Mine conveying belt deviation detection method and device based on multi-level vision algorithm fusion, electronic equipment and storage medium

Abstract

The invention relates to a mine conveying belt deviation detection method based on multi-level vision algorithm fusion. The method comprises the steps of obtaining an image to be processed, obtaining a processed image, calculating the deviation amount of a conveying belt and judging the abnormal condition of the conveying belt. The invention further relates to a mine conveying belt deviation detection device based on multi-level vision algorithm fusion. The mine conveying belt deviation detection device comprises a positioning module, a processing module, a calculation module and a judgment module. The invention further relates to electronic equipment and a storage medium. By means of the mine conveying belt deviation detection method and device based on multi-level vision algorithm fusion, the electronic equipment and the storage medium, the problems that in the prior art, a mechanical contact type detection device is mainly adopted in the conveying belt deviation detection technology, bidirectional abrasion between a detection component and a conveying belt body is detected, and the detection precision is poor can be solved. The service life of breakfast is shortened, the maintenance cost is increased, the accuracy degree is low, and the detection stability is poor.

Description

Technical Field

[0001] This invention relates to the fields of machine vision and industrial automation technology, specifically to a method, device, electronic equipment, and storage medium for detecting belt misalignment in mines using a multi-level vision algorithm fusion approach. Background Technology

[0002] Conveyor belts, as core material transport equipment in industries such as mining, metallurgy, cement, and coal, commonly experience belt misalignment during long-term operation. Misalignment not only accelerates wear on the conveyor belt edges but can also damage equipment components and even cause system shutdowns, resulting in production interruptions and economic losses. To ensure the stable operation of the conveyor belt system, real-time and accurate detection of belt misalignment is essential.

[0003] In existing technologies, conveyor belt misalignment detection mainly employs mechanical contact detection devices, such as mechanical limit switches and lever-type misalignment switches. While these methods are simple in structure and low in cost, the long-term contact between the detection element and the conveyor belt causes bidirectional wear on both the detection component and the conveyor belt itself, shortening its service life and increasing maintenance costs. Furthermore, mechanical detection devices primarily output switching signals with an accuracy typically around ±10mm, which is insufficient to meet precise quantification requirements and cannot provide high-quality data for subsequent intelligent analysis. In dusty and complex industrial environments, mechanical detection elements are prone to failure due to dust accumulation, corrosion, or foreign object jamming, reducing detection stability.

[0004] Therefore, existing technologies have shortcomings and need to be improved and developed. Summary of the Invention

[0005] This invention provides a method, apparatus, electronic device, and storage medium for detecting conveyor belt misalignment in mining operations using a multi-level visual algorithm fusion approach. This addresses the problems of existing conveyor belt misalignment detection technologies, which primarily employ mechanical contact detection devices such as mechanical limit switches and lever-type misalignment switches. While these methods are simple in structure and low in cost, the long-term contact between the detection element and the conveyor belt causes bidirectional wear, shortening service life and increasing maintenance costs. Furthermore, mechanical detection devices typically output switching signals with an accuracy of only ±10mm, which is insufficient for precise quantification and fails to provide high-quality data for subsequent intelligent analysis. In dusty and complex industrial environments, mechanical detection elements are prone to failure due to dust accumulation, corrosion, or foreign object obstruction, reducing detection stability.

[0006] In a first aspect, this invention provides a method for detecting conveyor belt misalignment in mines using a multi-level visual algorithm fusion approach, comprising:

[0007] Locate the conveyor belt area and acquire images of the conveyor belt's position and shape in real time;

[0008] The image to be processed is segmented and the background is denoised to obtain a processed image with conveyor belt edges;

[0009] The difference between the distance between the straight line at the edge of the conveyor belt obtained from the processed image using Hough transform and the theoretical straight line at the edge of the conveyor belt is calculated to obtain the calculated deviation of the conveyor belt.

[0010] Based on the calculated belt misalignment, an abnormal situation is determined. If the calculated belt misalignment exceeds the theoretical misalignment under normal operating conditions, an alarm system is triggered and an alarm message is sent to the monitoring center. Otherwise, the above process is repeated to update the calculated belt misalignment.

[0011] Furthermore, the positioning of the conveyor belt area involves real-time acquisition of an image to be processed, showing the position and shape of the conveyor belt, including:

[0012] Acquire images of the conveyor belt;

[0013] The conveyor belt image is imported into a deep learning object detection model to identify the conveyor belt image through a detection box;

[0014] The feature information of the conveyor belt in the image is extracted using a multi-layer convolutional neural network to obtain an image to be processed that has the position and shape of the conveyor belt.

[0015] The conveyor belt area is updated in real time through a dynamic ROI adjustment mechanism to locate the position and shape of the conveyor belt when it moves or is blocked, and to update the image to be processed.

[0016] Furthermore, importing the conveyor belt image into the deep learning object detection model to identify the conveyor belt image through detection boxes includes: optimizing the detection boxes based on prior geometric knowledge of the conveyor belt to remove regions that do not conform to the shape of the conveyor belt.

