CN114955814A - Overlong overhead escalator truss deformation monitoring system and detection method - Google Patents

Abstract

A super-long overhead escalator truss deformation monitoring system and detection method relate to the technical field of hidden danger detection of escalators and ladders. The inspection and detection of the bending of the bottom line of the long escalator can significantly reduce the safety risk of the inspection personnel and improve the efficiency of the inspection work. In the super-long overhead escalator truss deformation monitoring system, the acquisition module is connected with the industrial camera, the acquisition module adopts an image and video embedded microprocessor, and the image and video embedded microprocessor is connected with the HT82K629A circuit board, and the HT82K629A circuit board is connected with a button The PC terminal is connected with the acquisition module and has a display module; the power module is connected with the acquisition module and has a controller, and the controller is connected with a battery and 220V direct current.

Description

Super-long overhead escalator truss deformation monitoring system and detection method

technical field

The invention relates to the technical field of hidden danger detection of escalator ladders, in particular to a super-long overhead escalator truss deformation monitoring system and a detection method.

Background technique

Super long overhead escalator is a kind of means of transportation that can transport passengers up and down the floor with the support of the drive device. It is widely used in office buildings, shopping malls, high-speed rail stations and other public places with concentrated traffic.

In the process of installing the super-long overhead escalator, the middle will bend downward due to being too long or its own gravity, and also because the super-long escalator is fastened and connected by strong bolts in sections. The ultra-long overhead escalator will vibrate and wear during daily use, resulting in the loosening of the bolts, and the downward bending deformation of the ultra-long overhead escalator, burying potential safety hazards in daily use.

In the routine inspection, manual inspection is used, which is a more troublesome and tedious work, resulting in the inaccuracy of the final inspection result. Therefore, based on the above considerations and the research trends in recent years, how to provide a machine vision-based system and method for detecting the deformation of a super-long overhead escalator truss has become a technical problem that those skilled in the art need to solve urgently.

SUMMARY OF THE INVENTION

The purpose of the present invention is to provide a super-long overhead escalator truss deformation monitoring system and detection method. The method adopts a non-contact measurement method, which can well overcome the accuracy defect of the contact measurement method and effectively realize the super-long escalator. It can significantly reduce the safety risk of inspection personnel and improve the efficiency of inspection work.

In the super-long overhead escalator truss deformation monitoring system and detection method provided by the present invention, the images of the bending degree under the truss of the super-long overhead escalator are collected and recognized online in real time by means of machine vision, and then the Hough transform curve is transformed into The method and the standard straight line Hough transform method are improved and fused to obtain the classification Hough transform algorithm, which has the characteristics of classification and equidistant point selection. It is used to judge the bending deformation degree of the truss deformation of the super-long overhead escalator. A machine is proposed. An effective method for visual recognition and detection of bending of super-long overhead escalator trusses.

To achieve the above object, the present invention adopts the following technical solutions:

An ultra-long overhead escalator truss deformation monitoring system, comprising: an industrial camera, wherein the industrial camera is used to perform bending and straight line monitoring of a marked target from multiple different angles, and to collect image information;

an acquisition module, the acquisition module is connected with the industrial camera, the acquisition module adopts an image and video embedded microprocessor, and the image and video embedded microprocessor is connected with a HT82K629A circuit board, and the HT82K629A circuit board is connected with a button, The acquisition module is used to use the Hough transform curve transform and the standard straight line Hough transform method to fuse the improved classification Hough transform algorithm model to process the collected image information, identify and mark it, and judge the degree of deformation of the truss of the super-long overhead escalator and whether it exists or not. hidden danger;

PC terminal, the PC terminal is connected with the acquisition module and has a display module, and the PC terminal is used to display the result of the identification and judgment;

A power supply module, which is connected to the acquisition module and has a controller, and the controller is connected with a battery and 220V direct current, and the power supply module is used to supply power to the entire system.

The image and video embedded microprocessor is provided with a buffer and memory; the image and video embedded microprocessor communicates with the HT82K629A circuit board through a USB interface, and the image and video embedded microprocessor communicates with the HT82K629A circuit board through another USB interface Connect and communicate with the industrial camera, the image and video embedded microprocessor is connected and communicated with the display module of the PC terminal through a VGA interface, and the image and video embedded microprocessor communicates with the power supply through another USB interface. The controller of the module is connected to communicate.

