A method and system for monitoring the operating status of a mining belt conveyor

Abstract

This invention belongs to the field of coal mine transportation technology, specifically relating to a method and system for monitoring the operating status of a mining belt conveyor. The method includes: simultaneously acquiring multi-source video image data of the belt surface, transported materials, and trackless rubber-tired vehicles via several AI imaging components and mining network cameras spaced longitudinally along the belt conveyor; inputting the pre-processed standardized image data into a pre-trained AI recognition component to perform parallel foreign object detection, belt tear recognition, and trackless rubber-tired vehicle speed measurement and analysis, generating preliminary anomaly recognition results; and introducing an environmental-operating condition joint compensation coefficient to comprehensively consider multiple factors such as gas concentration, dust concentration, and belt speed, thereby generating quantitative indicators of hazard levels and precise linkage control commands. This ensures that alarms and control commands are no longer based on simple judgments of a single indicator, greatly improving the accuracy and reliability of the response.

Description

Technical Field

[0001] This invention belongs to the field of coal mine transportation technology, specifically relating to a method and system for monitoring the operating status of a mining belt conveyor. Background Technology

[0002] As a key transportation equipment in coal mine production, the operating status of belt conveyors directly affects the production efficiency and safety of the entire mine. Due to the harsh underground environment of coal mines, especially the presence of large amounts of dust, uneven lighting, and limited space, real-time and reliable monitoring of belt conveyors has always been a technical challenge in the industry.

[0003] Currently, monitoring the operating status of belt conveyors mainly relies on two methods: one is traditional manual periodic inspections and monitoring systems based on simple sensors; the other is the emerging video image-based monitoring technology. Manual inspections suffer from problems such as long intervals, susceptibility to subjective factors, and the inability to achieve 24-hour uninterrupted monitoring, making it difficult to detect potential faults such as belt misalignment, surface tearing, and idler jamming in a timely manner. Simple sensors (such as pull-cord switches and belt misalignment switches) have limited monitoring range, high false alarm rates, and cannot identify complex abnormal conditions, such as large foreign objects or belt joint damage.

[0004] While video surveillance technology provides intuitive visual information, in practical applications, especially in underground environments with strong dust interference, the image quality captured by conventional cameras deteriorates significantly, resulting in low recognition accuracy based on traditional image processing algorithms. Furthermore, existing technical solutions are mostly limited to single-function monitoring (e.g., detecting only belt misalignment or only identifying personnel), lacking the ability to comprehensively and collaboratively perceive and intelligently analyze the operating status of the conveyor belt system (e.g., material transport status, equipment health status, and surrounding vehicle behavior). Summary of the Invention

[0005] To address the aforementioned shortcomings of the prior art, this invention provides a method, system, terminal, and storage medium for monitoring the operating status of a mining belt conveyor.

[0006] In a first aspect, the present invention provides a method for monitoring the operating status of a mining belt conveyor, comprising: S1. By using several AI shooting components and mining network cameras distributed at intervals along the longitudinal direction of the belt conveyor, multi-source video image data of the belt surface, the transported materials and the trackless rubber-wheeled vehicle are collected simultaneously. S2. The multi-source video image data is preprocessed to resist dust interference by an intelligent recognition device to obtain standardized image data; S3. Input standardized image data into the pre-trained AI recognition component, and perform parallel foreign object unloaded detection, belt tear recognition and trackless rubber-tired vehicle speed measurement and analysis to generate preliminary anomaly recognition results; S4. Transmit the initial anomaly identification results to several parallel management servers for data fusion analysis, and generate linkage control instructions containing anomaly type, location and hazard level. S5. Send the linkage control command to the client to drive the external audio system, loudspeaker, and printer to perform graded sound and light alarms and evidence solidification output; S6. Synchronously store multi-source video image data, standardized image data, primary anomaly recognition results, and linkage control commands to the storage system.

[0007] Further improvements to this technical solution include step S1, which includes: S11. Obtain the rated conveying capacity of the belt conveyor, the running speed of the belt in the belt conveyor, and the wind speed at the cross-section of the roadway where the belt conveyor is located. Based on the obtained rated conveying capacity, belt running speed, and roadway cross-section wind speed, dynamically optimize the spacing along the longitudinal direction of the belt conveyor. Deploy several AI-powered imaging components and mining network cameras; dynamically optimize spacing. The calculation formula is: ; in, Rated conveying capacity of the belt conveyor, unit: t / h; The speed of the belt in a conveyor belt, in m / s; The wind speed at the cross-section of the roadway where the belt conveyor is located, in m / s; The maximum allowable response delay time for the system, in seconds; The safety factor is set at a value not less than 1.5. Design the maximum conveying capacity for the belt conveyor; This is the load influence coefficient; Dust settling height threshold, unit: m; Environmental dust concentration, unit: ; The field of view of the camera is half an angle, in rad. S12. At the beginning of each data collection cycle, based on the real-time monitored illuminance... and environmental dust concentration Adaptively adjust the exposure time of the AI ​​shooting component Gain of the image sensor in the AI ​​shooting component The adaptive adjustment formula is: ; ; in, Full well capacity for the preset image sensor; This is a function for the dust attenuation factor; Pixel photosensitivity, unit: V / lux·s; and This is the adjustment coefficient; For reference illuminance; For reference dust concentration; This is the reference gain value; S13. Synchronize the clocks of all AI shooting components with the mining network camera using the IEEE 1588v2 time protocol, and embed an absolute timestamp based on GPS / BeiDou dual-mode time synchronization into each video frame. ; S14. After the acquisition is completed, perform real-time quality assessment on each frame of multi-source video image data and calculate the image sharpness index. With dust blocking rate Only when the following conditions are met and If the data is in the correct state, send it to step S2; otherwise, re-collect the data. The quality assessment formula is: ; ; in, The gray value of the pixel (x, y); For the Laplace operator; and These are the image width and height in pixels, respectively. Low-light suppression coefficient; This is the maximum grayscale value; A collection of low grayscale regions; It is an indicator function; The threshold for determining dust obstruction.

[0008] Further improvements to this technical solution include step S2, which includes: S21. Through the intelligent recognition device integrated inside the mining network camera, image dust removal processing based on dark channel prior theory is performed on multi-source video image data to obtain a preliminary dust-removed image; S22. Perform contrast-limited adaptive histogram equalization enhancement on the preliminary dust removal image to obtain a contrast-enhanced image; the contrast limit threshold is based on the real-time dust concentration. Dynamic calculation, the calculation formula is as follows: ; in, Represents the local noise variance of the image; This represents the local average gray value; This is the dust environmental compensation coefficient; S23. Adaptive median filtering algorithm is used to remove impulse noise from contrast-enhanced images; S24. After completing the above three preprocessing steps, the intelligent recognition device performs real-time quality assessment on the obtained standardized image data and calculates the overall image quality index. Only when Exceeding the preset quality index threshold The image is only sent to step S3 if the condition is met; otherwise, the parameters in steps S21 to S23 are adjusted and reprocessed. The formula for calculating the overall quality index is: ; in, and These represent the signal-to-noise ratios of the images before and after preprocessing, respectively. This represents the increase in contrast. This represents the maximum permissible enhancement level. Residual noise rate; Maximum tolerance noise rate; , , These are the weighting coefficients; feedback adjustments are made through iterative optimization. It converges to the maximum value.

