Coal mine main conveying belt tearing detection method and system based on field step

Abstract

The invention relates to the technical field of equipment fault detection, in particular to a coal mine main conveying belt tearing detection method and system based on field step, and the system comprises the following components: a temperature field establishing device, a thermal imaging camera, an edge processor, a display terminal, a far-end server and a conveying belt control system. Comprising the following steps: establishing a temperature field of a conveying belt through a temperature field establishing device; acquiring a thermal infrared image of the step change of the temperature field of the conveying belt by using a thermal imaging camera; detecting the tearing fault of the conveying belt by utilizing the curved surface curvature change of the thermal feature image of the conveying belt; detecting the collected thermal characteristic image of the conveying belt by using the trained conveying belt tearing detection model; and the conveying belt under the mine is detected in real time. The invention provides a coal mine main conveyor belt tearing detection method and system based on field step, which improve the sensitivity and anti-noise capability of tearing detection and solve the problem of failure of traditional visual detection under a low light condition.

Description

Technical Field

[0001] This invention relates to the field of equipment fault detection technology, and in particular to a method and system for detecting tears in the main conveyor belt of a coal mine based on field step. Background Technology

[0002] In underground coal mine production, belt conveyors are widely used transportation equipment, especially for long-distance, high-capacity, and high-speed conveying, where they are indispensable. By transporting materials from the bottom of the mine to the surface, they significantly improve production efficiency. The conveyor belt design is highly adaptable, space-saving, and allows for automated control, further enhancing work efficiency and safety.

[0003] As the main carrying component of conveyors, conveyor belts experience wear and tear during actual underground coal mine transportation. Due to friction from materials and increasing service life, the tensile strength of the conveyor belt decreases, making it prone to scratches and tears, thus affecting the continuous operation of the belt conveyor. Failure to detect and address these issues promptly can lead to significant economic losses and even safety accidents. Therefore, real-time and accurate detection of conveyor belt tears is crucial.

[0004] According to the search, the currently published invention patents related to conveyor belt tear detection mainly include the following:

[0005] Based on material leakage detection, Chinese invention patent application CN 116081227A, published on May 9, 2023, discloses a belt conveyor tear detection device. Its feature is that by employing a backward-inclined side roller, a separating force is applied to the belt. If the belt is torn in the middle by a foreign object, the two torn sections will be forcibly separated when the torn section passes the detection point. At this time, material will leak from the torn section, and a photoelectric switch will send a stop signal, stopping the belt conveyor and preventing further tearing, thus effectively protecting the conveyor. However, if no material leakage occurs in the initial stage of the tear, the conveyor belt tear fault cannot be detected in time.

[0006] Based on stereo vision detection, methods such as point cloud scanning, laser scanning, and 3D structured light are used to detect conveyor belt tearing faults. For example, Chinese invention patent application CN 117699376 A, with a publication date of March 15, 2024, discloses a method and system for detecting longitudinal tearing of conveyor belts based on fusion perception. Its characteristic is that it uses a point cloud system to scan the state of the conveyor belt in real time, reflecting the current information of the conveyor belt in a high frame rate. The information scanned by the point cloud system is digitized, and the model parameters are checked for matching using an algorithm to detect whether the conveyor belt is torn. However, the working environment of main conveyor belts in coal mines is relatively harsh; when the conveyor belt is contaminated with coal particles or other foreign objects, false detections may occur.

[0007] Machine vision-based inspection utilizes visible light images, infrared images, or binary image fusion to detect conveyor belt tearing faults. For example, Chinese invention patent application CN 118037688A, published on May 14, 2024, discloses a method and system for detecting longitudinal tears in conveyor belts based on computer vision recognition. Its key feature is the use of a trained generative adversarial network to fuse visible light and thermal infrared images, using the fused image of the conveyor belt to detect belt cracks. However, when the conveyor belt has minor tears or the tear is not detected in time, the conveyor belt tends to reach a steady-state with a constant temperature after long-term operation. In this case, the tear in the thermal infrared image tends to match that of a healthy belt, leading to inaccurate detection results.

[0008] In summary, while some progress has been made in detecting longitudinal tear faults in conveyor belts, certain shortcomings remain, and the detection methods have significant limitations. To improve upon the deficiencies of existing conveyor belt tear detection methods, this application provides a method and system for detecting tears in main coal mine conveyor belts based on field step jumps. Summary of the Invention

[0009] The present invention aims to provide a method and system for detecting tearing of main conveyor belts in coal mines based on field step, which avoids interference from factors such as low light and high noise in the underground imaging environment of coal mines, and achieves stable detection of longitudinal tearing faults of conveyor belts.