A super-long overhead escalator truss deformation detection method, comprising the following steps:

Step S1, select the super-long overhead escalator to be detected;

Step S2, affix datum identification marks at equal intervals at the bottom or side of the escalator, connect the truss deformation system and start collecting the visual information of the escalator image, and simultaneously start the preprocessing of the image information;

Step S3: Integrate and improve the curve transformation algorithm and the standard straight line algorithm to obtain a classification Hough transform algorithm with the function of recognizing straight lines and curves, and this algorithm introduces a classification mechanism and an equidistant point selection mechanism to reduce global search and exhaustion. method or random sampling, and reduce the amount of calculation and improve the efficiency of the algorithm;

Step S4, judging whether the image and the straight line are completely fitted, if the full fitting proves that the truss of the escalator is not deformed, otherwise, there is a potential safety hazard in the bending deformation;

Step S5, displaying the result of the fusion and improved classification Hough transform algorithm model identification and the existence of hidden dangers on the PC client of the display module.

In practical application, the classification Hough transform algorithm specifically includes the following steps:

Step S31, using the Canny operator to obtain the edge of the image to be tested: it is assumed that the edge point field has K feature points (x _i , y _i ), i=1, 2, 3...K, which constitute the boundary point field D;

Step S32, extract the horizontal and vertical gradient information of boundary point with Sobel operator;

Step S33, initializing the parameter space H;

Step S34, construct a classifier, and describe it with 4 classifiers: the 4 classifier divides the candidate straight lines into 4 categories, horizontal, vertical, positive 45°, and negative 45°, and then according to the gradient information of the two directions in step S32. The boundary points are divided into 4 subclasses D _i ; it is assumed that there are K _i edge feature points (x _i , y _i ) in the 4 subclasses after the classifier, i=1, 2, 3...K _i ;

Step S35, carry out parameter space mapping in 4 subclasses: select m edge feature points at every distance t from subclass D _i as seed points; t=K _i /m, is the selection distance; the selection of the specific t value is not only It is related to the value of K and the noise level of the tested image; for each seed point, the edge feature points are randomly selected from D _i and paired with it at a distance t; for each pair of feature points, the virtual Line parameters, and projected on the parameter space;

The formula for calculating the straight line parameters from the feature point pair (x _i , y _i )(x _j , y _j ) is simply given as follows:

ρ _t = _xi cosθ _t +y _i sinθ _t

again:

ρ _t =x _j cosθ _t +y _j sinθ _t

It can be concluded that:

Put θ _t into the formula ρ _t = _xi cosθ _t +y _i sinθ _t , we can get ρ _t ;

Discretize H(θ _t ,ρ _t ), and add 1 to the value of the corresponding unit in the parameter space H; when each subclass space has completed the above operations, go to the next step;

Step S36: Statistical parameter space, looking for the peak value: look for the peak value in the parameter space H; when it is found that the value of a certain H(θ _c , ρ _c ) is greater than the global threshold T, then it is considered that there is an image with (θ c , ρ c ) in the original recognized image. _c , ρ _c ) is a straight line of parameters; all peak points are recorded.

Step S37, straight line fitting and restoration: the original curve can be obtained by calculating all the obtained peak points with the formula ρ _t = _xi cosθ _t +y _i sinθ _t .

Compared with the prior art, the super-long overhead escalator truss deformation monitoring system and detection method of the present invention have the following advantages:

In the super-long overhead escalator truss deformation monitoring system and detection method provided by the present invention, firstly, the identification object of the super-long escalator to be detected is selected, a straight line identification mark is made on the monitoring target, and the super-long overhead automatic machine vision based machine vision is used. The monitoring and identification system of the escalator, that is, the industrial camera monitors and identifies the image information of the marked target from multiple different angles; The classification Hough transform algorithm model processes the collected image information, identifies and labels, and judges the deformation degree of the truss of the super-long overhead escalator and whether there are hidden dangers; finally, the identification and judgment results are displayed in the super-long overhead escalator truss deformation detection system. Display on the module PC terminal. From this analysis, it can be seen that the deformation monitoring system and detection method of the super-long overhead escalator truss improved by the present invention does not rely on manual visual inspection for the deformation and bending detection of the super-long overhead escalator truss, thereby effectively improving the The inspection efficiency is improved, the inspection time of each super-long escalator is shortened, and the labor cost is effectively reduced.

Description of drawings

FIG. 1 is a schematic diagram of the use process of an ultra-long overhead escalator truss deformation monitoring system and a detection method provided by an embodiment of the present invention;

2 is a schematic structural diagram of a super-long overhead escalator truss deformation monitoring system provided by an embodiment of the present invention;

3 is a schematic flowchart of a method for detecting deformation of a super-long overhead escalator truss provided by an embodiment of the present invention.

