Coal mine anti-impact pressure relief drill rod identification algorithm and system

Abstract

The invention relates to the technical field of coal mine safety production and intelligent monitoring, in particular to a coal mine anti-impact pressure relief drill rod recognition algorithm and system, which comprises image acquisition equipment, communication transmission equipment, data processing equipment and data storage equipment. Video data in the coal mine underground anti-impact pressure-relief drill withdrawal process are collected in real time through an intrinsic safety type camera, and the states of a drill rod, a drilling machine and personnel in images are recognized and analyzed through a target detection algorithm and a video action sequence detection algorithm. The target detection model is responsible for recognizing the connection and disconnection states of the drilling machine and the drilling rods, and the video action sequence detection model dynamically counts the number of the drilling rods in the drilling retreating process by analyzing the movement direction of the drilling machine and the operation behaviors of personnel in continuous videos. According to the method, the detection precision and efficiency in the anti-impact pressure relief process are improved, the method adapts to the complex underground environment, the problems of missing detection and false detection caused by human factors are reduced, and the safety and compliance of coal mine operation are ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of safe production and intelligent monitoring of coal mines, and in particular to an algorithm and system for identifying a drill rod for preventing and unloading pressure in coal mines. Background Art

[0002] With the continuous increase in the depth and intensity of coal mining, the frequency and hazards of rock burst are gradually increasing, which seriously threatens the safety of underground workers and affects the production efficiency of coal mines. Therefore, anti-burst pressure relief has become one of the important links in coal mine safety production. In the process of anti-burst pressure relief, accurately detecting the number of drill rods withdrawn is the key to ensuring the qualified pressure relief operation. However, the existing anti-burst pressure relief drill rod detection method still has many shortcomings, which affects the operation efficiency and accuracy.

[0003] At present, the detection of the number of drill rods withdrawn for anti-bumping pressure relief is mainly carried out in the following two ways:

[0004] Manual review: When underground workers are performing the anti-blowout pressure relief rod withdrawal operation, they record the entire process through a camera and manually copy the video data to the surface. Relevant staff manually record the number of drill rods withdrawn based on the video and calculate the total length of the anti-blowout pressure relief rod withdrawal. This method is not only inefficient, but also relies on manual judgment, which is prone to recording errors or omissions, resulting in unqualified anti-blowout pressure relief.

[0005] Intelligent identification: Using motion recognition technology, the number of drill rods is counted by detecting the action of personnel removing the drill rods. This method can partially replace manual review, but due to the complex underground environment, the video clarity and picture integrity may be affected, especially the actions of personnel are easily confused with other operating actions, resulting in missed detection or false detection, thus affecting the accuracy of drill rod identification. In addition, when the drilling rig is deep, it is common for the drilling rig to adjust the rod withdrawal multiple times, which may cause multiple repeated counting, further reducing the accuracy of detection.

[0006] In view of the limitations of existing methods, especially the detection accuracy problem in complex environments, this paper proposes an anti-bumping and pressure-relief drill rod recognition algorithm and system based on target detection and video action sequence detection. The system uses artificial intelligence technology, combined with the connection status of the drilling rig and the drill rod and the rod-removing action of the personnel, and automatically detects and counts the number of rods withdrawn through multi-dimensional analysis of video data, thereby improving the accuracy and work efficiency of detection and avoiding missed detection and false detection problems caused by factors such as poor lighting in the underground environment and personnel occlusion. Summary of the invention

[0007] Aiming at the problems of low efficiency, false detection and missed detection in the existing methods for identifying drill rods for anti-blowout pressure relief in coal mines, the present invention proposes an anti-blowout pressure relief drill rod identification algorithm and system based on target detection and video action sequence detection. The system realizes automatic detection and statistics of the number of drill rods in the process of anti-blowout pressure relief by combining artificial intelligence technology, multi-algorithm fusion and image processing methods, which significantly improves the detection accuracy and operation efficiency.

[0008] The purpose of the present invention is to provide a system that can accurately identify the number of drill rods during the anti-blowout pressure relief process in a complex underground environment, so as to solve the problems of time-consuming and labor-intensive manual detection and large errors in intelligent detection in traditional methods. The present invention collects image and video data during the drill withdrawal process, and uses the target detection model and the video action sequence detection model for comprehensive analysis, so as to count the number of drill rods in real time and accurately.