[0017] Furthermore, the step of performing image segmentation and background denoising on the image to be processed to obtain the processed image includes:

[0018] Based on the image to be processed, image information of the conveyor belt is obtained; the image information includes the ambient light level of the conveyor belt, the image contrast of the conveyor belt, and the noise level of the conveyor belt image.

[0019] The cropping and limiting parameters and the block grid size parameters in the adaptive OTSU threshold segmentation algorithm are dynamically adjusted according to the illumination magnitude to remove background noise after temporal filtering of the conveyor belt image contrast and noise level, thereby obtaining a segmented processed image with conveyor belt edges.

[0020] Furthermore, the calculation utilizes the difference between the distance between the straight line at the edge of the conveyor belt obtained from the processed image using the Hough transform and the theoretical straight line at the edge of the conveyor belt to obtain the calculated conveyor belt deviation, including:

[0021] The straight line of the conveyor belt edge is obtained by introducing an angle constraint on the tilt angle of the straight line of the conveyor belt edge obtained from the processed image using Hough transform, and by introducing weight optimization on each point on the straight line of the conveyor belt edge obtained from the processed image using Hough transform.

[0022] By performing on-site calibration, the distance between the midpoints of the theoretical straight lines along the conveyor belt edge and the midpoints of the straight lines along the conveyor belt edge are calculated in real time using the Hough transform. This difference in pixel distances is obtained from the processed image, and the conveyor belt deviation is calculated.

[0023] Furthermore, the angle constraint introduced for the tilt angle of the straight line of the conveyor belt edge obtained from the processed image using the Hough transform is: the angle is increased at the upper and lower limits of the tilt angle, and the range of the increased angle is [5°, 15°]; the weight optimization introduced for each point on the straight line of the conveyor belt edge obtained from the processed image using the Hough transform is: each point on the straight line of the conveyor belt edge obtained from the processed image using the Hough transform is given a weight w, where the stronger the gradient of the straight line segment containing each point on the straight line of the conveyor belt edge obtained from the processed image using the Hough transform, the closer the gradient direction is to the theoretical straight line of the conveyor belt edge, and the longer the continuous segment, the larger w is, where w is a number greater than 0.

[0024] Furthermore, based on the calculated conveyor belt misalignment, an abnormal condition of the conveyor belt is determined. If the calculated misalignment exceeds the theoretical misalignment under permissible operating conditions, an alarm system is triggered and an alarm message is sent to the monitoring center; otherwise, the above process is repeated to update the calculated conveyor belt misalignment, including:

[0025] When the calculated deviation of the conveyor belt is greater than the theoretical deviation of the conveyor belt under the allowable working conditions, the calculated deviation of the conveyor belt and the conveyor belt information at this time are recorded; the conveyor belt information includes at least the conveyor belt position, the conveyor belt deviation time, the conveyor belt operating environment, speed, and load;

[0026] The calculated conveyor belt misalignment and conveyor belt information are sent to the operator's receiving terminal so that the operator can adjust and maintain the conveyor belt based on the conveyor belt information.

[0027] A second aspect of the present invention provides a mine conveyor belt misalignment detection device based on multi-level visual algorithm fusion, comprising:

[0028] The positioning module is used to locate the conveyor belt area and acquire images of the conveyor belt's position and shape in real time.

[0029] The processing module is used to perform image segmentation and background denoising on the image to be processed to obtain a processed image with conveyor belt edges;

[0030] The calculation module is used to calculate the difference between the distance between the straight line of the conveyor belt edge obtained from the processed image using Hough transform and the theoretical straight line of the conveyor belt edge, so as to obtain the calculated deviation of the conveyor belt.

[0031] The judgment module is used to determine the abnormal condition of the conveyor belt based on the calculated deviation amount. If the calculated deviation amount is greater than the theoretical deviation amount allowed under the working conditions, the alarm system is triggered and an alarm message is sent to the monitoring center; otherwise, the above process is repeated to update the calculated deviation amount.

[0032] In a third aspect, the present invention provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the multi-level visual algorithm fusion method for detecting belt misalignment in mines.

[0033] In a fourth aspect, the present invention provides a computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor, implements the multi-level visual algorithm fusion method for detecting belt misalignment in mines.

[0034] Beneficial effects:

[0035] As can be seen from the above technical solutions, this invention provides a multi-level visual algorithm fusion method for detecting conveyor belt deviation in mines. Through deep learning target detection, dynamic ROI adjustment, adaptive image segmentation, background denoising, and Hough transform linear fitting with angle constraints and weight optimization, it achieves high-precision, real-time, and stable detection of conveyor belt deviation. The overall solution has the following beneficial effects:

[0036] 1. Significantly Improved Detection Accuracy: By extracting conveyor belt features through a multi-layer convolutional neural network and combining it with a dynamic ROI mechanism to update the detection area in real time, the localization process can continuously focus on key areas of the conveyor belt, reducing background interference and improving the effectiveness of detection data. In the edge detection stage, adaptive OTSU threshold segmentation combined with temporal filtering not only adapts to changes in light intensity but also suppresses the impact of instantaneous noise and environmental fluctuations on the detection results, thereby obtaining conveyor belt edge information with clear contours and high continuity. Angle constraints are introduced into the Hough transform fitting to ensure that only straight edges conforming to the conveyor belt direction are extracted. Simultaneously, by assigning weights to pixels on the straight line, the contribution of key points to the straight line fitting is strengthened, significantly improving the fit between the fitted straight line and the actual conveyor belt edge, thus quantifying the deviation at the pixel level.

