Machine Vision-Based Method and System for Detecting Tear Defects in Belt Conveyors

Abstract

This invention provides a machine vision-based method and system for detecting tear defects in belt conveyors. The method involves forming laser stripes on the surface of the belt's back side; acquiring distorted images of the laser stripes caused by belt tearing under various operating conditions and laser stripe images under normal operating conditions; preprocessing and data augmentation of the acquired laser stripe images to form laser stripe datasets for various operating conditions; dividing the laser stripe datasets and using a deep neural network model for feature extraction and fusion to train the deep neural network model; using the trained deep neural network model to classify and detect the processed images; and introducing an alarm decision factor based on the belt tear defect classification detection results to determine the corresponding judgment level of the defect and issue a corresponding warning. This invention can improve the accuracy of belt conveyor tear defect detection and provide corresponding alarm decision schemes, thereby improving production and maintenance efficiency.

Description

Technical Field

[0001] This invention belongs to the field of online monitoring technology for coal mine automation, specifically relating to a method and system for detecting tear defects in belt conveyors based on machine vision. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] In the coal mining industry, belt conveyors are one of the core pieces of equipment for material handling, responsible for transporting mined coal from underground to surface storage or processing facilities. These conveying systems typically handle large quantities of coal, thus requiring extremely high performance and reliability. However, a major problem that belt conveyors often encounter during long-term operation is belt tearing.

[0004] Belt tears can be caused by a variety of factors, such as impact from sharp objects in the material, conveyor component malfunctions (e.g., wear of guide wheels or idlers), belt misalignment leading to tearing, and core-pulling belt tears. Belt tears not only cause coal leakage and increase material loss, but can also lead to the shutdown of the entire conveyor line, severely impacting the production efficiency of the coal mine. Furthermore, excessively large tears can pose significant safety hazards and increase the system's inspection and maintenance costs.

[0005] In recent years, with the increasing application of machine vision technology in the industrial field, using machine vision technology to replace traditional belt tear detection technology can effectively improve production efficiency and product quality. However, the alarm decision mechanism of existing machine vision inspection methods is not perfect. Often, it will alarm and stop the machine as soon as a tear defect is detected, which will cause additional losses to enterprises. Summary of the Invention

[0006] To address the aforementioned problems, this invention proposes a machine vision-based method and system for detecting tear defects in belt conveyors. This invention can improve the accuracy of tear defect detection in belt conveyors and provide corresponding alarm decision schemes, thereby improving production and maintenance efficiency.

[0007] According to some embodiments, the present invention adopts the following technical solution:

[0008] A machine vision-based method for detecting tear defects in belt conveyors includes the following steps:

[0009] Laser stripes are formed on the surface of the back of the belt;

[0010] Acquire laser stripe distortion images caused by belt tearing under various operating conditions and laser stripe images under normal operating conditions;

[0011] The acquired laser stripe images are preprocessed and data augmented to form laser stripe datasets under various working conditions;

[0012] The laser stripe dataset is divided, and features are extracted and fused using a deep neural network model to train the deep neural network model.

[0013] Acquire the image to be detected, and perform preprocessing and data augmentation on the image;

[0014] The trained deep neural network model is used to classify and detect the processed images to be detected;

[0015] Based on the classification and detection results of belt tear defects, an alarm decision factor is introduced to determine the judgment level corresponding to the defect and issue a corresponding warning.

[0016] As an alternative implementation, a line laser emitter is used to emit multiple line lasers toward the back of the belt, forming laser stripes on the surface of the back of the belt.

[0017] Industrial cameras were used to capture laser stripe distortion images caused by belt tearing under various operating conditions and laser stripe images under normal operating conditions.

[0018] As an alternative implementation, the preprocessing process includes smoothing the image using Gaussian filtering, performing grayscale processing on the image, converting the color image into a grayscale image with only black and white, and performing an erosion operation on the grayscale image to reduce noise in the image.

[0019] As an alternative implementation, the data augmentation operation includes: randomly stretching, rotating, scaling, translating, cropping, and brightness changing the preprocessed image.

[0020] As an alternative implementation, the process of generating laser stripe datasets under various working conditions includes annotating the image data after data augmentation according to four features: depth, length, width, and density of belt tear defects.

