areas from spectral signatures in the satellite data;

select a number of existing Internet of Things sensors associated with the infrastructure to monitor for the selected type of emission in the set of areas in real time, wherein the selecting comprises choosing from sensors already deployed in the infrastructure;

analyze emissions data generated by the number of Internet of Things sensors from monitoring for the selected type of emission to form an analysis, wherein the analysis comprises at least one of: a level of emissions for the selected type of emission in a number of locations in the set of areas, an identification of a set of sources for the selected type of emission in the set of areas, a spread of the selected type of emission, or a trend for the selected type of emission over time; and performing, by the number of processor units, a set of actions using the analysis; and

perform a set of actions using the analysis.

20 . The computer programming product of claim 19 , further comprising:

generating, by the number of processor units, a visual display for at least one of the set of actions or the analysis in a graphical user interface on a display system and wherein the analysis comprises at least one of: a level of emissions for the selected type of emission in a number of locations in the set of areas, an identification of a set of sources for the selected type of emission in the set of areas, a spread of the selected type of emission, or a trend for the selected type of emission over time.

Assignments (1)

Continuity (1)

References Cited (23)