[0010] During prolonged operation without an applied temperature field, the conveyor belt's temperature field tends to reach a uniform steady state. At this point, the uniform continuity of the temperature field makes it difficult for thermal imaging cameras to distinguish between tearing and normal states. Consequently, the cameras fail to capture temperature field step characteristics, resulting in their inability to detect longitudinal tearing faults. To address this issue, based on the concept of transient detection, a method and system for detecting tears in coal mine main conveyor belts based on field step is proposed.

[0011] To achieve the above objectives, the present invention provides the following technical solution: a coal mine main conveyor belt tear detection system based on field step, comprising the following components:

[0012] A temperature field establishment device is set in the front stage in the opposite direction of the conveyor belt running direction. It is used to apply a transient heat source to cause a step change in the temperature field of the conveyor belt. The step change is controlled by a Gaussian diffusion function to control the diffusion range of the heat source in the length direction of the conveyor belt, so that the difference in thermal diffusivity between the torn area and the healthy area becomes significant.

[0013] The thermal imaging camera, an intrinsically safe thermal imaging spherical camera for mining, is installed on the idler roller below the conveyor belt to acquire thermal characteristic images of the conveyor belt in real time. The images are detected based on the temperature increment difference between the torn area and the healthy area.

[0014] The edge processor has a built-in tear detection algorithm based on the YOLO v11 model, which is used to analyze the Gaussian curvature abrupt change features in thermal feature images and output the tear location and confidence score.

[0015] The display terminal shows the tear detection results, conveyor belt operation screen, and temperature distribution cloud map in real time.

[0016] The remote server is used to store detection data and support historical backtracking;

[0017] The conveyor belt control system triggers a shutdown and maintenance command upon receiving a tear fault signal.

[0018] Preferably, the temperature field establishing device applies a transient temperature source to cause a step change in the temperature field of the conveyor belt, that is, a step change occurs in the temperature surface of the thermal characteristic map in the tear area of ​​the conveyor belt, which is mathematically described as a sudden change in the curvature of the thermal characteristic map of the conveyor belt.

[0019] Preferably, the operating parameters of the temperature field establishment device include: the temperature increment of the transient heat source, the Gaussian diffusion coefficient, and the correlation ratio with the conveyor belt running speed;

[0020] The difference in thermal diffusivity is manifested in the fact that the thermal diffusivity rate in the torn region is lower than that in the healthy region, resulting in different rates of temperature decay.

[0021] Preferably, the detection logic of the thermal imaging camera includes:

[0022] The Gaussian curvature abrupt change in the temperature surface of the torn region is captured. The Gaussian curvature is calculated based on the local geometric parameters of the thermal feature surface, and anomalies are determined by comparing the coefficients of the first and second basic forms.

[0023] In the temperature decay model, the rate of temperature increment decay in the torn region is different from that in the healthy region.

[0024] Preferably, the detection process of the edge processor includes:

[0025] After normalizing the thermal imaging image, it is input into the target detection model, and the feature extraction network identifies regions with abrupt changes in Gaussian curvature.

[0026] Candidate boxes are generated and tearing probabilities are predicted. Depthwise separable convolution is used to optimize computational efficiency, and nonmaximum suppression is used to filter the final detection boxes.

[0027] The output includes the tear location coordinates, confidence score, and fault level.

[0028] A method for detecting tears in main conveyor belts in coal mines based on field step, applicable to the aforementioned tear detection system, includes the following steps:

[0029] S1. The temperature field of the conveyor belt is increased by a temperature field establishment device, so that the temperature field of the conveyor belt undergoes a step change;

[0030] S2. Use an explosion-proof thermal imaging camera to acquire thermal infrared images after a step change in the temperature field.

[0031] S3. Detect conveyor belt tearing faults by utilizing the surface curvature changes of the thermal characteristic image of the conveyor belt;

[0032] S4. Use the trained conveyor belt tear detection model to detect the collected thermal feature images of the conveyor belt;

[0033] S5. Real-time detection of underground conveyor belts based on the longitudinal tear detection model of conveyor belts.

[0034] Preferably, in step S1, the temperature field establishment device causes a step change in the temperature field of the conveyor belt through instantaneous heating, and the torn area has abnormal heat conduction performance due to structural damage, forming a temperature difference with the healthy area.