[0037] 2. Enhanced Robustness and Environmental Adaptability: The detection process fully considers adverse factors such as dust, lighting variations, and occlusion in mining and metallurgical environments. A dynamic ROI adjustment mechanism ensures rapid re-locking of the conveyor belt area even when it is partially obscured or its position changes. The adaptive segmentation and background denoising module automatically adjusts processing parameters based on real-time lighting and noise levels, ensuring stable performance in various environments, including daytime, nighttime, strong light, and weak light. Multi-level algorithm fusion avoids the problem of single algorithms failing in specific scenarios, improving the system's operational stability under different environmental conditions.

[0038] 3. Quantifiable Detection Results and Direct Guidance for Operation and Maintenance: The offset calculation module, combined with on-site calibration, achieves a precise mapping from pixel distance to actual physical distance, making the detection results quantifiable and providing accurate numerical data for equipment maintenance and operational adjustments. The anomaly handling module not only triggers alarms but also simultaneously records multi-dimensional information such as conveyor belt running position, offset occurrence time, operating environment, speed, and load, providing comprehensive decision-making references for operation and maintenance personnel.

[0039] 4. Reduced Equipment Maintenance and Operation Risks: Non-contact vision inspection avoids direct friction between mechanical sensors and the conveyor belt, reducing wear on the conveyor belt and sensor components, extending service life, and lowering maintenance frequency and replacement costs. The system can detect and warn of belt misalignment in its early stages, reducing the risk of accidents such as conveyor belt tearing and idler damage caused by severe misalignment, thereby reducing unplanned downtime and improving the continuous operation capability of the production line.

[0040] 5. Scalability and Intelligent Potential: The detection framework of this invention can seamlessly integrate with the centralized monitoring system of mine conveyor belts, enabling centralized management and operational data analysis of multiple conveyor belts. The detection data can be further used as training samples to optimize deep learning models, achieving adaptive improvement in detection capabilities, thereby continuously enhancing detection accuracy and environmental adaptability during long-term operation.

[0041] In summary, the overall technical solution of this invention has significant improvements over existing technologies in terms of accuracy, robustness, environmental adaptability, quantifiable analysis, and equipment protection. It can meet the high-standard requirements for conveyor belt misalignment detection in high-risk industrial scenarios such as mines, and has broad application prospects in ensuring production safety, reducing maintenance costs, and achieving intelligent management.

[0042] It should be understood that all combinations of the foregoing concepts and the additional concepts described in more detail below can be considered part of the inventive subject matter of this disclosure, provided that such concepts do not contradict each other.

[0043] The foregoing and other aspects, embodiments, and features of the teachings of the present invention will be more fully understood from the following description in conjunction with the accompanying drawings. Other additional aspects of the invention, such as features and / or beneficial effects of exemplary embodiments, will become apparent from the following description or may be learned through practice of specific embodiments according to the teachings of the present invention. Attached Figure Description

[0044] The accompanying drawings are not drawn to scale. In the drawings, each identical or nearly identical component shown in the various figures may be denoted by the same reference numeral. For clarity, not every component is labeled in each figure. Embodiments of various aspects of the invention will now be described by way of example and with reference to the accompanying drawings, wherein:

[0045] Figure 1 This is a flowchart illustrating the overall process of a multi-level visual algorithm fusion method for detecting conveyor belt misalignment in mines, as described in this application.

[0046] Figure 2 This is a flowchart of step S102 of a multi-level visual algorithm fusion method for detecting belt misalignment in a mine, as described in an embodiment of this application.

[0047] Figure 3 This is a flowchart of step S104 of a multi-level visual algorithm fusion method for detecting belt misalignment in a mine, as described in an embodiment of this application.

[0048] Figure 4 This is a flowchart of step S106 of a multi-level visual algorithm fusion method for detecting belt misalignment in a mine, as described in an embodiment of this application.

[0049] Figure 5 This is a flowchart of step S108 of a multi-level visual algorithm fusion method for detecting belt misalignment in a mine, as described in an embodiment of this application.

[0050] Figure 6 This is a target detection image of a mine conveyor belt deviation detection method based on multi-level visual algorithm fusion in an embodiment of this application.

[0051] Figure 7 This is an edge extraction image of a mine conveyor belt deviation detection method based on multi-level visual algorithm fusion in an embodiment of this application.