[0009] Further improvements to this technical solution include step S3, which includes: S31. Standardized image data is input in batches into a pre-trained lightweight YOLOv5s improved AI recognition component. The lightweight YOLOv5s improved AI recognition component is a dust suppression attention module embedded at the end of the backbone network of a general lightweight YOLOv5s model. This module processes the input feature maps. Channel weighting calibration was performed to obtain the feature map after dust suppression. The calculation formula is as follows: ; in, and This is the weight matrix of the fully connected layer. This refers to the channel reduction ratio; and It is the bias vector; This indicates a global average pooling operation; Use the Sigmoid activation function; It acts as a dust-inhibiting factor. S32. Feature map based on dust suppression The system executes three subtasks in parallel: foreign object unloaded detection, belt tear identification, and trackless rubber-tired vehicle speed measurement and analysis. Each subtask shares a feature extraction layer but has an independent detection head. The foreign object unloaded detection employs a background difference and texture feature fusion mechanism, and its foreign object confidence level... The calculation formula is: ; in, The pixel value of the current frame; These are the pixel values ​​of the background model. This refers to local texture contrast features based on the gray-level co-occurrence matrix; For fusion weights; Meanwhile, belt tear identification employs a multi-scale edge gradient direction histogram matching algorithm, which measures the tear damage quantification value. The calculation formula is: ; in, The number of detection sub-blocks to divide the belt surface; Let be the real-time gradient vector of the i-th subtask; This represents the gradient vector of the normal belt template at the corresponding position. This indicates element-wise division; The weighting coefficient for the i-th subtask; Let be the dust occlusion factor for the i-th subtask; Let be the dust attenuation coefficient for the i-th subtask; Furthermore, the speed measurement analysis of trackless rubber-tired vehicles employs an improved dense optical flow method, and its vehicle speed... The calculation formula is: ; in, Vehicle center of gravity coordinates The optical flow vector at that location; Image spatial resolution; The tracking efficiency loss factor under dust obstruction. ; To track the efficiency loss coefficient; The camera's frame rate; This is the physical length calibration value corresponding to a unit pixel; The angle between the vehicle's direction of travel and the optical axis; S33. The outputs of the three sub-tasks are spatiotemporally aligned and weighted by confidence in the detection result fusion layer to generate a unified primary anomaly detection result. The fusion formula is: ; in, This indicates a spatiotemporal nonmaximum suppression fusion operation; It is the kth type of exception; The coordinates of the k-th anomaly; is the original confidence level for the k-th type of anomaly; The environmental-operating condition joint compensation coefficient is... ; As a task type sensitivity factor, it is used to adaptively improve detection reliability in dusty environments and high-speed belt operation conditions; The maximum operating speed preset for the belt conveyor.

[0010] Further improvements to this technical solution include step S4, which includes: S41. The initial anomaly identification results... Transmitted via an intrinsically safe fiber optic ring network to a parallel management server cluster built on the Raft consensus protocol. The cluster contains at least three nodes, each managing its real-time processing load. Memory usage and network latency Dynamic election of master nodes, master node election weight The calculation formula is: ; in, , , These are the preset CPU load limit, memory limit, and network latency limit; , , These are the election weighting coefficients; S42. Perform multi-source heterogeneous data fusion analysis on the master node to analyze the preliminary anomaly identification results from different AI recognition components. Spatiotemporal alignment and confidence-weighted fusion are performed. An improved DS evidence theory algorithm is used to handle conflicts and redundancies in the results of foreign object detection, belt tear identification, and trackless rubber-tired vehicle speed measurement. The confidence of the fused abnormal events is then calculated. The calculation formula is: ; Among them, A is the final abnormal event after fusion; B is the foreign object detection evidence; C is the belt tear evidence; and D is the vehicle speed measurement evidence. , , The confidence levels of evidence for foreign object detection, belt tearing, and abnormal vehicle speed are respectively: The dynamic fusion weights based on spatiotemporal distance and sensor reliability are calculated as follows: ; in, , These represent the temporal and spatial deviations of the three types of test results; , For time air conditioning and parameters; The sensor reliability of the k-th AI recognition component; This represents the global conflict coefficient. For environmental reliability factor, ; , This is the environmental weighting coefficient; Gas concentration; The preset maximum gas concentration; This is the preset maximum dust concentration; S43, Based on the confidence level of fused anomaly events Combined with abnormal location information and the real-time operating speed of the belt Dynamically calculate the hazard level It generates a linkage control instruction package containing equipment control logic, alarm priority, and evidence storage strategy, and determines the hazard level. The calculation formula is: ; in, and These are the high confidence threshold and the medium confidence threshold, respectively. The critical safe speed for the belt; This is a quantification of tear damage; This represents the critical threshold for tear damage.

[0011] Further improvements to this technical solution include step S5, which includes: S51. After the client performs CRC check and digital signature verification on the received linkage control command packet, it parses and extracts the danger level. Abnormal location coordinates Evidence locking time window and alarm priority weight And query the local configuration database to get the number of workers currently working in the roadway. Compared with real-time background noise in decibels ; S52. Hazard level obtained from analysis Dynamically calculate the alarm volume of the sound system and speaker. Alarm frequency and duration of sound and light The sound system is installed in the underground control room, while loudspeakers are deployed along the conveyor belt tunnels. Together, they achieve zoned coverage alarms; the alarm volume... The calculation formula is: ; in, The rated maximum volume for explosion-proof speakers; Used as the baseline alarm volume; This is the hazard level increment coefficient; Numerical mapping of hazard levels; For reference number of workers; Background noise as reference; speaker trigger delay time Based on the distance from the abnormal location The calculation formula is as follows: ; in, The speed at which sound travels in the alleyway environment; This is the network transmission delay coefficient; For client command processing time; S53. While the client is activating the audible and visual alarm, it locks the time window based on the evidence. Automatically capture and cache multi-source video image data and standardized image data, depending on the hazard level. When Level I or Level II is reached, the external mining explosion-proof printer is triggered to perform evidence solidification output.

[0012] Further improvements to this technical solution include step S6, which includes: S61. After receiving the linkage control command packet, the client determines the risk level based on the severity of the hazard. Data types Based on downhole storage environment parameters, dynamically calculate the storage priority weight of various types of data. Retention period and the number of redundant copies Develop differentiated storage strategies and prioritize storage. The calculation formula is: ; in, Numerical mapping of hazard levels; Quantize values ​​for data types; , , As a weighting coefficient; when the environmental dust concentration More than 150 Time increases Prioritize the storage of critical data; S62. Based on the differentiated storage strategy, perform adaptive compression coding based on content importance on multi-source video image data, and automatically increase the coding bitrate when the danger level is Level I or Level II. Interval with keyframe For periods without abnormalities, low bitrate storage is used; compression ratio The adjustment is dynamically based on storage priority weights and available bandwidth, and the adjustment formula is as follows: ; ; in, The preset base compression ratio; This is the hazard level influence coefficient; The maximum storage priority weight; This is the time-varying bandwidth efficiency factor; Let t be the network load rate at time t; This represents the maximum network bandwidth capacity. This is the bandwidth attenuation coefficient; S63. Separate the compressed video stream, standardized image data, primary anomaly detection results, and linkage control instructions by timestamp. The components are aligned and packaged into a unified storage unit, and written to an explosion-proof certified NAS storage system via the iSCSI protocol.

