CN-121280365-B - Method and system for detecting staggered table defect of single track finger-shaped plate

CN121280365BCN 121280365 BCN121280365 BCN 121280365BCN-121280365-B

Abstract

The invention discloses a method and a system for detecting the fault of a finger-shaped plate of a monorail track, which relate to the technical field of track traffic detection and are used for collecting two-dimensional texture images and three-dimensional depth images of tracks to be inspected and extracting reference depth values of planes of adjacent track beams, preprocessing the two-dimensional texture images, then adopting target detection, extracting rectangular frame images of finger-shaped plate areas, cutting finger areas and positioning and restoring the finger areas into the three-dimensional depth images, obtaining depth values of central points of the finger areas and putting the depth values of the same sides into the same set, carrying out primary adjustment on the depth values in the set based on the reference depth values of the track beams, analyzing the primary adjusted depth values to obtain analysis results, carrying out secondary adjustment based on the depth value analysis results, finally combining the depth values of the adjacent track beams after secondary adjustment, and outputting fault data of the fault of the finger-shaped plate. The invention solves the problems of high cost, low accuracy and low efficiency of traditional manual inspection in the field of fault detection.

Inventors

MAO HONGJUN
HU CHENGKAI
Lan Yayi

Assignees

成都精工华耀科技有限公司

Dates

Publication Date: 20260508
Application Date: 20250929

Claims (10)