[0035] Preferably, when the temperature field establishment device is applied in step S1, the conveyor belt enters a transient temperature field process; when running for a long time without a temperature field applied, the temperature field of the conveyor belt tends to be steady.

[0036] Preferably, the Gaussian curvature abrupt change analysis in step S3 includes: detecting the torn region based on the curvature change of the temperature surface in the thermal feature image, and realizing anomaly localization by calculating the principal curvature and metric coefficient of the thermal feature surface.

[0037] The beneficial effects of this invention are:

[0038] 1. By using a temperature field establishment device to enhance the temperature field of the conveyor belt, the disruption of the conveyor belt's structure and medium consistency caused by tearing leads to transient changes in the conveyor belt's temperature field in the tear area. This results in a step-like change in the temperature surface of the thermal characteristic map in the tear area. Tear faults can be detected by identifying abrupt changes in the Gaussian curvature of the conveyor belt's thermal characteristic map. Effective highlighting of thermal radiation characteristics helps reduce the impact of environmental factors (such as dust, insufficient light, etc.) on image acquisition, thereby ensuring the system's stability and accuracy in various harsh environments.

[0039] 2. The system uses a target recognition network model to perform target detection on conveyor belt images, automatically identify and locate tearing parts, avoid the inefficiency of traditional manual inspection, greatly improve detection efficiency and accuracy, reduce labor costs, and continuously monitor the belt during operation to ensure the safe operation of the conveyor belt.

[0040] 3. This solution combines temperature difference and target detection technology, enabling real-time and accurate belt tear detection in complex environments. It has strong adaptability and is suitable for belt fault diagnosis in different environments, providing an effective equipment monitoring solution for high-risk industries such as mines. Attached Figure Description

[0041] Figure 1 This is a schematic diagram illustrating the implementation of the tear detection system of the present invention.

[0042] Figure 2 For the present invention Figure 1 A magnified view of a portion of point A in the middle.

[0043] Figure 3 This is a hardware architecture diagram of the tear detection system of the present invention.

[0044] Figure 4 This is a flowchart of the tear detection method of the present invention.

[0045] Figure 5 This is a flowchart of step S4 of the tear detection method of the present invention.

[0046] In the diagram: a is the direction of the conveyor belt; 1, 2, 3, and 4 are temperature field establishment devices; 5 is a thermal imaging camera. Detailed Implementation

[0047] In the description of this invention, it should also be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," and "link" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0048] 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 embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0049] Example 1

[0050] like Figures 1-3 As shown, a field step-based tear detection system for coal mine main conveyor belts is used to perform the aforementioned field step-based tear detection method for coal mine main conveyor belts, including:

[0051] Temperature field establishment device, used to enhance the temperature field of the conveyor belt;

[0052] Thermal imaging cameras are used to acquire thermal characteristic images of conveyor belts;

[0053] Both the temperature field establishment device and the thermal imaging camera meet strict explosion-proof standards and are certified with relevant qualifications such as coal mine safety certificates.

[0054] Hard disk video recorders use fiber optic cables to centrally store and manage videos captured by thermal imaging cameras.

[0055] The edge processor reads and processes the thermal imaging images stored in the hard disk recorder frame by frame through optical fiber to realize the above-mentioned method for detecting tearing of the main conveyor belt in coal mines based on field step.

[0056] The display terminal shows information such as tear fault detection results and real-time operation of the belt conveyor via video signal cable.

[0057] The conveyor belt control system detects a tear in the conveyor belt and controls the belt conveyor to stop for maintenance.

[0058] Remote servers are used for collecting, storing, and analyzing data, as well as for fault detection and alarm notifications.

[0059] The temperature field establishment device is used to apply a transient step change to the temperature field of the conveyor belt. A thermal imaging spherical camera is used to acquire thermal characteristic images of the conveyor belt to determine if it is torn. The display terminal can be a computer, LCD screen, or other display device, which can be installed outside the conveyor belt operating site or in the monitoring room to display tear fault detection results and real-time conveyor belt operation information, facilitating manual monitoring.

[0060] The temperature field establishment device applies a transient temperature source to cause a step change in the temperature field of the conveyor belt, that is, a step change occurs in the temperature surface of the thermal characteristic map in the tear area of ​​the conveyor belt, which is mathematically described as a sudden change in the curvature of the thermal characteristic map of the conveyor belt.