[0013] Secondly, the present invention provides an operating status monitoring system for a mining belt conveyor, comprising: The data acquisition module includes several AI shooting components and mining network cameras distributed at intervals along the longitudinal direction of the belt conveyor, used to simultaneously acquire multi-source video image data of the belt surface, the transported materials and the trackless rubber-wheeled vehicle on the belt conveyor; The image preprocessing module, which is connected in communication with the data acquisition module, includes an intelligent recognition device for performing anti-dust interference preprocessing on multi-source video image data to obtain standardized image data. The AI ​​recognition and analysis module communicates with the image preprocessing module and includes a pre-trained AI recognition component. This component is used to perform parallel foreign object detection, belt tear recognition, and trackless rubber-tired vehicle speed measurement and analysis on standardized image data, and generate preliminary anomaly recognition results. The data fusion and decision-making module communicates with the AI ​​recognition and analysis module, including several management servers set in parallel, which are used to perform data fusion analysis on the initial anomaly recognition results and generate linkage control instructions containing anomaly type, location and danger level. The alarm and evidence output module, which communicates with the data fusion and decision module, includes a client and the audio, loudspeaker and printer driven by it, and is used to execute graded sound and light alarms and evidence solidification output according to the linkage control command. The data storage module is communicatively connected to the data acquisition module, image preprocessing module, AI recognition and analysis module, and data fusion and decision-making module, and includes a storage system for synchronously storing multi-source video image data, standardized image data, primary anomaly recognition results, and linkage control commands.

[0014] Thirdly, the present invention provides a terminal, comprising: Processor, memory, among which, This memory is used to store computer programs. The processor is used to retrieve and run the computer program from memory, causing the terminal to perform the terminal method described above.

[0015] Fourthly, the present invention provides a computer storage medium storing instructions that, when executed on a computer, cause the computer to perform the methods described in the above aspects.

[0016] The beneficial effects of this invention are as follows: At the data acquisition end (S1), this invention maximizes the availability of image data from the source by using a dynamic optimization deployment model based on operating parameters such as conveying capacity and wind speed, and an adaptive exposure gain adjustment algorithm. At the preprocessing end (S2), image dust removal and dynamic contrast enhancement based on dark channel priors are used to effectively restore image details attenuated by dust. At the core recognition end (S3), a dust suppression attention module (DSAM) is innovatively embedded in the YOLOv5s model, enabling the model to focus on robust features that are not easily obscured by dust. Dedicated anti-interference quantification evaluation formulas (including foreign object confidence, tear damage, and vehicle speed) are designed for different tasks such as foreign objects, tears, and vehicle speed. These progressive and synergistic anti-interference designs ensure that the system maintains extremely high recognition accuracy and extremely low false alarm rate even in strong dust environments.

[0017] This invention overcomes the limitations of existing technologies, which are mostly single-function monitoring systems, and constructs a multi-task parallel, multi-source information fusion intelligent sensing system. Through a parallel management server cluster and an improved DS evidence theory fusion algorithm (S4), the system can perform spatiotemporal alignment and confidence-weighted fusion of abnormal information from different locations and types, effectively solving the problem of uncertainty or conflict in single sensor information. Based on this, an environmental-operating condition joint compensation coefficient is introduced, comprehensively considering multi-dimensional factors such as gas concentration, dust concentration, and belt speed, thereby generating quantitative indicators of hazard levels and precise linkage control commands. This system-level intelligent decision-making mechanism ensures that alarms and control commands (such as emergency stops) are no longer based on simple judgments of single indicators, but on scientific decisions based on multi-dimensional evidence, greatly improving the accuracy and reliability of the response.

[0018] This invention not only focuses on real-time monitoring but also emphasizes data management and traceability throughout the entire process. Through a refined hierarchical alarm strategy (S5), the alarm volume and range are dynamically adjusted based on hazard level, personnel distribution, and environmental noise, achieving accurate early warning and avoiding excessive disturbance. Simultaneously, the system automatically triggers evidence solidification output, strongly correlating key data with alarm events. At the data storage end (S6), an innovative dynamic differentiated storage strategy based on data value (hazard level, type) and storage environment is proposed. Through adaptive compression coding, storage resources are intelligently allocated, ensuring the integrity and long-term traceability of key data while improving storage efficiency and economy. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a schematic flowchart illustrating a method according to an embodiment of the present invention.

[0021] Figure 2 This is a schematic block diagram of a system according to an embodiment of the present invention.

[0022] Figure 3 This is a schematic diagram of the structure of a terminal provided in an embodiment of the present invention. Detailed Implementation

[0023] To make the objectives, features, and advantages of this invention more apparent and understandable, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings of the specific embodiments. Obviously, the embodiments described below are only some embodiments of this invention, and not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0024] Unless otherwise defined, 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. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.

[0025] Figure 1 This is a schematic flowchart illustrating a method for monitoring the operational status of a mining belt conveyor, provided by the present invention. Wherein, Figure 1The implementing entity can be a monitoring system for the operational status of a mining belt conveyor. Depending on different requirements, the order of the steps in this flowchart can be changed, and some steps can be omitted.

[0026] like Figure 1 As shown, the method includes: S1. By using several AI shooting components and mining network cameras distributed at intervals along the longitudinal direction of the belt conveyor, multi-source video image data of the belt surface, the transported materials and the trackless rubber-wheeled vehicle are collected simultaneously. S2. The multi-source video image data is preprocessed to resist dust interference by an intelligent recognition device to obtain standardized image data; S3. Input standardized image data into the pre-trained AI recognition component, and perform parallel foreign object unloaded detection, belt tear recognition and trackless rubber-tired vehicle speed measurement and analysis to generate preliminary anomaly recognition results; S4. Transmit the initial anomaly identification results to several parallel management servers for data fusion analysis, and generate linkage control instructions containing anomaly type, location and hazard level. S5. Send the linkage control command to the client to drive the external audio system, loudspeaker, and printer to perform graded sound and light alarms and evidence solidification output; S6. Synchronously store multi-source video image data, standardized image data, primary anomaly recognition results, and linkage control commands to the storage system.

[0027] To facilitate understanding of the present invention, the following description further illustrates the method for monitoring the operating status of a mining belt conveyor provided by the present invention, based on the principle of the method for monitoring the operating status of a mining belt conveyor and in conjunction with the process of monitoring the operating status of a mining belt conveyor in the embodiments.