[0021] As an alternative implementation method, the process of introducing alarm decision factors includes:

[0022] Based on the depth, length, width, and density characteristics of belt tear defects in the labeled data, an alarm decision function is set:

[0023] Δ=ω d d+ω l l+ω w w+ω s s;

[0024] Where Δ is the alarm decision factor, ω dω is the tear depth coefficient. l ω is the tear length coefficient. w ω is the tear width coefficient. s d is the tear density coefficient, l is the tear depth, w is the tear length, and s is the tear density.

[0025] As a further step, ω d ω l ω w ω s The specific value is determined using a regression model in machine learning.

[0026] As an alternative implementation method, the process of determining the judgment level corresponding to the defect and issuing the corresponding warning includes classifying the judgment level into normal, minor damage, moderate damage and severe damage according to the size of the alarm decision factor. When the judgment result is moderate damage, an alarm is issued and the location distribution of the damage is reported. When the judgment result is severe damage, an alarm is immediately triggered and the system is shut down.

[0027] A machine vision-based belt conveyor tear defect detection system includes:

[0028] The image acquisition module is configured to form laser stripes on the surface of the back of the belt, acquire laser stripe distortion images caused by belt tearing under various working conditions and laser stripe images under normal working conditions, and acquire the image to be detected.

[0029] The data processing module is configured to preprocess and enhance the acquired laser stripe images to form laser stripe datasets under various working conditions, and to preprocess and enhance the images to be detected.

[0030] The model training module is configured to divide the laser stripe dataset, extract and fuse its features using a deep neural network model, and train the deep neural network model.

[0031] The classification and detection module is configured to use a trained deep neural network model to classify and detect the processed image to be detected;

[0032] The early warning module is configured to introduce alarm decision factors based on the classification and detection results of belt tear defects, determine the judgment level corresponding to the defect, and issue corresponding early warnings.

[0033] An electronic device includes a memory and a processor, as well as computer instructions stored in the memory and running on the processor, wherein the computer instructions, when executed by the processor, perform the steps in the method described above.

[0034] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0035] This invention enables rapid identification of belt tear defects. While retaining the advantages of machine vision-based belt tear detection technology, this invention further improves the alarm decision-making mechanism for belt tear defect detection, avoiding additional losses caused by frequent alarm shutdowns and improving the efficiency of coal mine production.

[0036] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0037] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0038] Figure 1 This is a flowchart of a machine vision-based belt conveyor tear defect detection process according to one embodiment.

[0039] Figure 2 This is a flowchart of the preprocessing operation in the tear defect detection process of a belt conveyor based on machine vision, according to one embodiment.

[0040] Figure 3 This is a schematic diagram of an alarm decision mechanism in the process of detecting tear defects in a belt conveyor based on machine vision, according to one embodiment.

[0041] Figure 4 This is a schematic diagram showing the relative installation positions of various detection devices in one embodiment. Detailed Implementation

[0042] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0043] It should be noted that the following detailed description is illustrative and intended to provide further explanation of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0044] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0045] Where there is no conflict, the embodiments and features described in this application may be combined with each other.

[0046] Example 1

[0047] A machine vision-based method for detecting tear defects in belt conveyors is proposed. While retaining the advantages of machine vision-based belt tear detection technology, it proposes a more sophisticated alarm decision-making mechanism and a more practical method for locating belt tear defects. The specific steps are as follows: Figure 1 As shown:

[0048] S100: Uses a line laser emitter to emit a red line laser towards the back of the belt, which will form red laser stripes on the surface of the back of the belt.

[0049] S200: Uses a CCD industrial camera to capture laser stripe distortion images caused by belt tearing under various working conditions and laser stripe images under normal working conditions of belt conveyors;

[0050] Of course, in other embodiments, laser stripes can be formed in other ways, or laser images of other colors or types can be formed.

[0051] Similarly, in other embodiments, other devices may be used for image acquisition.

[0052] like Figure 4 As shown, in this embodiment, the line laser emitter and the CCD industrial camera are arranged side by side near the belt. However, in other embodiments, the laser emitting device and the image acquisition device may not be on the same side, as long as it is possible to form a laser image on the belt or acquire an image of the belt.