[0061] The operating parameters of the temperature field establishment device include: the temperature change of the transient heat source, the Gaussian diffusion coefficient, and the correlation ratio with the conveyor belt running speed;

[0062] The difference in thermal diffusivity is manifested in the fact that the thermal diffusivity rate in the torn region is different from that in the healthy region, resulting in different rates of temperature decay.

[0063] The detection logic of a thermal imaging camera includes:

[0064] The Gaussian curvature abrupt change in the temperature surface of the torn region is captured. The Gaussian curvature is calculated based on the local geometric parameters of the thermal feature surface, and anomalies are determined by comparing the coefficients of the first and second basic forms.

[0065] In the temperature decay model, the decay rate of temperature change in the torn region is different from that in the healthy region.

[0066] The edge processor detection process includes:

[0067] After normalizing the thermal imaging image, it is input into the target detection model, and the feature extraction network identifies regions with abrupt changes in Gaussian curvature.

[0068] Candidate boxes are generated and tearing probabilities are predicted. Depthwise separable convolution is used to optimize computational efficiency, and nonmaximum suppression is used to filter the final detection boxes.

[0069] The output includes the tear location coordinates, confidence score, and fault level.

[0070] Example 2

[0071] like Figure 4 and Figure 5 As shown, this invention discloses a method for detecting tearing of main conveyor belts in coal mines based on field step, which avoids interference from factors such as low light and high noise in the underground imaging environment of coal mines, and achieves stable detection of longitudinal tearing faults in conveyor belts.

[0072] Specifically, it includes the following steps:

[0073] Step S1: A temperature field establishment device is added to the front stage of the thermal imaging camera (in the opposite direction of the conveyor belt operation) to briefly enhance the temperature field of the conveyor belt, causing a step change in the temperature field of the conveyor belt.

[0074] When a conveyor belt operates for an extended period without an applied temperature field, the entire conveyor belt will tend towards a steady-state with constant temperature. At this point, the torn areas and healthy sections of the belt will appear similar in thermal infrared images. For steady-state heat conduction (i.e., minimal change over time), the fundamental equation is:

[0075]

[0076] For Laplace's equation, Let T be the Laplace operator.

[0077] By using a temperature field establishment device, the temperature field of the conveyor belt undergoes a step change, at which point the heat conduction equation becomes:

[0078]

[0079] In the formula, T(x,y,t) represents the temperature at position (x,y) at time t, and α health and α tear These are the thermal diffusivity coefficients of the healthy belt and the tear, respectively, defined as... Where k i It is the thermal conductivity, ρ i It's density, it's c i Specific heat capacity, i∈{health,tear}. ΔT heat It is the temperature increase caused by the temperature source. heart σ represents the position of the temperature source, and x(t) represents the position on the conveyor belt. The influence of the temperature source moves along the x-direction. σ controls the diffusion range of the heat source in the x-direction, i.e., the width of the temperature range. S is the Gaussian diffusion function, which describes the spatial influence of a temperature source on temperature, indicating that the influence of the heat source gradually decreases with distance from the heat source. health For the belt health zone, S tear This is the area where the conveyor belt is torn.

[0080] Furthermore, the temperature field establishment device causes a momentary change in the conveyor belt, after which heat begins to dissipate. The structure of the torn section of the conveyor belt may have been damaged, leading to abnormal thermal conductivity in that area. The abnormal thermal conductivity of the torn region will therefore produce a certain temperature difference. The temperature of the torn region can be represented by an exponential decay model:

[0081]

[0082] In the formula, T0 is the background temperature (i.e., the temperature before heating). ΔT health It is the temperature increment of the healthy region, ΔT tear λ is the temperature increment in the torn region. λ is the temperature decay coefficient. r(x,y,t) is the distance from the current position to the area of ​​effect of the temperature field establishing device.

[0083] Furthermore, since the conveyor belt is in motion, the location of the tear changes over time. Assuming the conveyor belt speed is υ, the location of the tear in the x-direction changes over time as follows: x tear (t)=x tear (0)+υ·t. Therefore, the location of the tear changes with time and can be expressed as:

[0084]

[0085] A step change in the temperature field makes the temperature change at the tear site more significant, thereby enhancing the temperature characteristics of the tear site in the thermal infrared image.

[0086] Step S2: The conveyor belt is located above the idler rollers. As the idler rollers rotate, an intrinsically safe thermal imaging spherical camera located below the belt collects data on the conveyor belt running on the idler rollers (e.g., ...). Figure 1 As shown in the figure, real-time thermal characteristic images of the conveyor belt operation were acquired and transmitted to the hard disk video recorder via optical fiber.

