Method for constructing infrared imaging dataset of gas leakage based on computational fluid dynamics

Patent metadata

Field	Value
Publication number	US-12467859-B2
Application number	US-202519215245-A
Country	US
Kind code	B2
Filing date	May 21, 2025
Priority date	May 22, 2024
Publication date	Nov 11, 2025
Grant date	Nov 11, 2025

Abstract

A method for constructing an infrared imaging dataset of gas leakage based on computational fluid dynamics includes steps of: establishing a three-dimensional physical model of a gas leakage field scene; meshing and simulating based on the computational fluid dynamics to obtain leaking gas mole fractions of each mesh under time steps, and constituting three-dimensional gas concentration data corresponding to each frame; using optical gas imaging based on a pinhole camera model, imaging the three-dimensional gas concentration data to obtain initial images, and calculating gas concentration path-lengths corresponding to pixel points in each frame of the initial images; and performing maximum-minimum value normalization and generating grayscale images, thereby constructing the infrared imaging dataset of the gas leakage. The method can not only be used in leakage classification tasks and localization tasks, but can also be used in training tasks related to the leakage concentration.

First claim

What is claimed is:

1 . A method for constructing an infrared imaging dataset of gas leakage based on computational fluid dynamics, comprising steps of:
S1: in a gas leakage field scene, collecting a geometric structure of a pipeline and a leakage aperture, so as to establish a three-dimensional physical model of the gas leakage field scene; S2: meshing the three-dimensional physical model, and determining an inlet surface, an outlet surface and a wall surface of the three-dimensional physical model; S3: based on the computational fluid dynamics, simulating with the meshed three-dimensional physical model; setting at least one inlet velocity of the inlet surface to obtain leaking gas mole fractions of each mesh under time steps, and constituting three-dimensional gas concentration data corresponding to each frame; wherein during simulating, constructing a component transportation equation based on components of a leaking gas and an ambient gas determined from the gas leakage field scene, and selecting a turbulence model based on the geometric structure of the pipeline and a flow rate of the leaking gas; S4: using optical gas imaging based on a pinhole camera model, imaging the three-dimensional gas concentration data to obtain initial images, and calculating gas concentration path-lengths corresponding to pixel points in each frame of the initial images; and S5: performing maximum-minimum value normalization on the gas concentration path-lengths corresponding to all the pixel points in the initial images, and generating grayscale images according to normalization results; and constructing the infrared imaging dataset of the gas leakage based on the grayscale images and the gas concentration path-lengths corresponding to the pixel points in the grayscale images.

2 . The method, as recited in claim 1 , wherein in the step S2, a size of the mesh is 1/100 to 1/1000 of a size of the geometrical structure of the pipeline in the gas leakage field scene.

3 . The method, as recited in claim 1 , wherein in the step S3, a coulomb number during simulating is kept no more than 2.

4 . The method, as recited in claim 1 , wherein in the step S3, the turbulence model is a standard k-ε model, a RNG (Re-normalization group) k-ε model, a realizable k-ε model, a standard k-ε model, or an SST (Shear Stress Transport) k-ε model.

5 . The method, as recited in claim 1 , wherein in the step S4, during imaging, the gas concentration path-lengths corresponding to the pixel points are obtained by processing the leaking gas mole fractions of all meshes contained in each pixel point with path integrating.

6 . The method, as recited in claim 1 , wherein in the step S4, before imaging, the three-dimensional gas concentration data are processed with nearest-neighbor interpolation.

Assignees

• Univ Electronic Sci & Tech China

Inventors

• Tai Huiling
• Zhang Yu
• Jiang Yadong
• Lai Wenjie
• Wu Yuanming

Classifications

• G06F30/28 Primary

  using fluid dynamics, e.g. using Navier-Stokes equations or computational fluid dynamics [CFD] · CPC title

• G06F2113/08

  Fluids · CPC title

• G01N21/3504 Primary

  for analysing gases, e.g. multi-gas analysis · CPC title

• G06F2111/10

  Numerical modelling · CPC title

• G06F2119/14

  Force analysis or force optimisation, e.g. static or dynamic forces · CPC title