[0028] First, step S1 includes: S11. Obtain the rated conveying capacity of the belt conveyor, the running speed of the belt in the belt conveyor, and the wind speed at the cross-section of the roadway where the belt conveyor is located. Based on the obtained rated conveying capacity, belt running speed, and roadway cross-section wind speed, dynamically optimize the spacing along the longitudinal direction of the belt conveyor. Deploy several AI-powered imaging components and mining network cameras; dynamically optimize spacing. The calculation formula is: ; in, Rated conveying capacity of the belt conveyor, unit: t / h; The speed of the belt in a conveyor belt, in m / s; The wind speed at the cross-section of the roadway where the belt conveyor is located, in m / s; The maximum allowable response delay time for the system, in seconds; The safety factor is set at a value not less than 1.5. Design the maximum conveying capacity for the belt conveyor; This is the load influence coefficient; Dust settling height threshold, unit: m; Environmental dust concentration, unit: ; The field of view of the camera is half an angle, in rad. S12. At the beginning of each data collection cycle, based on the real-time monitored illuminance... and environmental dust concentration Adaptively adjust the exposure time of the AI ​​shooting component Gain of the image sensor in the AI ​​shooting component The adaptive adjustment formula is: ; ; in, Full well capacity for the preset image sensor; It is a function of dust attenuation factor and satisfies , The dust scattering intensity coefficient is... For the nonlinear decay term of dust concentration, when hour, Indicates no attenuation, when hour, This indicates that the brightness of the background model needs to be increased to counteract the dust occlusion effect; Pixel photosensitivity, unit: V / lux·s; and The values ​​are adjustment coefficients, and their ranges are [0.2, 0.5] and [3, 7], respectively. For reference illuminance; For reference dust concentration; This is the reference gain value; S13. Synchronize the clocks of all AI shooting components with the mining network camera using the IEEE 1588v2 time protocol, and embed an absolute timestamp based on GPS / BeiDou dual-mode time synchronization into each video frame. Timestamp synchronization error The control formula is: ; in, The inherent latency of the network layer, in μs; The data size of a single frame of image data, in bits; Network bandwidth, unit: Mbps; To account for electromagnetic interference in dusty environments and satisfy the bandwidth efficiency factor. , The interference coefficient ranges from [0.01, 0.03]. Crystal oscillator frequency stability, unit: ppm / °C; The change in ambient temperature is expressed in °C. S14. After the acquisition is completed, perform real-time quality assessment on each frame of multi-source video image data and calculate the image sharpness index. With dust blocking rate Only when the following conditions are met and If the data is in the correct state, it will be sent to step S2; otherwise, the AI ​​PTZ camera's self-cleaning device will be triggered and data will be collected again. The quality assessment formula is: ; ; in, The gray value of the pixel (x, y); For the Laplace operator; and These are the image width and height in pixels, respectively. Low-light suppression coefficient; This is the maximum grayscale value; A collection of low grayscale regions; It is an indicator function; Determine the threshold for dust obstruction and based on Dynamic adjustment.

[0029] The exposure time calculated in step S12 It directly determines the image sharpness index in step S14. The calculation accuracy of the two is subject to the following constraint relationship: ; in, Let be the coefficient of dust's impact on sharpness, and when Exceeding the motion blur threshold It automatically switches to electronic shutter mode to avoid motion blur caused by belt movement. For pixel physical size, This refers to the lens focal length.

[0030] Synchronization error in step S13 The third component of crystal oscillator temperature drift Illuminance in step S12 An indirect correlation is established through ambient temperature, specifically manifested as: when illuminance... When infrared illumination is triggered below 10 lux, the LED light source generates heat, causing the internal temperature of the camera to rise. This leads to Increase, the system reduces The bandwidth allocated to synchronization messages is used for compensation, and the bandwidth allocation formula is as follows: ; in, Reserve bandwidth for synchronization messages; This represents the total network bandwidth.

[0031] Rated conveying capacity of belt conveyor Design maximum conveying capacity Read the belt speed from the equipment's manufacturer's technical specifications. Real-time data acquisition (sampling frequency 1Hz) is achieved via the speed sensor of the conveyor belt control system; wind speed at the tunnel cross-section. An intrinsically safe anemometer (e.g., model GFW15) was installed at the center of the tunnel roof, and the average value was recorded every 5 minutes; ambient dust concentration was also recorded. A laser dust collector (e.g., model CCZ1000) was used to collect data at a camera placement height (2.5m above the ground), with a sampling interval of 1 minute; the camera's field of view was half-angle. It is determined by the technical parameters of the selected equipment.

[0032] Secondly, step S2 includes: S21. Using an intelligent recognition device integrated into the mining network camera, image deduplication based on dark channel prior theory is performed on multi-source video image data to eliminate image degradation caused by suspended dust particles, resulting in a preliminary deduplicated image; the global atmospheric light value in the dark channel prior theory... With pixel-level transmittance The calculation formula is: ; ; in, This represents the grayscale value of the pixel (x,y) in color channel c of the original image; The set of pixels for the entire image; It is a local neighborhood window centered at (x,y); The dust scattering coefficient is defined as [0.6, 0.8], which is used to adapt to the particle size distribution characteristics of dust in underground coal mines. S22. Perform contrast-limited adaptive histogram equalization enhancement on the preliminary dust removal image to improve the detail discernibility in low-light areas and suppress overexposure in highlight areas, resulting in a contrast-enhanced image; the contrast limiting threshold is based on the real-time dust concentration. Dynamic calculation, the calculation formula is as follows: ; in, Represents the local noise variance of the image; This represents the local average gray value; It is the dust environment compensation coefficient, and its value range is [1.5, 2.0] under strong dust conditions; The reference dust concentration value is used to normalize the dust influence intensity; S23. An adaptive median filtering algorithm is used to remove impulse noise from the contrast-enhanced image. The filter kernel size is automatically adjusted according to the real-time dust concentration level to maintain the integrity of edge information in high-dust and high-noise scenes, resulting in standardized image data. The adaptive adjustment rule for the filter kernel size is as follows: ; S24. After completing the above three preprocessing steps, the intelligent recognition device performs real-time quality assessment on the obtained standardized image data and calculates the overall image quality index. Only when Exceeding the preset quality index threshold The image is only sent to step S3 if the condition is met; otherwise, the parameters in steps S21 to S23 are adjusted and reprocessed. The formula for calculating the overall quality index is: ; in, and These represent the signal-to-noise ratios of the images before and after preprocessing, respectively. This represents the increase in contrast. This represents the maximum permissible enhancement level. Residual noise rate; Maximum tolerance noise rate; , , These are the weighting coefficients; feedback adjustments are made through iterative optimization. It converges to the maximum value.