[0009] To achieve the above object, the present invention provides an anti-bumping pressure-relief drill rod identification system based on target detection and video action sequence detection, which mainly includes the following parts:

[0010] Image acquisition equipment

[0011] The image acquisition equipment uses an intrinsically safe camera that can operate in the harsh environment of coal mines and collect high-definition images and video data. The camera equipment is installed about 2 meters above the drilling rig to ensure that the entire drilling rig and drill rod are always in the picture, and can clearly capture the movement of the drill rod and the operator's operation. The camera resolution is 1920×1080 and above, the frame rate is 25 frames per second, and the total bit rate is 4096kbps and above to ensure the quality of video data and reduce detection errors caused by insufficient light or occlusion.

[0012] Communication transmission equipment

[0013] Communication transmission equipment includes 5G, industrial WiFi, USB3.0, Ethernet and other wired or wireless transmission methods, which transmit the collected image and video data to data processing equipment in real time to ensure the timeliness and stability of the data.

[0014] Data processing equipment

[0015] The data processing device is the core part of this system. It uses an AI edge computing server with sufficient computing power to perform real-time image processing and target detection. The data processing device uses the target detection model and the video action sequence detection model to analyze the collected images and videos.

[0016] The target detection model is based on the YOLO model of deep learning, which can identify the drill rods, drilling rigs, personnel and their action status in real time. The number of rods withdrawn can be determined by calculating the connection and disconnection status of the drilling rig and the drill rod, and the action of the personnel taking the drill rod.

[0017] The video action sequence detection model is based on the ViViT model. It dynamically detects the drilling process by analyzing the movement direction of the drilling rig, the connection status of the drill rod, and the operation of the personnel in the continuous video clips. The model uses the Transformer self-attention mechanism, which can simultaneously analyze the characteristics of the spatial and temporal dimensions to further improve the detection accuracy.

[0018] Data storage devices

[0019] The data storage device is used to store the identified images, videos and test results of the number of drill rods. The test data will be uploaded to the AI ​​application platform in real time for further analysis and confirmation by management personnel.

[0020] Algorithm Fusion and Counting Logic

[0021] The system of the present invention counts the number of drill rod withdrawals by combining a target detection algorithm with a video action sequence detection algorithm. In actual operation, the target detection model is first used to identify the drilling rig and personnel in the image, and the number of rod withdrawals is preliminarily determined by calculating the connection and disconnection status of the drilling rig and the drill rod; then, the video action sequence detection model is used to analyze the continuous drilling withdrawal process and the operating status of the personnel in combination with the time dimension to further confirm the number of rod withdrawals. The two algorithms complement each other. When the detection result of one of the algorithms has a higher accuracy, its result is used as the final number of rod withdrawals.

[0022] Online and offline detection modes

[0023] The system supports both online and offline detection modes. In online mode, the image acquisition device monitors the anti-collision and pressure relief process in real time, and the data processing device infers and generates detection results in real time. In offline mode, the video data is pre-recorded and uploaded to the AI ​​platform, and the data processing device then performs inference calculations. Whether in online or offline mode, the system will generate relevant alarm information and upload it to the AI ​​platform for secondary review to ensure the compliance of the anti-collision and pressure relief process.

[0024] Automatic alarm function

[0025] When the system detects that the number of anti-bumping pressure-removal rods does not meet the preset requirements, it automatically generates an alarm record and notifies the management personnel through the AI ​​platform for review and processing. This not only reduces human errors, but also improves overall safety.

[0026] The beneficial effects of the present invention are as follows: a target detection model and a video action sequence detection model are used to integrate multiple calculations, and the drill rod counting process and personnel actions, the movement state of the drilling rig, and the connection state between the drill rods are comprehensively analyzed and calculated to judge the number of drill rods. Compared with the current anti-bumping and pressure-unloading detection method in machine vision, it can cope with the multi-angle transformation problem in the rod withdrawal process and the missed detection caused by the light transformation problem. While improving the accuracy of drill rod counting, it can have better robustness and generalization than a single statistical algorithm. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0028] Figure 1 It is a schematic diagram of the implementation process of an embodiment of the present invention;

[0029] Figure 2 Schematic diagram of drilling rig motion state calculation according to an embodiment of the present invention. DETAILED DESCRIPTION

[0030] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. At the same time, it is explained here that in order to make the embodiments more detailed, the following embodiments are the best and preferred embodiments, and those skilled in the art may also adopt other alternatives to implement some known technologies; and the accompanying drawings are only for more specific description of the embodiments, and are not intended to specifically limit the present invention.

[0031] It should be noted that the references to "one embodiment", "an embodiment", "an exemplary embodiment", "some embodiments" and the like in the specification indicate that the embodiments described may include specific features, structures or characteristics, but not every embodiment may include the specific features, structures or characteristics. In addition, when a specific feature, structure or characteristic is described in conjunction with an embodiment, it should be within the knowledge of a person skilled in the art to implement such feature, structure or characteristic in conjunction with other embodiments (whether or not explicitly described).