[0087] Step S3: Under the action of the temperature field establishment device, the temperature field at the tear of the conveyor belt changes. Due to the damage to the consistency between the conveyor belt structure and the medium caused by the tear, there is a significant difference in the temperature field between the tear and the healthy belt. The thermal characteristic map shows a step change in the temperature surface of the tear area, which is described in a scientific sense as a sudden change in the Gaussian curvature of the thermal characteristic map of the conveyor belt.

[0088] Let the local coordinates of point p on the thermal characteristic map M of the conveyor belt be (u, v), then the Gaussian curvature K can be expressed as:

[0089]

[0090] Where k1 and k2 are the principal curvatures, and L, M, and N are the coefficients of the thermal characteristic surface (coefficients of the second basic form), defined as: Ⅰ=Edu 2 +2Fdudv+Gdv 2 , where E = <r u ,r u >、F= <r u ,r v >、G= <r v ,r v >, here r u and rv These are the tangent vectors of the thermal feature surface in the u and v directions, respectively;

[0091] E, F, and G are the coefficients of the first fundamental form, defined as: II = Ldu 2 +2Mdudv+Ndv 2 , where L = <r uu ,n>、M= <r uv ,n>、N= <r vv ,n>,here r uu r uv r vv These are the second-order partial derivatives of the surface, and n is the unit normal vector on the surface.

[0092] Step S4: In the NVIDIA Jestson Orin NX edge processor, the thermal feature image of the conveyor belt acquired by the intrinsically safe thermal imaging spherical camera is detected by combining the trained conveyor belt longitudinal tear detection model (Gaussian curvature anomaly detection). The detection principle is described in step S3.

[0093] Taking the YOLO v11 network as an example, the steps for detecting conveyor belt tears are as follows: Figure 5 As shown.

[0094] First, the thermal images of the conveyor belt acquired by the intrinsically safe thermal imaging spherical camera for mining are fed into the trained YOLOv11 model.

[0095] Next, image preprocessing is performed. To adapt to the model's input requirements, the input image size needs to be adjusted to the size expected by the model. Then, the image pixel values ​​are normalized. This can speed up model training and improve model stability.

[0096] Next, tear feature extraction is performed. Multi-scale features are extracted through a backbone network (such as C3k2 structure). Through a series of convolutional layers, pooling layers, and other operations, low-level features and high-level semantic features are gradually extracted from the original image. Multi-scale features are aggregated with the help of a neck network (PAN combined with C3k2, etc.). The features of different scales extracted by the backbone network are aggregated and optimized, so that the model can accurately detect small targets while detecting large targets.

[0097] To perform target prediction, candidate boxes are first generated at various locations in the feature map. Multiple candidate boxes of different scales and aspect ratios are generated at each location in the feature map obtained after feature extraction, covering all regions in the feature image that may contain Gaussian curvature anomalies. Then, the tear probability within the candidate boxes is predicted. For each generated candidate box, techniques such as depthwise separable convolution are used to predict the probability of conveyor belt tearing within the candidate box. Finally, the regression branch predicts the position and size of the candidate boxes (using DFL+CIoU Loss), adjusting the position and size of the candidate boxes to more accurately frame the tear area.

[0098] Finally, post-processing is performed. Non-maximum suppression (NMS) is used to filter redundant boxes. The best candidate box for each target is selected by comparing the class confidence (i.e., tear probability) and intersection-over-union (IoU) ratio of the candidate boxes. The final results (target class, location, and confidence score) are output. After NMS processing, what remains is the final detection result. These results include the tear location (usually represented by the coordinates of the bounding box, such as the coordinates of the top left and bottom right corners) and the confidence score (indicating the model's confidence in this detection result).

[0099] Step S5: Real-time detection is performed using the trained YOLO v11 tear detection model (Gaussian curvature anomaly detection model) to obtain tear detection results. The detection results and real-time conveyor belt operation footage are transmitted to a display terminal via fiber optic cable, and simultaneously, relevant information is transmitted to a remote server via a local area network. Upon detecting a tear fault in the conveyor belt, a signal is fed back to the conveyor belt control system, which then controls the belt conveyor to stop for maintenance. The fault is also marked on the real-time display terminal, facilitating timely identification and handling by operators.

