CN-121982106-A - Pipeline abnormal state identification and positioning method and inspection robot thereof

CN121982106ACN 121982106 ACN121982106 ACN 121982106ACN-121982106-A

Abstract

The application discloses a method for identifying and positioning abnormal states of a pipeline and a routing inspection robot thereof, relates to the technical field of machine vision identification, and can solve the problems of low accuracy and high false detection omission rate of abnormal structural postures such as falling, tilting, hanging sagging and the like of underground pipelines of a coal mine in the prior art. The method for identifying and positioning the abnormal state of the pipeline comprises the steps of S1, constructing a global map of a coal mine tunnel, establishing a coordinate corresponding relation between an image acquisition position and the global map, S2, obtaining a pipeline image in the tunnel and position information corresponding to the pipeline image, carrying out noise reduction treatment on the pipeline image to obtain an image to be identified, S3, judging whether the pipeline is in an abnormal state of the gesture, outputting an abnormal pipeline identification result, and S4, completing the positioning of the abnormal pipeline according to the position information corresponding to the abnormal pipeline identification result.

Inventors

MAO QINGHUA
WANG YU
ZHOU TONG
SU HAO
QIN SONG
XUE XUSHENG
WANG CHUANWEI
YANG WENJUAN
SU YINAN

Assignees

西安科技大学

Dates

Publication Date: 20260505
Application Date: 20260409

Claims (10)

1. The method for identifying and positioning the abnormal state of the pipeline is characterized by comprising the following steps of: S1, constructing a global map of a coal mine tunnel, and establishing a coordinate corresponding relation between an image acquisition position and the global map; S2, acquiring a pipeline image in a roadway and position information corresponding to the pipeline image, and performing noise reduction treatment on the pipeline image to obtain an image to be identified; S3, inputting the image to be identified into a neural network model for identification, obtaining a pipeline target area, judging whether the pipeline is in an abnormal state or not according to the geometric characteristics of the pipeline target area and the spatial relationship between the pipeline target area and the ground area, and outputting a pipeline abnormal identification result; And S4, mapping the position of the abnormal pipeline to a global map according to the position information corresponding to the pipeline abnormal identification result so as to finish positioning of the abnormal pipeline.
2. The method for identifying and locating abnormal conditions of a pipeline according to claim 1, wherein in S3, the geometric features of the target area of the pipeline include at least an inclination angle and an aspect ratio, the inclination angle is used for representing a deflection degree of the target area of the pipeline relative to a main direction of the target area of the pipeline, the aspect ratio is used for representing a projection shape feature of the target area of the pipeline, and the spatial relationship feature between the target area of the pipeline and the ground area includes at least one of an overlapping degree, a horizontal distance and a vertical distance.
3. The method for recognizing and locating abnormal states of pipelines according to claim 2, wherein in S3, geometric features are extracted based on the outline or circumscribed rectangle of the target area of the pipeline, spatial relationship features are calculated based on the positional relationship between the target area of the pipeline and the ground area, and the geometric features and the spatial relationship features are compared with a preset threshold to judge whether the pipeline is in abnormal states of posture.
4. The method for recognizing and locating abnormal states of pipeline according to claim 3, wherein in S3, it is judged whether the pipeline is in a falling state according to the overlapping degree between the target area of the pipeline and the ground area, and when it is judged that the pipeline is not in the falling state, it is judged whether the pipeline is in a hanging down or tilting state according to the inclination angle of the target area of the pipeline.
5. The method for identifying and locating abnormal states of a pipeline according to claim 1, wherein the neural network model is YOLOv model, and the YOLOv model includes: A backbone network, wherein MobileNetV4 is adopted to carry out multi-layer feature extraction on the input image; The characteristic enhancement module is arranged in the backbone network and used for enhancing the target characteristic expression of the abnormal pipeline; The space pyramid pooling module is used for expanding the receptive field; The feature fusion network is used for upsampling, splicing and fusing the features of different levels; The multi-scale detection head is used for outputting the position and type information of the abnormal pipeline target.
6. The method for identifying and positioning abnormal states of pipelines according to claim 1, wherein in S1, environmental characteristic data in a coal mine roadway is collected based on a laser radar, and a priori map of the coal mine roadway is constructed by utilizing Cartographer algorithm after denoising and registering preprocessing are performed on the environmental characteristic data.
7. The method for identifying and locating abnormal states of pipelines according to claim 6, wherein in S1, local environment characteristic data of a current position is obtained in real time through a sensor, the local environment characteristic data is matched with map characteristics in an priori map, fusion calculation is performed by combining with odometer data, and an extended kalman filtering algorithm is adopted in the fusion calculation to determine pose information of the current position in a global map coordinate system.
8. The method for identifying and locating abnormal states of pipelines according to claim 7, wherein in S4, when an abnormal pipeline is detected, pose information corresponding to the moment of abnormality identification is mapped as associated locating information of the abnormal pipeline into a global map to realize position marking of the abnormal pipeline in the global map, and alarm information is output.
9. The method for identifying and locating abnormal states of pipelines according to claim 1, wherein in S2, a median filtering method is adopted to perform noise reduction processing on the pipeline image so as to suppress dust interference and sensor noise, and an image to be identified is obtained.
10. The inspection robot is characterized by comprising a mobile chassis, an image acquisition device, an environment sensing device, an inertia measurement device, an alarm device and a controller, wherein the image acquisition device, the environment sensing device, the inertia measurement device, the alarm device and the controller are arranged on the mobile chassis; The image acquisition device is used for acquiring an intra-roadway pipeline image; the environment sensing device is used for acquiring environment characteristic data in the coal mine tunnel; the inertial measurement device is used for acquiring the attitude information of the inspection robot; The controller is respectively connected with the image acquisition device, the environment sensing device, the inertia measurement device and the alarm device and is configured to execute the pipeline abnormal state identification and positioning method according to any one of claims 1-9; the alarm device is used for outputting alarm information when an abnormal pipeline is detected.