Reference number:

1-industrial camera; 2-acquisition module; 21-image video embedded microprocessor; 211-buffer; 212-memory; 22-HT82K629A circuit board; 23-button; 3-PC terminal; 31-display module; 4 - Power module; 41 - controller; 42 - battery; 5 - escalator; 6 - truss; 7 - reference identification mark.

Detailed ways

For ease of understanding, the following describes the deformation monitoring system and detection method for the super-long overhead escalator truss provided by the embodiments of the present invention in detail with reference to the accompanying drawings.

An embodiment of the present invention provides an ultra-long overhead escalator truss deformation monitoring system, as shown in FIG. 1 and FIG. 2 , including: an industrial camera 1 , the industrial camera 1 is used to monitor the marked target from multiple different angles. Bending straight line monitoring and identification image information for collection;

The acquisition module 2, the acquisition module 2 is connected with the industrial camera 1, the acquisition module 2 adopts an image and video embedded microprocessor 21, and the image and video embedded microprocessor 21 is connected with the HT82K629A circuit board 22, and the HT82K629A circuit board 22 is connected with a button 23, the The acquisition module 2 is used to fuse the Hough transform curve transformation and the standard straight line Hough transform method. The improved classification Hough transform algorithm model processes the collected image information, identifies and labels, and judges the deformation degree of the truss of the super-long overhead escalator and whether there are hidden dangers. ;

PC terminal 3, the PC terminal 3 is connected with the acquisition module 2, and has a display module 31, and the PC terminal 3 is used to display the result of the identification and judgment;

The power supply module 4 is connected with the acquisition module 2 and has a controller 41, and the controller 41 is connected with a battery 42 and 220V DC, and the power supply module 4 is used to supply power to the entire system.

Among them, as shown in FIG. 2, the above-mentioned image and video embedded microprocessor 21 may be provided with a buffer 211 and a memory 212; and, the image and video embedded microprocessor 21 can be connected and communicated with the HT82K629A circuit board 22 through a USB interface, and the image and video embedded microprocessor 21 can communicate with the HT82K629A circuit board 22. The type microprocessor 21 can communicate with the industrial camera 1 through another USB interface, the image and video embedded microprocessor 21 can communicate with the display module 31 of the PC terminal 3 through the VGA interface, and the image and video embedded microprocessor 21 can communicate with the display module 31 of the PC terminal 3 through the VGA interface. A USB interface is connected and communicated with the controller 41 of the power module 4 .

As shown in FIG. 1 and FIG. 2 , in the super-long overhead escalator truss deformation monitoring system provided by the embodiment of the present invention, a linear datum identification mark 7 is made on the escalator 5, and then the escalator truss deformation detection system is used to detect the deformation from different The angular position is subjected to image monitoring, collection, identification and information collection, and the industrial camera 1 is used to process and identify the information. The power module 4 is used to supply power to the entire monitoring system, and finally the display is displayed on the PC terminal 3 with a display module 31; In other words, the truss 6 of the escalator 5 of the target to be detected is equidistantly pasted with the linear reference identification mark 7, and the image information is collected by the industrial camera 1 in a complete set of monitoring and identification system, and the image and video embedded microprocessor 21 process, input the image recognition model to identify and determine whether the escalator is bent and deformed, and display the result on the display module 31 .

The embodiment of the present invention further provides a method for detecting the deformation of an escalator truss with an ultra-long overhead, comprising the following steps:

Step S1, select the super-long overhead escalator to be detected;

Step S2, affix datum identification marks at equal intervals at the bottom or side of the escalator, connect the truss deformation system and start collecting the visual information of the escalator image, and simultaneously start the preprocessing of the image information;

Step S3: Integrate and improve the curve transformation algorithm and the standard straight line algorithm to obtain a classification Hough transform algorithm with the function of recognizing straight lines and curves, and this algorithm introduces a classification mechanism and an equidistant point selection mechanism to reduce global search and exhaustion. method or random sampling, and reduce the amount of calculation and improve the efficiency of the algorithm;

Step S4, judging whether the image and the straight line are completely fitted, if the full fitting proves that the truss of the escalator is not deformed, otherwise, there is a potential safety hazard in the bending deformation;

Step S5, displaying the result of the fusion and improved classification Hough transform algorithm model identification and the existence of hidden dangers on the PC client of the display module.