[0052] Figure 8 This is a linear fitting diagram of a mine conveyor belt deviation detection method based on multi-level visual algorithm fusion in an embodiment of this application. Detailed Implementation

[0053] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the described embodiments of the present invention without creative effort are within the scope of protection of the present invention. Unless otherwise defined, the technical or scientific terms used herein should have the ordinary meaning understood by those skilled in the art to which this invention pertains.

[0054] The terms "first," "second," and similar words used in the specification and claims of this patent application do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Similarly, unless the context clearly indicates otherwise, the singular forms of "an," "a," or "the," etc., do not indicate a quantity limitation, but rather indicate the presence of at least one. Terms such as "comprising" or "including" indicate that the element or object preceding "comprising" encompasses the features, integrals, steps, operations, elements, and / or components listed following "comprising" or "including," and do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components, and / or sets thereof. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; these relative positional relationships may change accordingly when the absolute position of the described object changes.

[0055] In existing technologies, conveyor belt misalignment detection mainly employs mechanical contact detection devices, such as mechanical limit switches and lever-type misalignment switches. While these methods are simple in structure and low in cost, the long-term contact between the detection element and the conveyor belt causes bidirectional wear on both the detection component and the conveyor belt itself, shortening its service life and increasing maintenance costs. Furthermore, mechanical detection devices primarily output switching signals with an accuracy typically around ±10mm, which is insufficient to meet precise quantification requirements and cannot provide high-quality data for subsequent intelligent analysis. In dusty and complex industrial environments, mechanical detection elements are prone to failure due to dust accumulation, corrosion, or foreign object jamming, reducing detection stability.

[0056] In view of this, refer to Figure 1 This invention provides a method for detecting conveyor belt misalignment in mines using a multi-level visual algorithm fusion approach, comprising:

[0057] Step S102: Locate the conveyor belt area and acquire an image of the conveyor belt's position and shape in real time.

[0058] Step S104: Perform image segmentation and background denoising on the image to be processed to obtain a processed image with conveyor belt edges.

[0059] Step S106: Calculate the difference between the distance between the straight line of the conveyor belt edge obtained from the processed image using Hough transform and the theoretical straight line of the conveyor belt edge, and obtain the calculated deviation of the conveyor belt.

[0060] Step S108: Based on the calculated belt misalignment, determine the abnormality of the conveyor belt. If the calculated belt misalignment is greater than the theoretical belt misalignment under the allowable operating conditions, trigger the alarm system and send an alarm message to the monitoring center; otherwise, repeat the above process to update the calculated belt misalignment.

[0061] In some embodiments, reference is made to Figure 2 Locate the conveyor belt area and acquire images of the conveyor belt's position and shape in real time, including:

[0062] Step S1021: Acquire images of the conveyor belt.

[0063] Step S1022: Import the conveyor belt image into the deep learning object detection model to identify the conveyor belt image through the detection box.

[0064] Step S1023: Use a multi-layer convolutional neural network to extract the feature information of the conveyor belt in the conveyor belt image to obtain an image to be processed that has the position and shape of the conveyor belt.

[0065] Reference Figure 6The document presents the target detection results. Steps S1021-S1023 employ a deep learning target detection algorithm to quickly locate the conveyor belt, effectively solving the problem of conveyor belt position drift in complex scenes, such as small displacements or slight occlusions during high-speed operation. In some embodiments, the YOLOv5 target detection algorithm is used for rapid localization of the conveyor belt area. YOLOv5 is a deep learning algorithm based on convolutional neural networks (CNN), possessing high accuracy and fast processing speed. By inputting the acquired conveyor belt image into the YOLOv5 model, the model can quickly identify the conveyor belt area in the image. YOLOv5 uses a single neural network for end-to-end training and inference, enabling it to capture the morphological features of the conveyor belt at multiple scales, effectively locating the conveyor belt's position regardless of whether the image is clear or the lighting conditions vary significantly. Through this step, the system can provide accurate regions for subsequent image segmentation and deviation calculation.

[0066] Step S1024: Update the conveyor belt area in real time through a dynamic ROI adjustment mechanism to locate the position and shape of the conveyor belt when it moves or is blocked, and update the image to be processed.

[0067] The dynamic ROI adjustment mechanism based on target detection can focus computational resources by automatically adjusting the detection frame. Conveyor belt width, speed, and environmental conditions frequently change, and traditional methods may fail to effectively handle bandwidth variations or occlusion, leading to ROI misalignment or wasted computational resources. However, this invention, through dynamic adjustment of the detection frame, can automatically identify the areas requiring the most attention. Regardless of changes in the conveyor belt's position, the detection area remains focused on critical parts. Especially in the unique environment of mines, where conveyor belts are often constrained by space, dust accumulation, and temperature variations, dynamic ROI adjustment not only improves computational efficiency but also ensures detection accuracy, maintaining high-efficiency and stable detection performance even under constrained or unstable working conditions.