Description

Pipeline abnormal state identification and positioning method and inspection robot thereof Technical Field The application relates to the technical field of machine vision identification, in particular to a pipeline abnormal state identification and positioning method and a patrol robot thereof. Background Coal is used as an important basic energy source in the energy structure of China, and the safe and efficient exploitation of the coal has important significance for the stable operation of national economy. In underground coal mine production systems, water supply and drainage pipelines, spray pipelines and other functional pipelines are key infrastructure for guaranteeing safe production and normal operation of equipment. Because the underground environment of the coal mine is complex and severe, adverse factors such as unstable geological structure, deformation of surrounding rock, high dust concentration, high humidity, poor illumination condition and the like exist for a long time, meanwhile, the pipeline is in a high-load running state for a long time, the problems of ageing, corrosion, connection looseness and the like are easy to occur, and the factors such as installation deviation, incomplete maintenance or improper manual operation are added, so that various pipeline anomalies are easy to be caused. The problems of pipeline inclination, hanging sagging, falling and the like caused by pipeline water leakage, fixed buckles or hanging hooks falling are particularly remarkable. The abnormality not only can cause waste of water resources and energy sources and increase production and operation costs, but also can further induce secondary safety accidents such as water burst, roof collapse, falling objects hurting people and the like. In the prior art, the detection method for the underground pipeline abnormality of the coal mine is mostly focused on typical problems such as pipeline leakage, corrosion, cracks, surface defects and the like, for example, the pipeline leakage points, heat abnormal areas or surface defect areas are identified and positioned through means such as negative pressure waves, electromagnetic induction, machine vision, infrared detection or binocular vision. However, the anomalies such as pipeline drop, suspension sagging and attitude deviation belong to sudden structural anomalies, and have essential differences with leakage or surface defects, the former is represented by that the whole or partial pipeline is separated from the original fixed structure, so that the spatial position change, the attitude inclination and even the falling to the ground are caused, the identification of the pipeline needs to judge whether the target pipeline is abnormal or not, the further analysis of the attitude characteristics of the pipeline and the spatial relationship between the pipeline and the surrounding environment such as the ground, the supporting structure and the like is needed, and the latter usually only relates to the texture, the temperature or the morphology change of the partial area and focuses on the surface characteristic detection. Therefore, the prior art is insufficient in adaptability to abnormal structural postures, and particularly the problems of poor recognition effect, insufficient detection accuracy, high false detection omission rate and the like are easy to occur under complex working conditions such as underground dust shielding, insufficient illumination, disordered background, narrow space, equipment vibration and the like of a coal mine. Disclosure of Invention Therefore, the application provides a method for identifying and positioning abnormal states of pipelines and a patrol robot thereof, which are used for solving the problems of low accuracy and high false detection omission rate of identifying abnormal structural postures such as falling, tilting, hanging sagging and the like of underground pipelines of a coal mine in the prior art. In order to achieve the above object, the present application provides the following technical solutions: a method for identifying and locating abnormal states of pipelines comprises the following steps: S1, constructing a global map of a coal mine tunnel, and establishing a coordinate corresponding relation between an image acquisition position and the global map; S2, acquiring a pipeline image in a roadway and position information corresponding to the pipeline image, and performing noise reduction treatment on the pipeline image to obtain an image to be identified; S3, inputting the image to be identified into a neural network model for identification, obtaining a pipeline target area, judging whether the pipeline is in an abnormal state or not according to the geometric characteristics of the pipeline target area and the spatial relationship between the pipeline target area and the ground area, and outputting a pipeline abnormal identification result; And S4, mapping the position of the abnormal pipeline to a global ma