Methods and systems for characterizing methane emission employing mobile methane emission detection
Approaches to characterizing methane emission at an industrial facility using an unmanned air vehicle (UAV) equipped with navigational instruments and methane gas sensors. A scan area associated with the industrial facility is defined. The scan area can be logically partitioned into a plurality of pixels based on information provided by an intelligent agent. A flight plan for a flight of the UAV can be created to cover the pixels of the scan area. The UAV can execute the flight plan and collect time-series navigational data and time-series sensor data as the UAV executes the flight plan. The flight path of the UAV can be dynamically adjusted during the flight by an agent to ensure that the flight path of the UAV covers the scan area. A two-dimensional (2D) image of methane concentration over the scan area can be constructed from the time-series navigational data and the time-series sensor data.
1 . A method for characterizing methane emission at an industrial facility using an unmanned air vehicle (UAV) equipped with at least one navigational instrument and at least one methane gas sensor, the method comprising:
defining a scan area associated with the industrial facility, wherein the scan area is a two-dimensional (2D) plane;
logically partitioning the scan area into a plurality of pixels based on information provided by an intelligent agent, wherein the plurality of pixels include a plurality of subsquares spatially distributed over the scan area;
defining a flight plan for a flight of the UAV, wherein the flight plan covers the plurality of pixels of the scan area;
operating the UAV to execute the flight plan and collect time-series navigational data from the at least one navigational instrument and time-series sensor data from the at least one methane gas sensor as the UAV executes the flight plan;
generating a predicted success of the flight by applying a mind map to the time-series navigational data and the time-series sensor data;
responsive to the predicted success being less than a threshold, dynamically adjusting the flight plan of the UAV during the flight by operation of the intelligent agent to ensure that the flight plan of the UAV covers the plurality of pixels of the scan area; and
constructing at least one 2D image of methane concentration over the plurality of pixels of the scan area from the time-series navigational data and the time-series sensor data.
2 . A method according to claim 1 , wherein the intelligent agent is configured to use at least one model to dynamically adjust speed or stabilization of the UAV during the flight.
3 . A method according to claim 2 , wherein the at least one model comprises at least one of a physics-based model and an artificial intelligence model.
4 . A method according to claim 1 , further comprising processing the at least one 2D image using artificial intelligence and/or computer vision algorithms to identify anomalies or patterns related to the methane emission at the industrial facility.
5 . A method according to claim 4 , wherein processing of the at least one 2D image employs an autoencoder configured to identify the anomalies or the patterns related to the methane emission at the industrial facility.
6 . A method according to claim 4 , wherein processing of the at least one 2D image is further configured to estimate a rate of the methane emission and a source location of the methane emission at the industrial facility.
7 . A method according to claim 6 , wherein processing of the at least one 2D image employs an artificial intelligence model configured to quantify the rate of the methane emission and estimate an approximate location of the source location of the methane emission at the industrial facility.
8 . A method according to claim 4 , wherein processing of the at least one 2D image is further configured to visualize a rate of the methane emission and a source location of the methane emission at the industrial facility.
9 . A method according to claim 4 , wherein the intelligent agent is configured to provide flight plan information used to define the flight plan such that a spatial distribution of pixels in the scan area and the time-series sensor data collected during the flight and the at least one 2D image derived therefrom comply with an input of the artificial intelligence and/or computer vision algorithms.
10 . A method according to claim 1 , wherein: the pixels comprise subsquares that cover the scan area; and/or the scan area is defined in at least one the 2D plane spaced from the industrial facility is a horizontal 2D plan spaced above the industrial facility.
11 . A method according to claim 1 , wherein a wind speed and a direction at or near the industrial facility at time of the flight are determined from data provided by a weather station or a weather monitoring service, and the wind speed and the direction are used to define the scan area.
12 . A method according to claim 1 , wherein the at least one 2D image is constructed by scaling raw methane concentration readings to a predefined range of pixel values over the plurality of pixels of the scan area.
13 . A method according to claim 1 , wherein the at least one 2D image is constructed using an image smoothing algorithm.
14 . A method according to claim 13 , wherein the image smoothing algorithm comprises a spline smoothing algorithm or a sinc smoothing algorithm.
15 . A method according to claim 1 , wherein the flight of the UAV operates under remote control by a human operator.
16 . A method according to claim 1 , wherein the flight of the UAV operates under remote control with a degree of autonomy.
17 . A method according to claim 16 , wherein the remote control includes autopilot assistance, or the remote control provides for fully autonomous control of the UAV with no provision for human intervention.
18 . The method of claim 1 , wherein the mind map includes a plurality of binary decision trees for evaluating a quality of at least one of the time-series navigational data or the time-series sensor data, wherein the predicted success is based at least in part on the quality.
19 . The method of claim 18 , further comprising generating, for each of the plurality of binary decision trees, a plurality of binary classification labels associated with the time-series navigational data and the time-series sensor data, and wherein the predicted success is based on the plurality of binary classification labels.
20 . A system, comprising:
a processor and memory, the memory including instructions that cause the processor to:
defining a scan area associated with an industrial facility, wherein the scan area is a two-dimensional (2D) plane;
logically partitioning the scan area into a plurality of pixels based on information provided by an intelligent agent, wherein the plurality of pixels include a plurality of subsquares spatially distributed over the scan area;
defining a flight plan for a flight of an unmanned aerial vehicle (UAV), wherein the flight plan covers the plurality of pixels of the scan area;
operating the UAV to execute the flight plan and collect time-series navigational data from at least one navigational instrument and time-series sensor data from at least one methane gas sensor as the UAV executes the flight plan;
generating a predicted success of the flight by applying a mind map to the time-series navigational data and the time-series sensor data;
responsive to the predicted success being less than a threshold, dynamically adjusting the flight plan of the UAV during the flight by operation of the intelligent agent to ensure that the flight plan of the UAV covers the plurality of pixels of the scan area; and
constructing at least one 2D image of methane concentration over the plurality of pixels of the scan area from the time-series navigational data and the time-series sensor data.