[0068] In some embodiments, importing a conveyor belt image into a deep learning object detection model to identify the conveyor belt image via a detection box includes: optimizing the detection box based on prior geometric knowledge of the conveyor belt to remove regions that do not conform to the shape of the conveyor belt.

[0069] In some embodiments, reference is made to Figure 3 The image to be processed is segmented and the background is denoised to obtain the processed image, including:

[0070] Step S1041: Obtain image information of the conveyor belt based on the image to be processed; the image information includes the illumination level of the environment where the conveyor belt is located, the image contrast of the conveyor belt, and the noise level of the conveyor belt image.

[0071] Step S1042: Dynamically adjust the cropping and limiting parameters and the block grid size parameters in the adaptive OTSU threshold segmentation algorithm according to the illumination size value, so as to remove background noise after temporal filtering of the conveyor belt image contrast and the conveyor belt image noise level, and obtain the segmented processed image with the conveyor belt edge.

[0072] Reference Figure 7 This image demonstrates the segmented image with conveyor belt edges after edge extraction. After locating the conveyor belt region, the next steps are image segmentation and background denoising. To handle complex backgrounds in mining environments (such as dust, lighting variations, and equipment vibration), this invention employs an improved adaptive OTSU threshold segmentation algorithm. This algorithm allows the system to dynamically adjust the segmentation threshold based on different lighting conditions, image contrast, and noise levels, thereby accurately segmenting the conveyor belt edges and removing background noise. This enables the system to accurately extract the conveyor belt outline even under strong lighting and dust conditions in mining environments, ensuring that subsequent analysis and calculations can be performed within the accurate image area.

[0073] Steps S1041 and S1042 employ an adaptive OTSU threshold segmentation algorithm to accurately separate the conveyor belt from the background, eliminating interference from lighting and dust. In complex environments such as mines and metallurgical plants, conveyor belts often encounter harsh conditions such as strong lighting changes, dust accumulation, and high temperatures. These factors significantly affect the stability and accuracy of traditional vision algorithms. To address this issue, this embodiment of the invention introduces an adaptive OTSU threshold segmentation algorithm, which automatically adjusts parameters according to different lighting conditions, ensuring consistently high segmentation accuracy. Furthermore, the application of temporal filtering technology effectively eliminates environmental vibrations and occasional image noise, thereby overcoming problems such as vibration interference, particulate contamination, and dynamic changes in mining environments, greatly improving the robustness and stability of the system. Even in extremely complex mining environments, stable detection is still possible, avoiding common problems in traditional technologies such as missed detections, false alarms, and instability.

[0074] In some embodiments, reference is made to Figure 4 The difference between the distance between the straight line at the edge of the conveyor belt obtained from the processed image using Hough transform and the theoretical straight line at the edge of the conveyor belt is calculated to obtain the calculated conveyor belt deviation, including:

[0075] Step S1061: Obtain the straight line of the conveyor belt edge by introducing an angle constraint on the tilt angle of the straight line of the conveyor belt edge obtained from the processed image using Hough transform, and by introducing weight optimization on each point on the straight line of the conveyor belt edge obtained from the processed image using Hough transform.

[0076] Step S1062: Through on-site calibration, the distance between the midpoints of the theoretical straight lines of the conveyor belt edge and the pixel distance difference between the midpoints of the straight lines of the conveyor belt edge obtained from the processed image using Hough transform is used to obtain the calculated deviation of the conveyor belt.

[0077] Reference Figure 8 This image demonstrates a straight line with conveyor belt edges after straight line fitting. In the segmented and denoised image, the Hough transform algorithm was used to accurately extract the conveyor belt edges. The Hough transform efficiently detects conveyor belt edges and calculates the offset of the conveyor belt relative to a reference straight line by extracting straight line features from the image. Figure 8 In the diagram, the green straight line represents the conveyor belt edge extracted by the Hough transform. By calculating the distance from the conveyor belt edge to the reference line, the system can determine the conveyor belt deviation and thus assess its operating status. Steps S1061-S1062 use the Hough transform to perform high-precision fitting of the edge, ensuring the accuracy of the deviation position. This embodiment of the invention employs an improved Hough transform algorithm for conveyor belt edge extraction. Through angle constraints and weight optimization, the fitted straight line can accurately extract the conveyor belt edge, avoiding the edge blurring or breakage problems that occur in conventional methods. Especially in the dusty and vibrating environment of mines, traditional edge extraction algorithms are often affected by noise, leading to unclear edges or incorrect identification. By introducing weight optimization technology, the system can automatically identify and correct errors, improving the accuracy of edge recognition. Furthermore, this invention, combined with on-site calibration parameters, calculates the pixel distance difference between the midpoints of the theoretically located straight line of the conveyor belt edge and the midpoints of the straight lines of the conveyor belt edge obtained from the processed image using the Hough transform, further realizing the accurate quantification of the conveyor belt offset. This precise edge fitting and error correction method ensures the accuracy of conveyor belt misalignment detection, meeting the high-precision monitoring needs of industrial sites. It provides stable and reliable detection results, especially when dealing with conveyor belts operating at high speeds and under complex conditions.