[0053] S300: Preprocess the acquired laser stripe images to create a laser stripe dataset of the belt conveyor under various working conditions;

[0054] In this embodiment, the image preprocessing step is as follows: Figure 2 As shown, it includes:

[0055] S310: Uses Gaussian filtering to smooth the image, reducing noise and other unwanted pixel information (such as dust generated during coal mine transportation), thereby improving image quality and clarity;

[0056] S320: Performs grayscale processing on images, using a weighted average method to convert a color image with three color channels (R, G, B) into a grayscale image with only one channel, thereby reducing the amount of data in the image, reducing memory usage, and improving processing speed.

[0057] S330: Performs erosion operation on grayscale images to further reduce noise and small objects that may exist in the image;

[0058] S340: Perform data augmentation on the image after the erosion operation to improve the model's generalization ability;

[0059] In this embodiment, the data augmentation operation includes random stretching, rotation, scaling, translation, cropping, and brightness transformation operations on the laser stripe image;

[0060] S350: The image data obtained by the above process is labeled according to the four characteristics of belt tear defects: depth, length, width, and density.

[0061] S400: Divide the dataset, use deep neural networks to perform deep learning on the image data, extract and fuse features, and use it for image recognition and classification tasks;

[0062] In this embodiment, the deep neural network model is YOLOv10. After the model predicts, it outputs predicted values ​​of the depth, length, width, and density of the belt tear, and marks the location of the tear with a rectangle.

[0063] Of course, in other embodiments, other deep neural network models, such as R-CNN, can be used.

[0064] S500: Deploy YOLOv10 on an NVIDIA GPU RTX4060 for image recognition and classification detection;

[0065] It should be noted that when performing real-time belt tear defect detection using a CCD industrial camera, the image to be detected is processed in the same way as the image used for training.

[0066] S600: The detection process continues. When a belt tear is detected, an early warning is issued based on a pre-set alarm decision mechanism, and the location of the belt tear is reported.

[0067] In this embodiment, the alarm decision mechanism is specifically as follows:

[0068] like Figure 3 As shown, the alarm decision function is set based on four characteristics of belt tear defects in the labeled data: depth, length, width, and density.

[0069] Δ=ω d d+ω l l+ω w w+ω s s;

[0070] In the formula: Δ is the alarm decision factor, ω d ω is the tear depth coefficient. l ω is the tear length coefficient. w ω is the tear width coefficient.s d is the tear density coefficient, l is the tear depth, w is the tear length, w is the tear width, and s is the tear density.

[0071] In this embodiment, ω d ω l ω w ω s The specific value is determined using a multiple regression model in machine learning, and the value will not be changed after it is determined.

[0072] For those familiar with this field, the training process of a multiple regression model is common knowledge, so the following steps will only be briefly explained:

[0073] Using the same dataset as for the deep learning task, the dataset is divided into three parts: 60% for training, 20% for cross-validation, and 20% for prediction.

[0074] Determine the multivariate regression model and loss function, update the weights using gradient descent, and determine the specific values ​​of the weights after the loss function converges and the loss is less than a set threshold.

[0075] The weight is ω. d ω l ω w ω s The specific value;

[0076] In this embodiment, the specific values ​​of d, l, w, and s are predicted by the deep neural network YOLOv10.

[0077] Based on the magnitude of Δ, the judgment level is divided into four categories: normal, minor damage, moderate damage, and severe damage. When the judgment result is moderate damage, an early warning is issued and the location distribution of the damage is reported. When the judgment result is severe damage, an alarm is immediately triggered and the system is shut down.

[0078] Example 2

[0079] A machine vision-based belt conveyor tear defect detection system includes:

[0080] The image acquisition module is configured to form laser stripes on the surface of the back of the belt, acquire laser stripe distortion images caused by belt tearing under various working conditions and laser stripe images under normal working conditions, and acquire the image to be detected.

[0081] The data processing module is configured to preprocess and enhance the acquired laser stripe images to form laser stripe datasets under various working conditions, and to preprocess and enhance the images to be detected.

[0082] The model training module is configured to divide the laser stripe dataset, extract and fuse its features using a deep neural network model, and train the deep neural network model.

[0083] The classification and detection module is configured to use a trained deep neural network model to classify and detect the processed image to be detected;

[0084] The early warning module is configured to introduce alarm decision factors based on the classification and detection results of belt tear defects, determine the judgment level corresponding to the defect, and issue corresponding early warnings.

