Scraper chain plate skew detection method based on machine vision
Through machine vision-based detection methods, the skewed state of the chain plate of the scraper machine is monitored in real time, which solves the problem of lack of chain plate detection function in the existing technology, and achieves efficient operation and safety improvement of the scraper machine.
Patent Information
- Application Number
- CN202411240695.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-05
- Publication Date
- 2024-12-03
AI Technical Summary
The existing scraper equipment lacks the chain plate skew detection function, which leads to inaccurate manual inspection and easy damage to the sensor, which cannot effectively monitor the operating status of the scraper, which in turn leads to equipment shutdown and potential accidents.
Using machine vision-based detection methods, images are collected through industrial cameras, image processing and analysis are carried out, skewed status of the chain plate is identified, real-time monitoring and alarms are issued, and hardware damage and downtime are avoided.
Real-time detection of the skewed chain plate of the scraper is realized, which reduces detection errors caused by human factors, avoids hardware damage, improves production efficiency and safety, prevents the expansion of faults in a timely manner, and reduces maintenance costs.
Smart Images

Figure CN119059207A_ABST
Abstract
Description
Technical Field
[0001] This patent application belongs to the field of scraper conveyor visual detection technology, and more specifically, it relates to a method for detecting the skewness of a scraper conveyor chain based on machine vision. Background Art
[0002] Scrapers are used in the mining industry to transport materials, especially in the coal washing process. They are indispensable equipment. The main mechanism includes a certain number of chain plates connected to two chains. One end of the chain is driven by a driving mechanism. The chain plate moves with the chain in the groove, hanging the material and moving forward to achieve the purpose of conveying the material. However, the problem of chain plate skewing will occur during the conveying process, which will force the equipment to stop operating. The current detection method is to perform manual inspections and add sensors that can respond to certain state parameters of the scraper. Since the internal working conditions of the scraper are random and uncontrollable, if the chain plate is skewed, it cannot be monitored, the sensor is easily damaged, and manual inspections cannot be carried out in real time. Therefore, the existing scraper equipment does not have the function of detecting whether the chain plate is skewed. Summary of the invention
[0003] The technical problem to be solved by the present invention is to provide a method for detecting the skewness of a scraper chain based on machine vision, which solves the drawbacks of the related technology, has the characteristics of being able to detect in real time, avoids the problem of hardware damage caused by poor working conditions, and ensures accuracy.
[0004] In order to solve the above problems, the technical solution adopted by the present invention is:
[0005] A method for detecting the skewness of a scraper chain based on machine vision comprises the following steps:
[0006] Step 1: The industrial camera is fixedly installed in the box, the box is connected and fixed to the scraper, and the industrial camera shoots vertically downward to collect images of the working surface of the scraper;
[0007] Step 2: Select a local analysis area in the image. The requirements for selecting this area are: the upper and lower ranges are located in the center between the two chain plates, the length is 80% to 90% of the distance between the chain plates, and the chain plates cannot be included in the area. The length of the left and right ranges is 50% to 90% of the distance between the two chain plates. The requirements for selecting the left and right ranges are: only the chain plates are included in the range and there are no other obstructions.
[0008] Step 3: Convert the image selected in the local analysis area into a grayscale image;
[0009] Step 4: The horizontal x direction in the image is the direction of chain movement, and the gradient threshold is used along the direction of chain movement: abs_sobelx = √(sobel x ) 2Perform image processing and convert the absolute value image to 8 bits;
[0010] step5: Convert binary image format;
[0011] Step 6: Perform a closing operation on the binary image to connect discontinuous lines, and then perform two opening operations to remove noise points in the binary image and filter out horizontal lines;
[0012] Step7: Perform contour search on the processed binary image;
[0013] Step 8: Traverse all identified contours, calculate the contour circumscribed rectangle to obtain the coordinates (x, y), width and height (w, h) of the upper left corner of the circumscribed rectangle. If the contour height value h is greater than the required value of the local analysis area height H, it is confirmed that it may be a chain plate, and the circumscribed rectangle is marked in green, and other circumscribed rectangles are marked in red, such as Figure 2 As shown, the contour information that meets the above conditions is counted and sorted from left to right according to the position to obtain [contours1, contours2...contoursN];
[0014] Step 9: Calculate the first fitting straight line of contours1 in the above statistical contour information, and obtain the coefficient value of the one-variable linear expression of the fitting straight line and the coordinates of a certain point through which the fitting straight line passes;
[0015] Step 10: Use the inclination of the fitted straight line line 1 of the first contours 1 on the left to represent the scraper posture;
[0016] Step 11: Calculate the angle between the scraper's fitting straight line line 1 and the vertical line in the local analysis area. If the angle value exceeds the threshold, go to step 12, otherwise continue to the next frame of image recognition;
[0017] step12: Send out an oblique pull alarm signal.