[0078] Steps S102-S106 introduce a multi-level visual fusion algorithm, including deep learning object detection, dynamic ROI adaptive adjustment, OTSU adaptive threshold segmentation, and improved Hough transform, to accurately detect conveyor belt misalignment. Especially in complex mining environments, the system can accurately identify the conveyor belt edge and calculate the misalignment amount under conditions of high noise and interference, ensuring a detection accuracy of ±2mm, providing real-time and reliable detection results, and timely warning of potential faults. Traditional misalignment detection relies on direct contact between mechanical sensors and the conveyor belt, which not only leads to sensor wear and frequent replacement but may also affect the service life of the conveyor belt. In contrast, this invention uses a non-contact visual detection method, completely avoiding physical contact between the equipment and the conveyor belt, thus avoiding mechanical wear problems. Through precise image processing and deep learning algorithms, the system can operate continuously and stably, thereby reducing equipment failures and maintenance frequency, and lowering the operation and maintenance costs of mining enterprises. At the same time, the system has high scalability and can be customized according to the specific needs of different mines, providing more flexible application solutions for different production environments.

[0079] In some embodiments, the angle constraint introduced for the tilt angle of the straight line of the conveyor belt edge obtained from the processed image using the Hough transform is: increasing the angle at the upper and lower limits of the tilt angle, with the increased angle ranging from [5°, 15°]; the weight optimization introduced for each point on the straight line of the conveyor belt edge obtained from the processed image using the Hough transform is: adding a weight w to each point on the straight line of the conveyor belt edge obtained from the processed image using the Hough transform, wherein the stronger the gradient of the straight line segment containing each point on the straight line of the conveyor belt edge obtained from the processed image using the Hough transform, the closer the gradient direction is to the theoretical straight line of the conveyor belt edge, and the longer the continuous segment, the larger w is, where w is a number greater than 0.

[0080] In some embodiments, reference is made to Figure 5 Based on the calculated conveyor belt misalignment, an abnormal situation is identified. If the calculated misalignment exceeds the theoretical misalignment under normal operating conditions, an alarm system is triggered and an alarm message is sent to the monitoring center. Otherwise, the above process is repeated to update the calculated conveyor belt misalignment, including:

[0081] Step S1081: When the calculated deviation of the conveyor belt is greater than the theoretical deviation of the conveyor belt under the allowable working conditions, record the calculated deviation of the conveyor belt and the conveyor belt information at this time; the conveyor belt information includes at least the conveyor belt position, the conveyor belt deviation time, the operating environment of the conveyor belt, the speed, and the load.

[0082] Step S1082: Send the calculated conveyor belt deviation and conveyor belt information to the operator's receiving terminal so that the operator can adjust and maintain the conveyor belt according to the information.

[0083] This invention, through real-time calculation of belt misalignment and dynamic threshold determination, can respond rapidly when abnormal belt misalignment occurs. The system's alarm device immediately triggers an alarm when misalignment exceeds a preset threshold, uploading relevant information (such as time, location, and images) to the operator's receiving terminal or monitoring center. This assists operators in timely adjustments and maintenance, providing timely data support for subsequent fault handling and production scheduling. The system can also dynamically adjust the alarm threshold based on factors such as the conveyor belt's operating environment, speed, and load, ensuring detection accuracy and response speed under different working conditions. Field verification shows that the system's response time is controlled within 800ms, meeting the stringent real-time requirements of industrial production.

[0084] Another embodiment of the present invention provides a multi-level visual algorithm fusion-based mine conveyor belt misalignment detection device, comprising:

[0085] The positioning module is used to locate the conveyor belt area and acquire images of the conveyor belt's position and shape in real time.

[0086] The processing module is used to perform image segmentation and background denoising on the image to be processed, so as to obtain a processed image with conveyor belt edges.

[0087] The calculation module is used to calculate the difference between the distance between the straight line of the conveyor belt edge obtained from the processed image using Hough transform and the theoretical straight line of the conveyor belt edge, thereby obtaining the calculated conveyor belt deviation.

[0088] The judgment module is used to determine the abnormal condition of the conveyor belt based on the calculated belt deviation. If the calculated belt deviation is greater than the theoretical belt deviation under the allowable working conditions, the alarm system is triggered and an alarm message is sent to the monitoring center; otherwise, the above process is repeated to update the calculated belt deviation.

[0089] Another embodiment of the present invention provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is executed by the processor, it implements a method for detecting belt misalignment in mines by fusing multi-level visual algorithms.

[0090] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field-Programmable Gate Arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the gateway's data processing device, connecting various parts of the gateway's data processing device through various interfaces and lines.

[0091] Another embodiment of the present invention provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements a method for detecting belt misalignment in mines by fusing multi-level visual algorithms.

[0092] The memory may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created by the processor, etc. Furthermore, the memory is preferably, but not limited to, high-speed random access memory (RAM). For example, it may also be non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, the memory may also optionally include memory remotely located relative to the processor, which can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks (LANs), mobile communication networks, and combinations thereof.