[0033] Next, step S3 includes: S31. Standardized image data is input in batches into a pre-trained lightweight YOLOv5s improved AI recognition component. The lightweight YOLOv5s improved AI recognition component is a dust suppression attention module embedded at the end of the backbone network of a general lightweight YOLOv5s model. This module processes the input feature maps. Channel weighting calibration was performed to obtain the feature map after dust suppression. The calculation formula is as follows: ; in, and This is the weight matrix of the fully connected layer. This is the channel reduction ratio, and its value is 16. and It is the bias vector; This indicates a global average pooling operation; Use the Sigmoid activation function; It serves as a dust environment suppression factor and is adaptively adjusted within the [0.5, 1.0] range during training based on the proportion of dusty samples, used to dynamically reduce the feature weights of dusty areas; S32. Feature map based on dust suppression The system executes three subtasks in parallel: foreign object unloaded detection, belt tear identification, and trackless rubber-tired vehicle speed measurement and analysis. Each subtask shares a feature extraction layer but has an independent detection head. The foreign object unloaded detection employs a background difference and texture feature fusion mechanism, and its foreign object confidence level... The calculation formula is: ; in, The pixel value of the current frame; These are the pixel values ​​of the background model. This refers to local texture contrast features based on the gray-level co-occurrence matrix; To combine weights and their values ; Meanwhile, belt tear identification employs a multi-scale edge gradient direction histogram matching algorithm, which measures the tear damage quantification value. The calculation formula is: ; in, The number of detection sub-blocks to divide the belt surface; Let be the real-time gradient vector of the i-th subtask; This represents the gradient vector of the normal belt template at the corresponding position. This indicates element-wise division; The weighting coefficient for the i-th subtask; Let be the dust occlusion factor for the i-th subtask; Let be the dust attenuation coefficient for the i-th subtask. Used to reduce false detection rates in high-dust areas. Let be the real-time dust concentration of the i-th sub-block; Furthermore, the speed measurement analysis of trackless rubber-tired vehicles employs an improved dense optical flow method, and its vehicle speed... The calculation formula is: ; in, Vehicle center of gravity coordinates The optical flow vector at that location; Image spatial resolution; The tracking efficiency loss factor under dust obstruction. ; The tracking efficiency loss coefficient is used to quantify the rate of decrease in tracking efficiency of feature points caused by a unit increase in dust concentration. The camera's frame rate; This is the physical length calibration value corresponding to a unit pixel; The angle between the vehicle's direction of travel and the optical axis; S33. The outputs of the three sub-tasks are spatiotemporally aligned and weighted by confidence in the detection result fusion layer to generate a unified primary anomaly detection result. The fusion formula is: ; in, This indicates a spatiotemporal nonmaximum suppression fusion operation; For the kth type of exception, ; The coordinates of the k-th anomaly; is the original confidence level for the k-th type of anomaly; The environmental-operating condition joint compensation coefficient is... ; As a task type sensitivity factor, it is used to adaptively improve detection reliability in dusty environments and high-speed belt operation conditions; The maximum operating speed preset for the belt conveyor.

[0034] Then, step S4 includes: S41. The initial anomaly identification results... Transmitted via an intrinsically safe fiber optic ring network to a parallel management server cluster built on the Raft consensus protocol. The cluster contains at least three nodes, each managing its real-time processing load. Memory usage and network latency The master node is dynamically elected, and log data is synchronously replicated to ensure data consistency. Master node election weights are also specified. The calculation formula is: ; in, , , These are the preset CPU load limit, memory limit, and network latency limit; , , This refers to the election weighting coefficient; the election weighting coefficient is dynamically adjusted based on the reliability of the downhole environment, for example, when the dust concentration... Above 150 Time increases Prioritize selecting nodes with stable network connectivity; S42. Perform multi-source heterogeneous data fusion analysis on the master node to analyze the preliminary anomaly identification results from different AI recognition components. Spatiotemporal alignment and confidence-weighted fusion are performed. An improved DS evidence theory algorithm is used to handle conflicts and redundancies in the results of foreign object detection, belt tear identification, and trackless rubber-tired vehicle speed measurement. The confidence of the fused abnormal events is then calculated. The calculation formula is: ; Among them, A is the final abnormal event after fusion; B is the foreign object detection evidence; C is the belt tear evidence; and D is the vehicle speed measurement evidence. , , The confidence levels of evidence for foreign object detection, belt tearing, and abnormal vehicle speed are respectively: The dynamic fusion weights based on spatiotemporal distance and sensor reliability are calculated as follows: ; in, , These represent the temporal and spatial deviations of the three types of test results; , For time air conditioning and parameters; The sensor reliability of the k-th AI recognition component; This represents the global conflict coefficient. For environmental reliability factor, ; , This is the environmental weighting coefficient; Gas concentration; The preset maximum gas concentration; This is the preset maximum dust concentration; S43, Based on the confidence level of fused anomaly events Combined with abnormal location information and the real-time operating speed of the belt Dynamically calculate the hazard level It generates a linkage control instruction package containing equipment control logic, alarm priority, and evidence storage strategy, and determines the hazard level. The calculation formula is: ; in, and These are the high confidence threshold and the medium confidence threshold, respectively. The critical safe speed for the belt; This is a quantification of tear damage; This represents the critical threshold for tear damage. The generation of linkage control commands follows a hazard level mapping table. Level I commands include immediate stop, full-area audible and visual alarms, high-definition video recording lock, and multiple backup storage. Level II commands include deceleration, area-specific alarms, and standard video recording storage. Level III commands only record logs and periodic inspection reminders. The mapping table is obtained through training and optimization using a downhole historical accident database.

[0035] In addition, step S5 includes: S51. After the client performs CRC check and digital signature verification on the received linkage control command packet, it parses and extracts the danger level. Abnormal location coordinates Evidence locking time window and alarm priority weight And query the local configuration database to get the number of workers currently working in the roadway. Compared with real-time background noise in decibels ; S52. Hazard level obtained from analysis Dynamically calculate the alarm volume of the sound system and speaker. Alarm frequency and duration of sound and light The sound system is installed in the underground control room, while loudspeakers are deployed along the conveyor belt tunnels. Together, they achieve zoned coverage alarms. Volume calculation follows the principle that the higher the hazard level, the louder the volume, but it must not exceed the rated power of the explosion-proof equipment. The alarm volume... The calculation formula is: ; in, The rated maximum volume for explosion-proof speakers; Used as the baseline alarm volume; This is the hazard level increment coefficient; Numerical mapping of hazard levels (Level I = 3, Level II = 2, Level III = 1); For reference number of workers; Background noise as reference; speaker trigger delay time Based on the distance from the abnormal location The calculation formula is as follows: ; in, The speed at which sound travels in the alleyway environment; This is the network transmission delay coefficient; For client command processing time; S53. While the client is activating the audible and visual alarm, it locks the time window based on the evidence. Automatically capture and cache multi-source video image data and standardized image data, depending on the hazard level. When an anomaly is reached at Level I or II, the external explosion-proof mining printer is triggered to perform evidence solidification output. The printed content includes the anomaly type, location coordinates, occurrence time, confidence level value, and keyframe images; the number of copies printed... The paper size is determined by the hazard level and archiving requirements, and the calculation formula is as follows: ; ; in, For abnormal confidence levels; , These are the high confidence threshold and the medium confidence threshold, respectively. After printing, the client sends the printed log and alarm execution status back to the management server cluster via heartbeat messages to complete the alarm closed-loop control.