[0018] Due to the adoption of the above technical solution, the beneficial effects achieved by the present invention are:
[0019] This method can perform real-time detection to avoid the impact of detection errors caused by human factors such as the intermittent nature of manual detection; and the device structure can be arranged on the outside of the scraper instead of the inside, to avoid hardware damage caused by poor internal working conditions, and it is more practical. Through the detection device, the chain skipping or inclined pull failure of the scraper can be discovered in time to prevent equipment damage or more serious accidents caused by failures not being discovered in time. The operating status of the scraper can be monitored in real time, so that intervention can be made at the early stage of the failure, avoiding high maintenance costs caused by the expansion of the failure. At the same time, reducing downtime caused by equipment failure can ensure the continuity of production, thereby improving overall production efficiency; the real-time monitoring can issue an alarm in time to avoid or reduce personal injury. BRIEF DESCRIPTION OF THE DRAWINGS
[0020] Figure 1 It is a schematic diagram of the overall structure of the present invention;
[0021] Figure 2 is a binary image processing result of a certain frame in the present invention;
[0022] Figure 3 is a processing result of a certain frame of image in the present invention;
[0023] Figure 4 It is a flow chart of the visual inspection method of the present invention.
[0024] In the figure: industrial camera 1, edge computer 2, box 3, chain plate 4, track 5. DETAILED DESCRIPTION
[0025] The present invention is further described in detail below with reference to the embodiments.
[0026] A method for detecting the skewness of a scraper chain based on machine vision, such as Figure 4 As shown, the following steps are included:
[0027] step1: Figure 1 As shown, the industrial camera 1 is fixedly installed in the box 3, and the box 3 is connected and fixed to the scraper. The industrial camera 1 shoots vertically downward to collect images of the working surface of the scraper. The scraper is provided with a chain plate 4, and the chain plate 4 has a matching track 5. An edge computer 2 is installed outside the scraper. The edge computer 2 is used to connect to the industrial camera 1 through a network cable, and is used to receive the images collected by the industrial camera 1 and process the images.
[0028] Step 2: Select a local analysis area in the image. The requirements for selecting this area are: the upper and lower ranges are located in the middle between the two chain plates 4, the length is 80% to 90% of the distance between the chain plates 4, and the chain plates 4 cannot be included in the area. The length of the left and right ranges is 50% to 90% of the distance between the two chain plates 4. The requirements for selecting the left and right ranges are: only the chain plates 4 are included in the range and there are no other obstructions. Figure 3 The green box shown is used to segment the region of interest in the image.
[0029] Step 3: Convert the selected image in the local analysis area into a grayscale image.
[0030] Step 4: The horizontal x direction in the image is the direction of the chain plate movement. The gradient threshold is used along the direction of the chain plate movement. The Sobel operator includes two convolution kernels, which are used to detect the edges in the horizontal and vertical directions to obtain the gradient image in each direction. Then, the "abs_sobelx" operation calculates the sum of the absolute values of this x-direction gradient image to obtain a single image that highlights the edges of the numerical line segments in the image: abs_sobelx = √(sobel x ) 2 Perform image processing and convert the absolute value image into an 8-bit image format with a value range of 0 to 255.
[0031] Step 5: Convert the binary image format and use Otsu's method to automatically determine the optimal threshold. Otsu's method is a global thresholding technique that determines the optimal threshold by minimizing the intra-class variance, thereby dividing the image into two categories: foreground and background.
[0032] Step 6: Perform a closing operation on the binary image with the kernel (3, 3) to connect the discontinuous lines, then perform an opening operation with the kernel (3, 3) to remove the noise in the image, and finally use a kernel of size (1, 3) to perform an opening operation to filter out the horizontal lines. The movement direction of the chain plate is horizontal.
[0033] Step 7: Search for contours in the processed binary image, only retrieve the outermost contours and ignore the inner contours.
[0034] Step 8: Traverse all identified contours. If the coordinates (x, y), width and height (w, h) of the upper left corner of the contour circumscribed rectangle are calculated, and if the contour height value h is greater than the required value of the local analysis area height H, the required value can be 70% to 90%. Taking 80% as an example, if h>H*80%, it is confirmed that it may be the chain plate 4, and the circumscribed rectangle is marked in green, and other circumscribed rectangles are marked in red or other distinguishable colors, such as Figure 2As shown, the contour information that meets the above conditions is counted and sorted from left to right according to the position to obtain [contours1, contours2...contoursN].
[0035] Step 9: Calculate the fitting line of the first contours1 in the above statistical contour information, and obtain the coefficient value of the one-variable linear expression of the fitting line and the coordinates of a certain point through which the fitting line passes.
[0036] Step 10: Use the inclination of the fitted straight line line 1 of the first contours 1 on the left to represent the scraper posture. The fitted straight line line 1 can be displayed as follows: Figure 3 The blue line is shown.
[0037] Step 11: Calculate the angle between the scraper fitting line 1 and the vertical line of the local analysis area. If the angle value exceeds the threshold of 5°, continue to detect and count the angle values of 10 consecutive frames of images. If all 10 consecutive frames exceed the threshold of 5°, enter step 12, otherwise continue to recognize the next frame of image. Of course, the angle value does not necessarily have to exceed the threshold of 5°, it can also exceed 10°, 8°, 6°, 4° and other degree values.
[0038] step12: Send out an oblique pull alarm signal.