[0093] Those skilled in the art will understand that implementing all or part of the processes in the above embodiments can be accomplished by a computer program instructing related hardware, and can be stored in a computer-readable storage medium. When executed, the program can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk drive (HDD), or solid-state drive (SSD), etc.; the storage medium can also include combinations of the above types of memory.

[0094] In summary, this invention provides a multi-level visual algorithm fusion method for detecting conveyor belt deviation in mines. Through deep learning target detection, dynamic ROI adjustment, adaptive image segmentation, background denoising, and Hough transform line fitting with angle constraints and weight optimization, it achieves high-precision, real-time, and stable detection of conveyor belt deviation. The overall solution has the following advantages: Significantly improved detection accuracy: By extracting conveyor belt features through a multi-layer convolutional neural network and combining it with a dynamic ROI mechanism to update the detection area in real time, the localization process can continuously focus on key areas of the conveyor belt, reducing background interference and improving the effectiveness of the detection data. In the edge detection stage, the illumination-adaptive OTSU threshold segmentation combined with temporal filtering not only adapts to changes in illumination intensity but also suppresses the influence of instantaneous noise and environmental fluctuations on the detection results, thereby obtaining conveyor belt edge information with clear contours and high continuity. Introducing angle constraints in the Hough transform fitting ensures that only straight edges conforming to the conveyor belt direction are extracted. Simultaneously, by assigning weights to pixels on the straight line, the contribution of key points to the straight line fitting is strengthened, significantly improving the fit between the fitted straight line and the actual conveyor belt edge, thus quantifying the deviation amount at the pixel level. Enhanced Robustness and Environmental Adaptability: The detection process fully considers adverse factors such as dust, lighting variations, and shading in mining and metallurgical environments. A dynamic ROI adjustment mechanism ensures rapid re-locking of the conveyor belt area even when it is partially obscured or its position changes. The adaptive segmentation and background denoising module automatically adjusts processing parameters based on real-time lighting and noise levels, ensuring stable performance in various environments, including daytime, nighttime, strong light, and weak light. Multi-level algorithm fusion avoids the failure of single algorithms in specific scenarios, improving system stability under different environmental conditions. Quantifiable Detection Results and Direct Maintenance Guidance: The offset calculation module, combined with on-site calibration, achieves precise mapping from pixel distance to actual physical distance, making the detection results quantifiable and providing accurate numerical data for equipment maintenance and operational adjustments. The anomaly handling module not only triggers alarms but also simultaneously records multi-dimensional information such as conveyor belt position, offset occurrence time, operating environment, speed, and load, providing comprehensive decision-making references for maintenance personnel. Reduced Equipment Maintenance and Operation Risks: Non-contact visual inspection avoids direct friction between mechanical sensors and the conveyor belt, reducing wear on the conveyor belt and sensor components, extending service life, and lowering maintenance frequency and replacement costs. The system can detect and warn of belt misalignment in its early stages, reducing the risk of accidents such as conveyor belt tearing and idler damage caused by severe misalignment, thereby reducing unplanned downtime and improving the continuous operation capability of the production line. Scalability and Intelligent Potential: The detection framework of this invention can seamlessly interface with the centralized monitoring system of mining conveyor belts, enabling centralized management and operational data analysis of multiple conveyor belts.The detection data can be further used as training samples to optimize the deep learning model, thereby improving the adaptive detection capability and continuously enhancing detection accuracy and environmental adaptability in long-term operation.

[0095] While the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the invention. Those skilled in the art can make various modifications and refinements without departing from the spirit and scope of the invention. Therefore, the scope of protection of the present invention shall be determined by the claims.

Claims

1. A method for detecting conveyor belt misalignment in mines using multi-level visual algorithm fusion, characterized in that, include: Locate the conveyor belt area and acquire images of the conveyor belt's position and shape in real time; The image to be processed is segmented and the background is denoised to obtain a processed image with conveyor belt edges; The difference between the distance between the straight line at the edge of the conveyor belt obtained from the processed image using Hough transform and the theoretical straight line at the edge of the conveyor belt is calculated to obtain the calculated deviation of the conveyor belt. Based on the calculated belt misalignment, an abnormal situation is determined. If the calculated belt misalignment exceeds the theoretical misalignment under permissible operating conditions, an alarm system is triggered and an alarm message is sent to the monitoring center; otherwise, the above process is repeated to update the calculated belt misalignment.

2. The method for detecting conveyor belt misalignment in mines using a multi-level visual algorithm fusion as described in claim 1, characterized in that, The positioning of the conveyor belt area involves acquiring, in real-time, images of the conveyor belt's position and shape, including: Acquire images of the conveyor belt; The conveyor belt image is imported into a deep learning object detection model to identify the conveyor belt image through a detection box; The feature information of the conveyor belt in the image is extracted using a multi-layer convolutional neural network to obtain an image to be processed that has the position and shape of the conveyor belt. The conveyor belt area is updated in real time through a dynamic ROI adjustment mechanism to locate the position and shape of the conveyor belt when it moves or is blocked, and to update the image to be processed.