[0100] The above are merely preferred embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A tear detection system for main conveyor belts in coal mines based on field step, characterized in that, Includes the following components: A temperature field establishment device is set in the front stage in the opposite direction of the conveyor belt running direction. It is used to apply a transient heat source to cause a step change in the temperature field of the conveyor belt. The step change is controlled by a Gaussian diffusion function to control the diffusion range of the heat source in the length direction of the conveyor belt, so that the difference in thermal diffusivity between the torn area and the healthy area becomes significant. The thermal imaging camera, an intrinsically safe thermal imaging camera for mining, is installed on the idler roller below the conveyor belt to acquire thermal characteristic images of the conveyor belt in real time. The images are detected based on the temperature increment difference between the torn area and the healthy area. The edge processor has a built-in tear detection algorithm based on the target detection model. It is used to analyze the Gaussian curvature abrupt change features in thermal feature images and output the tear location and confidence score. The display terminal shows the tear detection results, conveyor belt operation screen, and temperature distribution cloud map in real time. The remote server is used to store detection data and support historical backtracking; The conveyor belt control system triggers a shutdown and maintenance command upon receiving a tear fault signal.

2. The coal mine main conveyor belt tear detection system based on field step as described in claim 1, characterized in that, The temperature field establishment device applies a transient temperature source to cause a step change in the temperature field of the conveyor belt, that is, a step change occurs in the temperature surface of the thermal characteristic map in the tear area of ​​the conveyor belt, which is mathematically described as a sudden change in the curvature of the thermal characteristic map of the conveyor belt.

3. The coal mine main conveyor belt tear detection system based on field step as described in claim 1, characterized in that, The operating parameters of the temperature field establishment device include: the temperature change of the transient heat source, the Gaussian diffusion coefficient, and the correlation ratio with the conveyor belt running speed; The difference in thermal diffusivity is manifested in the fact that the thermal diffusivity rate in the torn region is different from that in the healthy region, resulting in different rates of temperature decay.

4. The coal mine main conveyor belt tear detection system based on field step as described in claim 1, characterized in that, The detection logic of the thermal imaging camera includes: The Gaussian curvature abrupt change in the temperature surface of the torn region is captured. The Gaussian curvature is calculated based on the local geometric parameters of the thermal feature surface, and anomalies are determined by comparing the coefficients of the first and second basic forms. In the temperature decay model, the decay rate of temperature change in the torn region is different from that in the healthy region.

5. The coal mine main conveyor belt tear detection system based on field step as described in claim 1, characterized in that, The detection process of the edge processor includes: After normalizing the thermal imaging image, it is input into the target detection model, and the feature extraction network identifies regions with abrupt changes in Gaussian curvature. Candidate boxes are generated and tearing probabilities are predicted. Depthwise separable convolution is used to optimize computational efficiency, and nonmaximum suppression is used to filter the final detection boxes. The output includes the tear location coordinates, confidence score, and fault level.

6. A method for detecting tears in a coal mine main conveyor belt based on a field step, applicable to the tear detection system as described in any one of claims 1 to 5, characterized in that, Includes the following steps: S1. The temperature field of the conveyor belt is increased by a temperature field establishment device, so that the temperature field of the conveyor belt undergoes a step change; S2. Use an explosion-proof thermal imaging camera to acquire thermal infrared images after a step change in the temperature field. S3. Detect conveyor belt tearing faults by utilizing the surface curvature changes of the thermal characteristic image of the conveyor belt; S4. Use the trained conveyor belt tear detection model to detect the collected thermal feature images of the conveyor belt; S5. Real-time detection of underground conveyor belts based on the longitudinal tear detection model of conveyor belts.

7. The method for detecting tearing of a main conveyor belt in a coal mine based on a field step, as described in claim 5, is characterized in that... In step S1, the temperature field establishment device causes a step change in the temperature field of the conveyor belt through instantaneous temperature field construction, and the torn area has abnormal heat conduction performance due to structural damage, forming a temperature field difference with the healthy area.

8. The method for detecting tearing of a main conveyor belt in a coal mine based on a field step, as described in claim 5, is characterized in that... When the temperature field establishment device is applied in step S1, the conveyor belt enters a transient temperature field process; when running for a long time without a temperature field applied, the temperature field of the conveyor belt tends to be steady.

9. The method for detecting tearing of a main conveyor belt in a coal mine based on a field step, as described in claim 5, is characterized in that... The Gaussian curvature mutation analysis in step S3 includes: detecting the torn region based on the curvature change of the temperature surface in the thermal feature image, and realizing anomaly localization by calculating the principal curvature and metric coefficient of the thermal feature surface.

Images