1. A method for detecting the dislocation defect of a single track finger-shaped plate is characterized by comprising the following steps: Acquiring a two-dimensional texture image and a three-dimensional depth image of a track to be inspected, extracting depth values of planes of adjacent track beams based on the three-dimensional depth image, and recording the depth values as reference depth values of corresponding track beams; inputting the preprocessed two-dimensional texture image into a target detection network for target detection, and extracting a rectangular frame image containing a finger-shaped plate area; Cutting all finger areas of each side of the track beam based on the rectangular frame image to obtain all finger images, extracting corresponding position information, and positioning and restoring to the corresponding finger areas in the three-dimensional depth image based on the position information of all finger images; Based on the positioned three-dimensional depth image, acquiring a depth value of a central point of each finger area, and putting the depth values of all finger areas of the track beam on the same side into the same set; Based on the reference depth values corresponding to the adjacent track beams, adjusting all depth values in the corresponding set once; analyzing the depth value after the primary adjustment in the corresponding set to obtain an analysis result, and secondarily adjusting the depth value after the primary adjustment in the corresponding set based on the analysis result; and combining the depth values after secondary adjustment in the corresponding sets of the adjacent track beams, and outputting fault data of the staggered platform of the finger-shaped plates.
2. The method for detecting the dislocation defect of the finger-shaped plate of the monorail track as claimed in claim 1, wherein extracting depth values of adjacent rail beam planes based on the three-dimensional depth image comprises: based on the acquired three-dimensional depth image, cutting the three-dimensional depth image related to the track beam, and recording the three-dimensional depth image as a first three-dimensional depth image; And obtaining depth values at a plurality of preset points in the first three-dimensional depth image, and calculating the average value of the depth values at all the preset points to serve as the depth value of the corresponding track beam plane.
3. A method of detecting a monorail track finger slab staggering defect in accordance with claim 1, wherein preprocessing the two-dimensional texture image comprises: and denoising and filtering the two-dimensional texture image.
4. The method for detecting the dislocation defect of the finger-shaped plate of the monorail track as claimed in claim 1, wherein the step of cutting all finger areas of each side of the track beam based on the rectangular frame image comprises the steps of: acquiring the rectangular block image and performing binarization processing to obtain a binarized image; Performing primary processing on the binarized image by adopting a global threshold value, and optimizing the primarily processed binarized image by adopting morphological operation; detecting the boundary of each finger area by adopting a contour based on the optimized binarization image, and screening out an effective finger contour based on the priori characteristics of the finger plates; And cutting all finger areas of each side of the track beam based on all the screened finger outlines.
5. The method for detecting the dislocation defect of the finger-shaped plate of the monorail track as claimed in claim 1, wherein the step of obtaining the depth value of the center point of each finger area based on the positioned three-dimensional depth image comprises the steps of: Acquiring a gray scale image of each finger area based on all finger images; setting a gray threshold value, processing the gray map of each finger area, and screening to obtain effective pixels of each finger area; Calculating global gravity centers of all effective pixels in each finger area as corresponding center points, and synchronously extracting the position information of the center points; And acquiring a depth value of each finger region center point based on the positioned three-dimensional depth image and the position information of the corresponding finger region center point.
6. The method for detecting the dislocation defect of a finger-shaped plate of a monorail track as claimed in claim 1, wherein the step of adjusting all depth values in the corresponding set once based on the reference depth values corresponding to the adjacent track beams comprises: Recording one reference depth value corresponding to the adjacent track beam as a first reference depth value, and recording the other reference depth value as a second reference depth value; Calculating the difference between the first reference depth value and the second reference depth value to be a first difference value, judging whether the first difference value exceeds a first set threshold value, if so, calculating the difference between the first difference value and the first set threshold value to be a compensation value, and compensating all depth values in a corresponding set based on the compensation value to finish one-time adjustment, otherwise, compensating all depth values in the corresponding set with a null value to finish one-time adjustment.
7. The method for detecting the dislocation defect of the finger-shaped plate of the monorail track as claimed in claim 1, wherein the step of analyzing the depth value once adjusted in the corresponding set to obtain the analysis result comprises the steps of: Acquiring all depth values in the corresponding set after one-time adjustment, and arranging the depth values based on the position information of the finger images to obtain a depth value sequence; Calculating the variance corresponding to the depth value sequence, judging whether the variance is smaller than a set value, if yes, judging that the plane where the corresponding finger-shaped plate is positioned is parallel to the plane of the corresponding track beam, and if not, judging that the plane where the corresponding finger-shaped plate is positioned is intersected with the plane of the corresponding track beam; And traversing corresponding depth value sequences when judging that the plane where the corresponding finger-shaped plate is positioned is parallel to the plane of the corresponding track beam or judging that the plane where the corresponding finger-shaped plate is positioned is intersected with the plane of the corresponding track beam, constructing a depth value change curve, performing fitting treatment to obtain a fitting straight line, calculating the geometric distance between each depth value in the corresponding depth value sequence and the fitting straight line to obtain a calculation result, and taking the calculation result as the analysis result.
8. The method for detecting a dislocation defect in a finger-shaped plate of a monorail track as claimed in claim 7, wherein based on the analysis result, performing secondary adjustment on the depth value once adjusted in the corresponding set comprises: and judging whether the geometric distance between each depth value in the corresponding depth value sequence and the fitting straight line falls within a set range or not based on the calculation result, if so, performing null operation on the depth values which are once adjusted in the corresponding set to finish secondary adjustment, and if not, screening out the depth values which are corresponding to the geometric distance of the fitting straight line exceeding the set range, and performing secondary adjustment to ensure that the distances between the depth values after secondary adjustment and the fitting straight line fall within the set range.
9. The method for detecting a dislocation defect of a finger-shaped plate of a monorail track according to claim 8, wherein the step of outputting the dislocation fault data of the finger-shaped plate in combination with the depth value after the secondary adjustment in the corresponding set of adjacent track beams comprises: The finger-shaped plates corresponding to the adjacent track beams are respectively marked as a first finger-shaped plate and a second finger-shaped plate; The depth value after secondary adjustment in the first finger-shaped plate corresponding set is recorded as a first depth value, and the depth value after secondary adjustment in the second finger-shaped plate corresponding set is recorded as a second depth value; Constructing a fitting plane corresponding to the first finger-shaped plate based on all the first depth values, and marking the fitting plane as a first plane, and constructing a fitting plane corresponding to the second finger-shaped plate based on all the second depth values, and marking the fitting plane as a second plane; Judging whether the first plane is parallel to the second plane, if so, calculating the geometric distance between the first plane and the second plane, and when the geometric distance exceeds a second set threshold, judging that the first finger-shaped plate and the second finger-shaped plate have longitudinal parallel dislocation faults and quantitatively outputting finger-shaped plate dislocation values according to the geometric distance, and if not, judging that the first finger-shaped plate and the second finger-shaped plate have longitudinal cross dislocation faults and quantitatively outputting finger-shaped plate looseness angle values according to fitting straight lines corresponding to the first finger-shaped plate or the second finger-shaped plate.
10. A monorail track finger slab staggering defect detection system, comprising: the acquisition unit is used for acquiring a two-dimensional texture image and a three-dimensional depth image of a track to be inspected, extracting depth values of planes of adjacent track beams based on the three-dimensional depth image, and recording the depth values as reference depth values of the corresponding track beams; the target detection unit is used for preprocessing the two-dimensional texture image, inputting the preprocessed two-dimensional texture image into a target detection network to perform target detection, and extracting a rectangular frame image containing a finger-shaped plate area; The positioning unit is used for cutting all finger areas of each side of the track beam based on the rectangular frame image, obtaining all finger images, extracting corresponding position information, and positioning and restoring to the corresponding finger areas in the three-dimensional depth image based on the position information of all the finger images; the depth value unit is used for acquiring the depth value of the center point of each finger area based on the positioned three-dimensional depth image and placing the depth values of all the finger areas of the track beam on the same side into the same set; the primary adjusting unit is used for adjusting all depth values in the corresponding set once based on the reference depth values corresponding to the adjacent track beams; The secondary adjustment unit is used for analyzing the depth value after the primary adjustment in the corresponding set to obtain an analysis result, and performing secondary adjustment on the depth value after the primary adjustment in the corresponding set based on the analysis result; and the dislocation detection unit is used for combining the depth values after secondary adjustment in the corresponding set of the adjacent track beams and outputting fault data of the dislocation of the finger-shaped plate.