[0036] Finally, step S6 includes: S61. After receiving the linkage control command packet, the client determines the risk level based on the severity of the hazard. Data types Based on downhole storage environment parameters, dynamically calculate the storage priority weight of various types of data. Retention period and the number of redundant copies Develop differentiated storage strategies and prioritize storage. The calculation formula is: ; in, Numerical mapping of hazard levels (Level I = 3, Level II = 2, Level III = 1); Quantize the data types (video=3, image=2, structured data=1); , , As a weighting coefficient; when the environmental dust concentration More than 150 Time increases Prioritize the storage of critical data; S62. Based on the differentiated storage strategy, perform adaptive compression coding based on content importance on multi-source video image data, and automatically increase the coding bitrate when the danger level is Level I or Level II. Interval with keyframe For periods without abnormalities, low bitrate storage is used; compression ratio The adjustment is dynamically based on storage priority weights and available bandwidth, and the adjustment formula is as follows: ; ; in, The preset base compression ratio; It is the hazard level influence coefficient, and its value ranges from [0.5, 1.5]. The maximum storage priority weight; This is the time-varying bandwidth efficiency factor; Let t be the network load rate at time t; This represents the maximum network bandwidth capacity. It is the bandwidth attenuation coefficient and its value ranges from [0.01, 0.05]. S63. Separate the compressed video stream, standardized image data, primary anomaly detection results, and linkage control instructions by timestamp. The components are aligned and packaged into a unified storage unit, and written to an explosion-proof certified NAS storage system via the iSCSI protocol. The storage system employs a RAID 5 redundancy architecture and is configured with a hot spare disk. Parity checks across disks are calculated before writing. The verification formula is: ; in, Store block data for the i-th data disk; This is a bitwise XOR operation; This represents the total number of disks in the RAID group; immediately after writing, a data integrity check is performed, calculating the checksum and comparing it with the original value. If they do not match, the checksum is adjusted based on the number of redundant replicas. The retransmission mechanism is triggered, with a maximum number of retransmissions allowed. The calculation formula is: ; in, This is the critical value for dust concentration. When this value is exceeded, the number of retransmissions is increased to ensure data reliability. After a successful retransmission, the storage log is updated and synchronized to the parallel management server cluster.

[0037] In some embodiments, the operating status monitoring system 200 for the mining belt conveyor may include multiple functional modules composed of computer program segments. The computer programs for each program segment in the operating status monitoring system 200 may be stored in the memory of a computer device and executed by at least one processor to perform (see details). Figure 1 (Description) Monitoring function for the operating status of mining belt conveyors.

[0038] In this embodiment, the operation status monitoring system 200 of the mining belt conveyor can be divided into multiple functional modules according to its functions, such as... Figure 2As shown. The functional modules may include: a data acquisition module 210, an image preprocessing module 220, an AI recognition and analysis module 230, a data fusion and decision-making module 240, an alarm and evidence output module 250, and a data storage module 260. The module referred to in this invention is a series of computer program segments that can be executed by at least one processor and perform a fixed function, and are stored in memory. In this embodiment, the functions of each module will be described in detail in subsequent embodiments.

[0039] Specifically, the data acquisition module 210 includes several AI shooting components and a mining network camera distributed at intervals along the longitudinal direction of the belt conveyor, used to simultaneously acquire multi-source video image data of the belt surface, transported materials, and trackless rubber-wheeled vehicles on the belt conveyor; the image preprocessing module 220, communicatively connected to the data acquisition module, includes an intelligent recognition device, used to perform anti-dust interference preprocessing on the multi-source video image data to obtain standardized image data; the AI ​​recognition and analysis module 230, communicatively connected to the image preprocessing module, includes a pre-trained AI recognition component, used to perform parallel foreign object unloaded detection, belt tear recognition, and trackless rubber-wheeled vehicle speed measurement analysis on the standardized image data, and generate preliminary anomaly recognition results; the data fusion and decision module 240... The system includes a management server connected in parallel to the AI ​​recognition and analysis module. This server performs data fusion analysis on the initial anomaly recognition results and generates linkage control commands containing anomaly type, location, and hazard level. An alarm and evidence output module 250, connected in parallel to the data fusion and decision module, includes a client and the audio, loudspeaker, and printer driven by it. This module executes graded audio-visual alarms and evidence solidification output according to the linkage control commands. A data storage module 260, connected in parallel to the data acquisition module, image preprocessing module, AI recognition and analysis module, and data fusion and decision module, includes a storage system for synchronously storing multi-source video image data, standardized image data, initial anomaly recognition results, and linkage control commands.

[0040] Figure 3 This is a schematic diagram of a terminal 300 provided in an embodiment of the present invention. The terminal 300 can be used to execute the method for monitoring the operating status of a mining belt conveyor provided in an embodiment of the present invention.

[0041] The terminal 300 may include a processor 310, a memory 320, and a communication module 330. These components communicate via one or more buses. Those skilled in the art will understand that the server structure shown in the figure does not constitute a limitation of the present invention. It may be a bus topology or a star topology, and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0042] The memory 320 can be used to store the execution instructions of the processor 310. The memory 320 can be implemented by any type of volatile or non-volatile memory terminal or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. When the execution instructions in the memory 320 are executed by the processor 310, the terminal 300 is able to perform some or all of the steps in the above method embodiments.

[0043] The processor 310 serves as the control center of the storage terminal, connecting various parts of the electronic terminal via various interfaces and lines. It executes software programs and / or modules stored in the memory 320, and calls data stored in the memory to perform various functions of the electronic terminal and / or process data. The processor can be composed of integrated circuits (ICs), such as a single packaged IC or multiple packaged ICs with the same or different functions connected together. For example, the processor 310 may consist only of a central processing unit (CPU). In this embodiment of the invention, the CPU may have a single processing core or include multiple processing cores.

[0044] The communication module 330 is used to establish a communication channel, enabling the storage terminal to communicate with other terminals. It receives user data sent by other terminals or sends user data to other terminals.

[0045] The present invention also provides a computer storage medium, wherein the computer storage medium may store a program, which, when executed, may include some or all of the steps provided in the embodiments of the present invention. The storage medium may be a magnetic disk, an optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0046] Those skilled in the art will clearly understand that the techniques in the embodiments of the present invention can be implemented using software plus necessary general-purpose hardware platforms. Based on this understanding, the technical solutions in the embodiments of the present invention, or the parts that contribute to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium such as a USB flash drive, mobile hard drive, read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk, or other media capable of storing program code. It includes several instructions to cause a computer terminal (which may be a personal computer, server, or a second terminal, network terminal, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention.

[0047] The same or similar parts between the various embodiments in this specification can be referred to mutually. In particular, the terminal embodiments are basically similar to the method embodiments, so the description is relatively simple, and the relevant parts can be referred to the description in the method embodiments.

[0048] In the embodiments provided by this invention, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative. For instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between systems or modules may be electrical, mechanical, or other forms.