3. The method for detecting conveyor belt misalignment in mines using a multi-level visual algorithm fusion as described in claim 2, characterized in that, The step of importing the conveyor belt image into a deep learning object detection model to identify the conveyor belt image through detection boxes includes: optimizing the detection boxes based on prior geometric knowledge of the conveyor belt to remove regions that do not conform to the shape of the conveyor belt.

4. The method for detecting conveyor belt misalignment in mines using a multi-level visual algorithm fusion as described in claim 1, characterized in that, The step of performing image segmentation and background denoising on the image to be processed to obtain the processed image includes: Based on the image to be processed, image information of the conveyor belt is obtained; the image information includes the ambient light level of the conveyor belt, the image contrast of the conveyor belt, and the noise level of the conveyor belt image. The cropping and limiting parameters and the block grid size parameters in the adaptive OTSU threshold segmentation algorithm are dynamically adjusted according to the illumination magnitude to remove background noise after temporal filtering of the conveyor belt image contrast and noise level, thereby obtaining a segmented processed image with conveyor belt edges.

5. The method for detecting conveyor belt misalignment in mines using a multi-level visual algorithm fusion as described in claim 1, characterized in that, The calculation utilizes the difference between the distance between the straight line at the edge of the conveyor belt obtained from the processed image using the Hough transform and the theoretical straight line at the edge of the conveyor belt to obtain the calculated conveyor belt deviation, including: The straight line of the conveyor belt edge is obtained by introducing an angle constraint on the tilt angle of the straight line of the conveyor belt edge obtained from the processed image using Hough transform, and by introducing weight optimization on each point on the straight line of the conveyor belt edge obtained from the processed image using Hough transform. By performing on-site calibration, the distance between the midpoints of the theoretical straight lines along the conveyor belt edge and the midpoints of the straight lines along the conveyor belt edge are calculated in real time using the Hough transform. This difference in pixel distances is obtained from the processed image, and the conveyor belt deviation is calculated.

6. The method for detecting conveyor belt misalignment in mines using a multi-level visual algorithm fusion as described in claim 5, characterized in that, The angle constraint introduced for the tilt angle of the straight line of the conveyor belt edge obtained from the processed image using the Hough transform is: the angle is increased at the upper and lower limits of the tilt angle, and the range of the increased angle is [5°, 15°]. The weight optimization introduced for each point on the straight line of the conveyor belt edge obtained from the processed image using the Hough transform is: each point on the straight line of the conveyor belt edge obtained from the processed image using the Hough transform is given a weight w, where the stronger the gradient of the straight line segment containing each point on the straight line segment of the conveyor belt edge obtained from the processed image using the Hough transform, the closer the gradient direction is to the theoretical straight line of the conveyor belt edge, and the longer the continuous segment, the larger w is, where w is a number greater than 0.

7. The method for detecting conveyor belt misalignment in mines using a multi-level visual algorithm fusion as described in claim 1, characterized in that, The system calculates the belt deviation based on the belt deviation and determines the abnormality of the belt. If the calculated belt deviation exceeds the theoretical deviation under the allowable operating conditions, the system triggers an alarm and sends an alarm message to the monitoring center. Otherwise, repeat the above process to update the calculated conveyor belt misalignment, including: When the calculated deviation of the conveyor belt is greater than the theoretical deviation of the conveyor belt under the allowable working conditions, the calculated deviation of the conveyor belt and the conveyor belt information at this time are recorded; the conveyor belt information includes at least the conveyor belt position, the conveyor belt deviation time, the conveyor belt operating environment, speed, and load; The calculated conveyor belt misalignment and conveyor belt information are sent to the operator's receiving terminal so that the operator can adjust and maintain the conveyor belt based on the conveyor belt information.

8. A multi-level visual algorithm fusion-based conveyor belt misalignment detection device for mines, characterized in that, include: The positioning module is used to locate the conveyor belt area and acquire images of the conveyor belt's position and shape in real time. The processing module is used to perform image segmentation and background denoising on the image to be processed to obtain a processed image with conveyor belt edges; The calculation module is used to calculate the difference between the distance between the straight line of the conveyor belt edge obtained from the processed image using Hough transform and the theoretical straight line of the conveyor belt edge, so as to obtain the calculated deviation of the conveyor belt. The judgment module is used to determine the abnormal condition of the conveyor belt based on the calculated deviation amount. If the calculated deviation amount is greater than the theoretical deviation amount allowed under the working conditions, the alarm system is triggered and an alarm message is sent to the monitoring center; otherwise, the above process is repeated to update the calculated deviation amount.

9. An electronic device, characterized in that, The method includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When executed by the processor, the computer program implements the multi-level vision algorithm fusion method for detecting conveyor belt misalignment in mines as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the multi-level visual algorithm fusion method for detecting conveyor belt deviation in mines as described in any one of claims 1 to 7.

Images