[0085] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0086] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0087] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0088] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0089] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made by those skilled in the art without creative effort within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A machine vision-based method for detecting tear defects in belt conveyors, characterized in that, Includes the following steps: Laser stripes are formed on the surface of the back of the belt; Acquire laser stripe distortion images caused by belt tearing under various operating conditions and laser stripe images under normal operating conditions; The acquired images are preprocessed and augmented to form laser stripe datasets under various working conditions; The laser stripe dataset is divided, and features are extracted and fused using a deep neural network model to train the deep neural network model. Acquire the image to be detected, and perform preprocessing and data augmentation on the image; The trained deep neural network model is used to classify and detect the processed images to be detected; Based on the classification and detection results of belt tear defects, an alarm decision factor is introduced to determine the judgment level of the defect and issue a corresponding warning. The process of generating laser stripe datasets under various working conditions includes labeling the image data after data augmentation according to four features: depth, length, width, and density of belt tear defects. The process of introducing alarm decision factors includes: Based on the depth, length, width, and density characteristics of belt tear defects in the labeled data, an alarm decision function is set: ； in, As an alarm decision factor, This is the tear depth coefficient. This is the tear length factor. This is the tear width coefficient. This represents the tear density coefficient. For tear depth, For tear length, For tear width, The degree of tearing density; The specific value is determined using a regression model in machine learning.

2. The method for detecting tear defects in belt conveyors based on machine vision as described in claim 1, characterized in that, Multiple line laser beams are emitted towards the back of the belt using a line laser emitter, forming laser stripes on the surface of the back of the belt. Industrial cameras were used to capture laser stripe distortion images caused by belt tearing under various operating conditions and laser stripe images under normal operating conditions.

3. The method for detecting tear defects in belt conveyors based on machine vision as described in claim 1, characterized in that, The preprocessing process includes smoothing the image using Gaussian filtering, performing grayscale processing on the image to convert the color image into a grayscale image with only black and white, and performing erosion on the grayscale image to reduce noise in the image.

4. The method for detecting tear defects in belt conveyors based on machine vision as described in claim 1, characterized in that, The data augmentation operations include: randomly stretching, rotating, scaling, translating, cropping, and brightness changing the preprocessed image.

5. The method for detecting tear defects in belt conveyors based on machine vision as described in claim 1, characterized in that, The process of determining the corresponding judgment level of a defect and issuing the corresponding warning includes classifying the judgment level into normal, minor damage, moderate damage, and severe damage based on the magnitude of the alarm decision factor. When the judgment result is moderate damage, an alarm is issued and the location distribution of the damage is reported. When the judgment result is severe damage, an alarm is immediately triggered and the system is shut down.

6. A machine vision-based tear defect detection system for belt conveyors, characterized in that, include: The image acquisition module is configured to form laser stripes on the surface of the back of the belt, acquire laser stripe distortion images caused by belt tearing under various working conditions and laser stripe images under normal working conditions, and acquire the image to be detected. The data processing module is configured to preprocess and enhance the acquired images to form laser stripe datasets under various working conditions, and to preprocess and enhance the images to be detected. The model training module is configured to divide the laser stripe dataset, extract and fuse its features using a deep neural network model, and train the deep neural network model. The classification and detection module is configured to use a trained deep neural network model to classify and detect the processed image to be detected; The early warning module is configured to introduce alarm decision factors based on the belt tear defect classification and detection results, determine the judgment level corresponding to the defect, and issue corresponding early warnings. The process of generating laser stripe datasets under various working conditions includes labeling the image data after data augmentation according to four features: depth, length, width, and density of belt tear defects. The process of introducing alarm decision factors includes: Based on the depth, length, width, and density characteristics of belt tear defects in the labeled data, an alarm decision function is set: ； in, As an alarm decision factor, This is the tear depth coefficient. This is the tear length factor. This is the tear width coefficient. This represents the tear density coefficient. For tear depth, For tear length, For tear width, The degree of tearing density; The specific value is determined using a regression model in machine learning.

7. An electronic device, characterized in that, It includes a memory and a processor, as well as computer instructions stored in the memory and running on the processor, which, when executed by the processor, perform the steps of the method according to any one of claims 1-5.

Images