[0032] In general, a term can be understood, at least in part, from its use in context. For example, depending, at least in part, on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in the singular sense, or can be used to describe a combination of features, structures, or characteristics in the plural sense. Additionally, the term "based on" can be understood as not necessarily intended to convey an exclusive set of factors, but can instead, depending, at least in part, on the context, allow for the presence of other factors that are not necessarily explicitly described.

[0033] Example 1

[0034] The present invention provides a method and device for identifying the number of anti-shock unloading drill rods based on target detection. The invention uses an intrinsically safe camera to pre-collect relevant data images and videos within the anti-shock unloading drill rod withdrawal time period, and establish an anti-shock unloading drill rod withdrawal stage data set. The target detection model is trained using images in the image data set, and the action sequence detection model is established using the video of the drill withdrawal process. Among them, the target detection model identifies the personnel, drilling rigs and drill rods in the image, and uses the actions of the personnel holding the drill rods in the image, the analysis of the drilling rig operation status, and the analysis of the connection status between the drilling rig and the drill rod to comprehensively judge and analyze the anti-shock unloading drilling process; the action sequence detection model detects the drill withdrawal process in the anti-shock unloading process in a continuous time segment, the personnel taking the drill rod process and the drilling rig operation process, and finally merges and sums the number of drill withdrawals each time, and comprehensively counts whether the anti-shock unloading process is qualified. If the drilling process is qualified, an anti-shock unloading management record is generated, and if it is unqualified, a relevant alarm record is generated and uploaded to the AI ​​application platform.

[0035] 1. Image acquisition and annotation requirements

[0036] Video acquisition and recording requires that the drilling rig is in the picture throughout the whole process. It is recommended that the drilling rig occupies about 80% of the entire picture, and the rod-removing worker faces the camera directly without blocking the drill rod. The camera angle requires shooting from top to bottom. The camera is located between the drilling rig and the borehole, and is 2 meters vertically away from the drill rod on the drilling rig. The video resolution requires pixels of 1920*1080 and above, a frame rate of 25 frames, and a total bit rate of 4096kbps and above. The on-site lighting requires that the drill rod and the moving parts of the drilling rig can be clearly seen. It is recommended to add diffuse light to fill in the light. To avoid obstruction, the drill rod must be visible, the process of removing the drill rod can be seen clearly, and the drill bit on the last rod can be seen without obstruction.

[0037] Target detection dataset: The anti-bumping and pressure relief target detection dataset needs to include relevant actions of the drill withdrawal process, such as the state of a worker holding a drill rod, the movement state of an empty drill rig and a drill rig with a rod, etc. It is annotated using the deep learning image labeling software CVAT. The annotation type is a rectangular box, and the annotation categories are drill rod, drill rig, personnel, and personnel holding a drill rod.

[0038] 2. Training the object detection model

[0039] Based on the relevant requirements in step one, a detection data set is collected, divided and the target detection model is trained. The data collected for anti-blowout pressure relief is divided into a training set and a validation set. Since the angle of the anti-blowout pressure relief drilling rig is not fixed during drilling, a simulated angle rotation is performed to further improve the robustness of the data set. The simulated rotation is to rotate the image at multiple angles such as 15°, 30°, 45°, and 90°. The actual anti-blowout pressure relief drilling rotation is simulated through image rotation, and then the rotated image and the original image are added to the model training set, that is, image enhancement is performed through preprocessing. At the same time, due to the complexity of the light underground, it also has a certain degree of anti-interference with the underground glare and the occlusion of workers when taking the drill rod.

[0040] 3. Target detection method and result display

[0041] The anti-bumping pressure relief algorithm detection has multiple algorithm categories to switch according to different drilling processes.

[0042] To prevent missed detection and false detection due to screen occlusion and glare during anti-impact unloading drilling, in the target detection model reasoning, two algorithms count the drill rods separately. First, when the target detects that the worker is identified as holding a drill rod for a period of time and then changes to an ordinary worker category, the number of drill rods withdrawn is increased by 1. Then, when the drill rig changes from the drill rig withdrawal category to the empty drill rig category after a period of time and then changes to the drill rig withdrawal category, the number of drill rods withdrawn is increased by 1 in combination with the drill rig motion state. Both algorithms calculate the drill rod withdrawal. When one of the algorithms has a higher accuracy rate in a complete drill withdrawal process, it is determined that the current drill rod withdrawal process is completed. The algorithm adopts the target detection algorithm, and the reasoning results include (x1, y1, x2, y2, cls, conf), x1 and y1 represent the coordinates of the upper left corner of the rectangular detection box, x2 and y2 represent the coordinates of the lower right corner of the rectangular detection box, cls represents the category of the detection box, and conf represents the accuracy of the detection box. This accuracy is determined based on which of the accuracies is higher. , at which point the counting logic of another algorithm will be refreshed.