In practical application, the classification Hough transform algorithm of the above step S3 may specifically include the following steps:

Step S31, using the Canny operator to obtain the edge of the image to be tested: it is assumed that the edge point field has K feature points (x _i , y _i ), i=1, 2, 3...K, which constitute the boundary point field D;

Step S32, extract the horizontal and vertical gradient information of boundary point with Sobel operator;

Step S33, initializing the parameter space H;

Step S34, construct a classifier, and describe it with 4 classifiers: the 4 classifier divides the candidate straight lines into 4 categories, horizontal, vertical, positive 45°, and negative 45°, and then according to the gradient information of the two directions in step S32. The boundary points are divided into 4 subclasses D _i ; it is assumed that there are K _i edge feature points (x _i , y _i ) in the 4 subclasses after the classifier, i=1, 2, 3...K _i ;

Step S35, carry out parameter space mapping in 4 subclasses: select m edge feature points at every distance t from subclass D _i as seed points; t=K _i /m, is the selection distance; the selection of the specific t value is not only It is related to the value of K and the noise level of the tested image; for each seed point, the edge feature points are randomly selected from D _i and paired with it at a distance t; for each pair of feature points, the virtual Line parameters, and projected on the parameter space;

The formula for calculating the straight line parameters from the feature point pair (x _i , y _i )(x _j , y _j ) is simply given as follows:

ρ _t = _xi cosθ _t +y _i sinθ _t

again:

ρ _t =x _j cosθ _t +y _j sinθ _t

It can be concluded that:

Put θ _t into the formula ρ _t = _xi cosθ _t +y _i sinθ _t , we can get ρ _t ;

Discretize H(θ _t ,ρ _t ), and add 1 to the value of the corresponding unit in the parameter space H; when each subclass space has completed the above operations, go to the next step;

Step S36: Statistical parameter space, looking for the peak value: look for the peak value in the parameter space H; when it is found that the value of a certain H(θ _c , ρ _c ) is greater than the global threshold T, then it is considered that there is an image with (θ c , ρ c ) in the original recognized image. _c , ρ _c ) is a straight line of parameters; all peak points are recorded.

Step S37, straight line fitting and restoration: the original curve can be obtained by calculating all the obtained peak points with the formula ρ _t = _xi cosθ _t +y _i sinθ _t .

As shown in FIG. 3 , in the method for detecting the deformation of the super-long overhead escalator truss provided by the embodiment of the present invention, the escalator to be inspected is selected first, and identification marks are attached to the escalator and the installation and debugging of the monitoring system are performed according to the needs of the site; Perform machine vision recognition on the escalator sticker label position at each angle position, and carry out image processing work, and then establish a new monitoring and recognition model by integrating and improving the recognition model algorithm, and use it to detect the target to be inspected to determine whether there is bending deformation. hidden safety hazards; fuse the improved new monitoring and identification algorithm process, (1) input the collected video image, (2) perform image preprocessing on it, (3) obtain the detection edge point D, (4) extract the gradient, (5) Initialize parameter space, (6) algorithm parameter setting, (7) classifier selection and construction, (8) select seed point Gi, (9) select feature point Pj and pairing, (10) calculate line parameters, (11) parameter space Mapping and accumulation, (12) judge whether i is less than m, (13) yes, return to (8) to re-select the seed point Gi, continue to execute until i is greater than m, (14) judge whether j is less than m, (15) yes , return to (8) to re-select the feature point Pj and pairing, continue to execute until j is greater than m, until the line fitting and restoration results are completed; finally, judge whether the image and the marked line are completely fitted, and the complete fitting is used to determine the super-long escalator The truss is safe, otherwise it is judged to be a safety hazard.

Compared with the prior art, the super-long overhead escalator truss deformation monitoring system and detection method described in the embodiments of the present invention have the following advantages:

In the super-long overhead escalator truss deformation monitoring system and detection method provided by the embodiment of the present invention, firstly, the identification object of the super-long escalator to be detected is selected, a straight line identification mark is made on the monitoring target, and the super-long escalator based on machine vision is used. The monitoring and identification system of overhead escalator, that is, the industrial camera monitors and recognizes the image information of the marked target from multiple different angles; Integrate the improved classification Hough transform algorithm model to process, identify and label the collected image information, and judge the deformation degree of the truss of the super-long overhead escalator and whether there are hidden dangers; finally, the identification and judgment results are used in the super-long overhead escalator truss deformation detection system. display on the PC terminal of the display module. From this analysis, it can be seen that the deformation monitoring system and detection method of the super-long overhead escalator truss improved by the present invention does not rely on manual visual inspection for the deformation and bending detection of the super-long overhead escalator truss, thereby effectively improving the The inspection efficiency is improved, the inspection time of each super-long escalator is shortened, and the labor cost is effectively reduced.