[0049] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0050] In addition, the functional modules in the various embodiments of the present invention can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0051] Although the present invention has been described in detail with reference to the accompanying drawings and preferred embodiments, the present invention is not limited thereto. Various equivalent modifications or substitutions can be made to the embodiments of the present invention by those skilled in the art without departing from the spirit and essence of the invention, and such modifications or substitutions should all be within the scope of the present invention. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should also be covered within the protection scope of the present invention.

Claims

1. A method of monitoring the operating state of a mine belt conveyor, characterized in that include: S1. By using several AI shooting components and mining network cameras distributed at intervals along the longitudinal direction of the belt conveyor, multi-source video image data of the belt surface, the transported materials and the trackless rubber-wheeled vehicle are collected simultaneously. S2. The multi-source video image data is preprocessed to resist dust interference by an intelligent recognition device to obtain standardized image data; S3. Input standardized image data into the pre-trained AI recognition component, and perform parallel foreign object unloaded detection, belt tear recognition and trackless rubber-tired vehicle speed measurement and analysis to generate preliminary anomaly recognition results; S4. Transmit the initial anomaly identification results to several parallel management servers for data fusion analysis, and generate linkage control instructions containing anomaly type, location and hazard level. S5. Send the linkage control command to the client to drive the external audio system, loudspeaker, and printer to perform graded sound and light alarms and evidence solidification output; S6. Synchronously store multi-source video image data, standardized image data, primary anomaly recognition results, and linkage control commands to the storage system.

2. The method of monitoring the operating state of a mine belt conveyor according to claim 1, characterized in that, Step S1 includes: S11, acquire the rated conveying capacity of the belt conveyor, the running speed of the belt in the belt conveyor and the wind speed of the roadway section where the belt conveyor is located, and according to the acquired rated conveying capacity of the belt conveyor, the running speed of the belt and the wind speed of the roadway section, dynamically optimize the interval along the longitudinal direction of the belt conveyor deploy several AI shooting components and mine network cameras; dynamically optimize the interval The calculation formula is: ； wherein, is the rated conveying capacity of the belt conveyor, unit: t / h; is the running speed of the belt in the belt conveyor, unit: m / s; is the air speed of the roadway section where the belt conveyor is located, unit: m / s; is the maximum allowable response delay time of the system, unit: s; is the safety factor and is not less than 1.5; is the designed maximum conveying capacity of the belt conveyor; is the load influence coefficient; is the dust deposition height threshold, unit: m; is the environmental dust concentration, unit: ; is the half angle of the camera field of view, unit: rad; S12, at the start of each acquisition cycle, according to the real-time monitored illumination and ambient dust concentration , the exposure time of the AI shooting component is adaptively adjusted and the gain of the image sensor in the AI shooting component ; the adaptive adjustment formula is: ； ； wherein, is a pre-set image sensor full well capacity; is a dust attenuation factor function; is a pixel photosensitivity in V / lux-s; and is an adjustment coefficient; is a reference illuminance; is a reference dust concentration; is a reference gain value; S13, all AI shooting components are clock-synchronized with the mine network camera through IEEE 1588v2 time protocol, and absolute time stamps based on GPS / Beidou dual-mode time service are embedded in each video frame ; S14, after the collection is completed, real-time quality assessment is performed on each frame of multi-source video image data, and image definition index is calculated With dust shielding rate , only when and are met, the data is sent to step S2, otherwise, re-collection; the quality assessment formula is: ； ； wherein, is a gray value of a pixel point (x, y); is a Laplacian operator; and are respectively a width and height pixel number of an image; is a low-illumination suppression coefficient; is a maximum gray value; is a low-gray region set; is an indicator function; is a dust shielding determination threshold.

3. The method of monitoring the operating state of a mine belt conveyor according to claim 2, characterized in that, Step S2 includes: S21. Through the intelligent recognition device integrated inside the mining network camera, image dust removal processing based on dark channel prior theory is performed on multi-source video image data to obtain a preliminary dust-removed image; S22, contrast-limited adaptive histogram equalization is performed on the preliminary dust-removed image to obtain a contrast-enhanced image; the contrast limit threshold is determined according to the real-time dust concentration Dynamic calculation, the calculation formula is: ； in, Represents the local noise variance of the image; This represents the local average gray value; This is the dust environmental compensation coefficient; S23. Adaptive median filtering algorithm is used to remove impulse noise from contrast-enhanced images; S24. After completing the above three preprocessing steps, the intelligent recognition device performs real-time quality assessment on the obtained standardized image data and calculates the overall image quality index. Only when Exceeding the preset quality index threshold The image is only sent to step S3 if the condition is met; otherwise, the parameters in steps S21 to S23 are adjusted and reprocessed. The formula for calculating the overall quality index is: ； in, and These represent the signal-to-noise ratios of the images before and after preprocessing, respectively. This represents the increase in contrast. This represents the maximum permissible enhancement level. Residual noise rate; Maximum tolerance noise rate; , , These are the weighting coefficients; feedback adjustments are made through iterative optimization. It converges to the maximum value.

4. The method for monitoring the operating status of a mining belt conveyor according to claim 3, characterized in that, Step S3 includes: S31. Standardized image data is input in batches into a pre-trained lightweight YOLOv5s improved AI recognition component. The lightweight YOLOv5s improved AI recognition component is a dust suppression attention module embedded at the end of the backbone network of a general lightweight YOLOv5s model. This module processes the input feature maps. Channel weighting calibration was performed to obtain the feature map after dust suppression. The calculation formula is as follows: ； in, and This is the weight matrix of the fully connected layer. This refers to the channel reduction ratio; and It is the bias vector; This indicates a global average pooling operation; Use the Sigmoid activation function; It acts as a dust-inhibiting factor. S32. Feature map based on dust suppression The system executes three subtasks in parallel: foreign object unloaded detection, belt tear identification, and trackless rubber-tired vehicle speed measurement and analysis. Each subtask shares a feature extraction layer but has an independent detection head. The foreign object unloaded detection employs a background difference and texture feature fusion mechanism, and its foreign object confidence level... The calculation formula is: ； in, The pixel value of the current frame; These are the pixel values ​​of the background model. This refers to local texture contrast features based on the gray-level co-occurrence matrix; For fusion weights; Meanwhile, belt tear identification employs a multi-scale edge gradient direction histogram matching algorithm, which measures the tear damage quantification value. The calculation formula is: ； in, The number of detection sub-blocks to divide the belt surface; Let be the real-time gradient vector of the i-th subtask; This represents the gradient vector of the normal belt template at the corresponding position. This indicates element-wise division; The weighting coefficient for the i-th subtask; Let be the dust occlusion factor for the i-th subtask; Let be the dust attenuation coefficient for the i-th subtask; Furthermore, the speed measurement analysis of trackless rubber-tired vehicles employs an improved dense optical flow method, and its vehicle speed... The calculation formula is: ； in, Vehicle center of gravity coordinates The optical flow vector at that location; Image spatial resolution; The tracking efficiency loss factor under dust obstruction. ; To track the efficiency loss coefficient; The camera's frame rate; This is the physical length calibration value corresponding to a unit pixel; The angle between the vehicle's direction of travel and the optical axis; S33. The outputs of the three sub-tasks are spatiotemporally aligned and weighted by confidence in the detection result fusion layer to generate a unified primary anomaly detection result. The fusion formula is: ； in, This indicates a spatiotemporal nonmaximum suppression fusion operation; It is the kth type of exception; The coordinates of the k-th anomaly; is the original confidence level for the k-th type of anomaly; The environmental-operating condition joint compensation coefficient is... ; As a task type sensitivity factor, it is used to adaptively improve detection reliability in dusty environments and high-speed belt operation conditions; The maximum operating speed preset for the belt conveyor.