Description

Method and system for detecting staggered table defect of single track finger-shaped plate Technical Field The invention relates to the technical field of rail transit detection, in particular to a method and a system for detecting the fault of a single track finger-shaped plate. Background The straddle type monorail line is composed of a plurality of track beams, and expansion joints are required to be arranged at the beam ends because the track beams can generate expansion deformation due to the factors of foundation settlement, shrinkage and creep of concrete, temperature change and the like. The train can generate larger vibration when passing through the expansion joint, thereby reducing the riding comfort of passengers and increasing the abrasion of the tires of the monorail transit train and the long-term maintenance cost of the track beam. In order to ensure that Liang Tizheng running surfaces of the side wheels can drive vehicles stably and safely, the rail joints are required to be provided with mutually crossed and horizontally connected finger-shaped plates, so that free movable gaps of the beam bodies can be ensured, and the vehicles can drive stably. However, the deformation and deflection of the track beam can occur during the long-term running process of the circuit due to the influence of external factors, the loosening and falling of the fixing bolts of the finger plate can also occur, the finger plate cannot be fixed, and finally the loosening and the tilting of the finger plate and the dislocation abnormality of the monorail beam can be caused, if the deformation and the tilting of the finger plate and the dislocation abnormality of the monorail beam are not treated in time, the tire and the vibration of the vehicle can be damaged during the running process of the vehicle, and the running safety of the vehicle is seriously endangered. At present, the detection mode of the disease of the finger-shaped plate Liang Cuotai still stays in the manual inspection mode, so that the labor cost is high, and the inspection accuracy and the inspection efficiency are low. Disclosure of Invention In view of the above, the application provides a method and a system for detecting the dislocation defect of a single track finger-shaped plate, so as to solve the defects in the prior art. The first aspect of the application provides a method for detecting the dislocation defect of a single track finger-shaped plate, which comprises the following steps: Acquiring a two-dimensional texture image and a three-dimensional depth image of a track to be inspected, extracting depth values of planes of adjacent track beams based on the three-dimensional depth image, and recording the depth values as reference depth values of corresponding track beams; inputting the preprocessed two-dimensional texture image into a target detection network for target detection, and extracting a rectangular frame image containing a finger-shaped plate area; Cutting all finger areas of each side of the track beam based on the rectangular frame image to obtain all finger images, extracting corresponding position information, and positioning and restoring to the corresponding finger areas in the three-dimensional depth image based on the position information of all finger images; Based on the positioned three-dimensional depth image, acquiring a depth value of a central point of each finger area, and putting the depth values of all finger areas of the track beam on the same side into the same set; Based on the reference depth values corresponding to the adjacent track beams, adjusting all depth values in the corresponding set once; analyzing the depth value after the primary adjustment in the corresponding set to obtain an analysis result, and secondarily adjusting the depth value after the primary adjustment in the corresponding set based on the analysis result; and combining the depth values after secondary adjustment in the corresponding sets of the adjacent track beams, and outputting fault data of the staggered platform of the finger-shaped plates. In a possible implementation manner of the first aspect, based on the three-dimensional depth image, extracting depth values of adjacent track beam planes includes: based on the acquired three-dimensional depth image, cutting the three-dimensional depth image related to the track beam, and recording the three-dimensional depth image as a first three-dimensional depth image; And obtaining depth values at a plurality of preset points in the first three-dimensional depth image, and calculating the average value of the depth values at all the preset points to serve as the depth value of the corresponding track beam plane. In a possible implementation manner of the first aspect, preprocessing the two-dimensional texture image includes: and denoising and filtering the two-dimensional texture image. In a possible implementation manner of the first aspect, based on the rectangular frame image, cutt