The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention. should be included within the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims (4)

1. a super-long overhead escalator truss deformation monitoring system is characterized in that, comprising: An industrial camera, which is used to monitor and identify the image information of the marked target from a plurality of different angles; an acquisition module, the acquisition module is connected with the industrial camera, the acquisition module adopts an image and video embedded microprocessor, and the image and video embedded microprocessor is connected with a HT82K629A circuit board, and the HT82K629A circuit board is connected with a button, The acquisition module is used to use the Hough transform curve transform and the standard straight line Hough transform method to fuse the improved classification Hough transform algorithm model to process the collected image information, identify and mark it, and judge the degree of deformation of the truss of the super-long overhead escalator and whether it exists or not. hidden danger; PC terminal, the PC terminal is connected with the acquisition module and has a display module, and the PC terminal is used to display the result of the identification and judgment; A power supply module, which is connected to the acquisition module and has a controller, and the controller is connected with a battery and 220V direct current, and the power supply module is used to supply power to the entire system. 2. The super-long overhead escalator truss deformation monitoring system according to claim 1, wherein the image and video embedded microprocessor is provided with a buffer zone and a memory; the image and video embedded microprocessor is provided with a USB interface Connect and communicate with the HT82K629A circuit board, the image and video embedded microprocessor communicates with the industrial camera through another USB interface, and the image and video embedded microprocessor communicates with the PC terminal through the VGA interface. The display module is connected and communicated, and the image and video embedded microprocessor is connected and communicated with the controller of the power module through another USB interface. 3. a super-long overhead escalator truss deformation detection method, is characterized in that, comprises the following steps: Step S1, select the super-long overhead escalator to be detected; Step S2, affix datum identification marks at equal intervals at the bottom or side of the escalator, connect the truss deformation system and start collecting the visual information of the escalator image, and simultaneously start the preprocessing of the image information; Step S3: Integrate and improve the curve transformation algorithm and the standard straight line algorithm to obtain a classification Hough transform algorithm with the function of recognizing straight lines and curves, and this algorithm introduces a classification mechanism and an equidistant point selection mechanism to reduce global search and exhaustion. method or random sampling, and reduce the amount of calculation and improve the efficiency of the algorithm; Step S4, judging whether the image and the straight line are completely fitted, if the full fitting proves that the truss of the escalator is not deformed, otherwise, there is a potential safety hazard in the bending deformation; Step S5, displaying the result of the fusion and improved classification Hough transform algorithm model identification and the existence of hidden dangers on the PC client of the display module. 4. super-long overhead escalator truss deformation detection method according to claim 3, is characterized in that, described classification Hough transform algorithm, specifically comprises the following steps: Step S31, using the Canny operator to obtain the edge of the image to be tested: it is assumed that the edge point field has K feature points (x _i , y _i ), i=1, 2, 3...K, which constitute the boundary point field D; Step S32, extract the horizontal and vertical gradient information of boundary point with Sobel operator; Step S33, initializing the parameter space H; Step S34, construct a classifier, and describe it with 4 classifiers: the 4 classifier divides the candidate straight lines into 4 categories, horizontal, vertical, positive 45°, and negative 45°, and then according to the gradient information of the two directions in step S32. The boundary points are divided into 4 subclasses D _i ; it is assumed that there are K _i edge feature points (x _i , y _i ) in the 4 subclasses after the classifier, i=1, 2, 3...K _i ; Step S35, carry out parameter space mapping in 4 subclasses: select m edge feature points at every distance t from subclass D _i as seed points; t=K _i /m, is the selection distance; the selection of the specific t value is not only It is related to the value of K and the noise level of the tested image; for each seed point, the edge feature points are randomly selected from D _i and paired with it at a distance t; for each pair of feature points, the virtual Line parameters, and projected on the parameter space; The formula for calculating the straight line parameters from the feature point pair (x _i , y _i )(x _j , y _j ) is simply given as follows: ρ _t = _xi cosθ _t +y _i sinθ _t again: ρ _t =x _j cosθ _t +y _j sinθ _t It can be concluded that:

Put θ _t into the formula ρ _t = _xi cosθ _t +y _i sinθ _t , we can get ρ _t ; Discretize H(θ _t ,ρ _t ), and add 1 to the value of the corresponding unit in the parameter space H; when each subclass space has completed the above operations, go to the next step; Step S36: Statistical parameter space, looking for the peak value: look for the peak value in the parameter space H; when it is found that the value of a certain H(θ _c , ρ _c ) is greater than the global threshold T, then it is considered that there is an image with (θ c , ρ c ) in the original recognized image. _c , ρ _c ) is a straight line of parameters; all peak points are recorded. Step S37, straight line fitting and restoration: the original curve can be obtained by calculating all the obtained peak points with the formula ρ _t = _xi cosθ _t +y _i sinθ _t .

Images