[0039] In addition, the present invention also provides a detection device for the skewness of a scraper chain based on machine vision, which can implement the method of the present invention. The detection device includes an image acquisition unit, an image processing unit, and an image display unit. The image processing unit connects the image acquisition unit and the image display unit. The image acquisition unit is used to execute step 1, the image processing unit is used to execute step 2 to step 12, and the image display unit is used to display the detection result. Although the industrial camera 1 is mentioned above as the image acquisition unit and the edge computer 2 as the image processing unit, there are no specific requirements for the image acquisition unit, the image processing unit, and the image display unit, as long as the purpose of the present invention can be achieved. For example, the image display unit can be a liquid crystal screen, an industrial computer, etc., which are not described one by one.
[0040] In summary, the present invention can timely detect chain skipping or inclined pull failures of the scraper, prevent equipment damage or more serious accidents caused by failures not being discovered in time, monitor the operating status of the scraper in real time, and intervene in the early stage of the failure, avoiding high maintenance costs caused by the expansion of the failure. It can also reduce downtime caused by equipment failure, issue alarms in time, ensure production continuity, improve overall production efficiency, and avoid or reduce personal injury.
Claims
1. A method for detecting the skewness of a scraper chain based on machine vision, characterized in that The steps include: Step 1: The industrial camera is fixedly installed in the box, the box is connected and fixed to the scraper, and the industrial camera shoots vertically downward to collect images of the working surface of the scraper; Step 2: Select a local analysis area in the image. The requirements for selecting this area are: the upper and lower ranges are located in the center between the two chain plates, the length is 80% to 90% of the distance between the chain plates, and the chain plates cannot be included in the area. The length of the left and right ranges is 50% to 90% of the distance between the two chain plates. The requirements for selecting the left and right ranges are: only the chain plates are included in the range and there are no other obstructions. Step 3: Convert the image selected in the local analysis area into a grayscale image; Step 4: The horizontal x direction in the image is the direction of chain movement, and the gradient threshold abs_sobelx=√(sobel x ) 2 Perform image processing and convert the absolute value image into 8-bit format; step5: Convert binary image format; Step 6: Perform a closing operation on the binary image to connect discontinuous lines, and then perform two opening operations to remove noise points in the binary image and filter out horizontal lines; Step7: Perform contour search on the processed binary image; Step 8: Traverse all identified contours, calculate the contour circumscribed rectangle to obtain the coordinates (x, y), width and height (w, h) of the upper left corner of the circumscribed rectangle. If the contour height value h is greater than the required value of the local analysis area height H, it is confirmed to be a chain plate, and the circumscribed rectangle is marked in green, and the other circumscribed rectangles are marked in red, as shown in Figure 2. The contour information that meets the above conditions is counted and sorted by position to obtain [contours1,contours2...contoursN]; Step 9: Calculate the first fitting straight line of contours1 in the above statistical contour information, and obtain the coefficient value of the one-variable linear expression of the fitting straight line and the coordinates of a certain point through which the fitting straight line passes; Step 10: Use the inclination state of the fitted straight line line1 of the first contours1 to represent the scraper posture; Step 11: Calculate the angle between the scraper's fitting straight line line 1 and the vertical line in the local analysis area. If the angle value exceeds the threshold, go to step 12, otherwise continue to the next frame of image recognition; step12: Send out an oblique pull alarm signal.
2. According to the method for detecting the skewness of a scraper chain based on machine vision according to claim 1, it is characterized in that: In step 2, the length is 85% to 90% of the distance between the chain plates, and the length of the left and right range is 60% to 80% of the distance between the two chain plates.
3. According to the method for detecting the skewness of a scraper chain based on machine vision of claim 1, it is characterized in that: In step 6, a closing operation with kernel (3, 3) is performed on the binary image to connect discontinuous lines, and then an opening operation with kernel (3, 3) is performed to remove noise in the image. Finally, an opening operation with kernel size (1, 3) is performed to filter out horizontal lines.
4. The method for detecting the skewness of a scraper chain based on machine vision according to claim 1 is characterized in that: In step 8, the contours whose contour height value h is greater than 80% of the local analysis area height H are screened out, and the contour information is sorted from left to right according to the position.
5. The method for detecting the skewness of a scraper chain based on machine vision according to claim 1 is characterized in that: In step 10, the inclination state of the first fitting straight line line 1 on the left is taken to represent the chain plate posture.
6. The method for detecting the skewness of a scraper chain based on machine vision according to claim 1 is characterized in that: In step 11, in order to avoid misjudgment, the angle values of the vertical line angles of 10 frames of images are continuously recorded. If all the angle values exceed the set threshold of 5°, an alarm signal is triggered.
7. A device for detecting the skewness of a scraper chain based on machine vision, using the method described in any one of claims 1 to 6, characterized in that: It includes an image acquisition unit, an image processing unit, and an image display unit. The image processing unit is connected to the image acquisition unit and the image display unit. The image acquisition unit is used to execute step 1, the image processing unit is used to execute step 2 to step 12, and the image display unit is used to display the results of the analysis and processing.