5. The method for monitoring the operating status of a mining belt conveyor according to claim 4, characterized in that, Step S4 includes: S41. The initial anomaly identification results... Transmitted via an intrinsically safe fiber optic ring network to a parallel management server cluster built on the Raft consensus protocol. The cluster contains at least three nodes, each managing its real-time processing load. Memory usage and network latency Dynamic election of master nodes, master node election weight The calculation formula is: ； in, , , These are the preset CPU load limit, memory limit, and network latency limit; , , These are the election weighting coefficients; S42. Perform multi-source heterogeneous data fusion analysis on the master node to analyze the preliminary anomaly identification results from different AI recognition components. Spatiotemporal alignment and confidence-weighted fusion are performed. An improved DS evidence theory algorithm is used to handle conflicts and redundancies in the results of foreign object detection, belt tear identification, and trackless rubber-tired vehicle speed measurement. The confidence of the fused abnormal events is then calculated. The calculation formula is: ； Among them, A is the final abnormal event after fusion; B is the foreign object detection evidence; C is the belt tear evidence; and D is the vehicle speed measurement evidence. , , The confidence levels of evidence for foreign object detection, belt tearing, and abnormal vehicle speed are respectively: The dynamic fusion weights based on spatiotemporal distance and sensor reliability are calculated as follows: ； in, , These represent the temporal and spatial deviations of the three types of test results; , For time air conditioning and parameters; The sensor reliability of the k-th AI recognition component; This represents the global conflict coefficient. For environmental reliability factor, ; , This is the environmental weighting coefficient; Gas concentration; The preset maximum gas concentration; This is the preset maximum dust concentration; S43, Based on the confidence level of fused anomaly events Combined with abnormal location information and the real-time operating speed of the belt Dynamically calculate the hazard level It generates a linkage control instruction package containing equipment control logic, alarm priority, and evidence storage strategy, and determines the hazard level. The calculation formula is: ； in, and These are the high confidence threshold and the medium confidence threshold, respectively. The critical safe speed for the belt; This is a quantification of tear damage; This represents the critical threshold for tear damage.

6. The method for monitoring the operating status of a mining belt conveyor according to claim 5, characterized in that, Step S5 includes: S51. After the client performs CRC check and digital signature verification on the received linkage control command packet, it parses and extracts the danger level. Abnormal location coordinates Evidence locking time window and alarm priority weight And query the local configuration database to get the number of workers currently working in the roadway. Compared with real-time background noise in decibels ; S52. Hazard level obtained from analysis Dynamically calculate the alarm volume of the sound system and speaker. Alarm frequency and duration of sound and light The sound system is installed in the underground control room, while loudspeakers are deployed along the conveyor belt tunnels. Together, they achieve zoned coverage alarms; the alarm volume... The calculation formula is: ； in, The rated maximum volume for explosion-proof speakers; Used as the baseline alarm volume; This is the hazard level increment coefficient; Numerical mapping of hazard levels; For reference number of workers; Background noise as reference; speaker trigger delay time Based on the distance from the abnormal location The calculation formula is as follows: ； in, The speed at which sound travels in the alleyway environment; This is the network transmission delay coefficient; For client command processing time; S53. While the client is activating the audible and visual alarm, it locks the time window based on the evidence. Automatically capture and cache multi-source video image data and standardized image data, depending on the hazard level. When Level I or Level II is reached, the external mining explosion-proof printer is triggered to perform evidence solidification output.

7. The method for monitoring the operating status of a mining belt conveyor according to claim 6, characterized in that, Step S6 includes: S61. After receiving the linkage control command packet, the client determines the risk level based on the severity of the hazard. Data types Based on downhole storage environment parameters, dynamically calculate the storage priority weight of various types of data. Retention period and the number of redundant copies Develop differentiated storage strategies and prioritize storage. The calculation formula is: ； in, Numerical mapping of hazard levels; Quantize values ​​for data types; , , As a weighting coefficient; when the environmental dust concentration More than 150 Time increases Prioritize the storage of critical data; S62. Based on the differentiated storage strategy, perform adaptive compression coding based on content importance on multi-source video image data, and automatically increase the coding bitrate when the danger level is Level I or Level II. Interval with keyframe For periods without abnormalities, low bitrate storage is used; compression ratio The adjustment is dynamically based on storage priority weights and available bandwidth, and the adjustment formula is as follows: ； ； in, The preset base compression ratio; This is the hazard level influence coefficient; The maximum storage priority weight; This is the time-varying bandwidth efficiency factor; Let t be the network load rate at time t; This represents the maximum network bandwidth capacity. This is the bandwidth attenuation coefficient; S63. Separate the compressed video stream, standardized image data, primary anomaly detection results, and linkage control instructions by timestamp. The components are aligned and packaged into a unified storage unit, and written to an explosion-proof certified NAS storage system via the iSCSI protocol.

8. A monitoring system for the operating status of a mining belt conveyor, characterized in that, include: The data acquisition module includes several AI shooting components and mining network cameras distributed at intervals along the longitudinal direction of the belt conveyor, used to simultaneously acquire multi-source video image data of the belt surface, the transported materials and the trackless rubber-wheeled vehicle on the belt conveyor; The image preprocessing module, which is connected in communication with the data acquisition module, includes an intelligent recognition device for performing anti-dust interference preprocessing on multi-source video image data to obtain standardized image data. The AI ​​recognition and analysis module communicates with the image preprocessing module and includes a pre-trained AI recognition component. This component is used to perform parallel foreign object detection, belt tear recognition, and trackless rubber-tired vehicle speed measurement and analysis on standardized image data, and generate preliminary anomaly recognition results. The data fusion and decision-making module communicates with the AI ​​recognition and analysis module, including several management servers set in parallel, which are used to perform data fusion analysis on the initial anomaly recognition results and generate linkage control instructions containing anomaly type, location and danger level. The alarm and evidence output module, which communicates with the data fusion and decision module, includes a client and the audio, loudspeaker and printer driven by it, and is used to execute graded sound and light alarms and evidence solidification output according to the linkage control command. The data storage module is communicatively connected to the data acquisition module, image preprocessing module, AI recognition and analysis module, and data fusion and decision-making module, and includes a storage system for synchronously storing multi-source video image data, standardized image data, primary anomaly recognition results, and linkage control commands.

9. A terminal, characterized in that, include: processor; Memory used to store the processor's execution instructions; The processor is configured to perform the method of any one of claims 1-7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the program is executed by the processor, it implements the method as described in any one of claims 1-7.

Images