[0043] 4. Calculation of drilling rig motion state

[0044] During the process of drilling and retracting the drill, the direction of the drill is uncertain. In the visual plane, to determine the movement state of the drill, it is necessary to determine the direction of the drill retracting rod based on the movement direction of the drill. First, obtain the center point P of the drill target detection frame. 1 (x 1 ,y 1) Target detection can obtain the coordinates of the upper left corner point x1, y1, and the coordinates of the lower right corner point x2, y2. The average of the coordinates of these two points is obtained to obtain the coordinates of the center point x = (x1 + x2) / / 2, y = (y1 + y2) / / 2;

[0045] When the center point of the drilling rig moves and the distance P 1 After a certain displacement, the center point P is determined. 2 (x 2 ,y 2 ), the slope k of the straight line is determined according to the coordinates of the two points:

[0046]

[0047] The equation of the line is:

[0048] y=k(xx 1 )+y 1

[0049] According to the straight line equation, calculate the intersection coordinates C(x c ,y c ) with each edge:

[0050] Left boundary x=0:

[0051] y c = k(0-x 1 )+y 1

[0052] Right boundary x=w:

[0053] y c = k(wx 1 )+y 1

[0054] Upper boundary y=0:

[0055]

[0056] Lower boundary y=h:

[0057]

[0058] For each calculated intersection point, check C(x c ,y c ) is within the valid range of the corresponding boundary, that is, y c ∈[0,h] or x c ∈[0,w], thereby determining which edges of the image intersect. At the same time, calculate the intersection point C and the drill rod center point P 3 (x 3 ,y 3 )’s dynamic displacement distance d.

[0059]

[0060] According to the dynamic displacement distance d, the drill rod center point P is determined 3 The movement direction of the drill rod is then determined, and the movement state of the drill rod is determined. When the movement state of the drill rod goes through a complete movement cycle, the rod withdrawal number is increased by 1.

[0061] 5. Video action sequence detection

[0062] On the basis of target detection and identification of the state of workers holding drill rods, the algorithm introduces a method based on video action sequence detection. The action sequence detection algorithm adopts the ViViT model, which is a deep learning model based on the Transformer self-attention mechanism. It has achieved industry-leading accuracy on the current mainstream action sequence detection data set. In the process of video action sequence detection, in addition to the spatial dimension, there is also a time dimension. Taking a continuous video sequence as a unit, the process of withdrawing drill rods in anti-bumping and pressure relief, the process of personnel taking drill rods, and the process of drilling rig operation in continuous time segments are dynamically detected. When the process of drilling rig withdrawing drill rods and the process of personnel taking drill rods are successively identified in a continuous video, the number of drill rods withdrawn is recorded and recognized. When the continuous video identifies that the drilling rig is running in the direction of the drill rod, the current state of withdrawing drill rods is recorded and refreshed to make relevant preparations for the next number of drill rods withdrawn.

[0063] 6. Drill rod withdrawal counting logic

[0064] The drill rod withdrawal count is mainly calculated by the target detection algorithm and the video action sequence detection algorithm. The target detection algorithm is divided into the detection of the drilling rig operation state and the identification of the person holding the drill rod. A total of three algorithms are included to count the drill rod withdrawal in parallel. First, the two calculation methods in the target detection algorithm are judged. When the drilling rig movement state is identified from the connection of the drilling rig and the drill rod to the separation of the drilling rig and the drill rod, if it is detected that the person is holding and removing the drill rod, the detection results of the two algorithms are merged. At this time, the algorithm is merged to improve the accuracy. There are two state judgments in the algorithm merger. One is the process of the separation of the drilling rig and the drill rod. At this time, the categories of two target detection frames will be detected. If occlusion occurs, only one part can be detected, that is, the position of the drilling rig and the drill rod is occluded. As long as the person holding the drill rod can be detected, the drill rod is counted, and vice versa. If there is no occlusion, both states need to be detected. If both algorithms have a withdrawal count result, the drill rod is counted once.

[0065] Then the number of rod withdrawals counted by the target detection method and the number of drill rods detected by the video action sequence are combined and judged. At this time, the algorithm is combined to prevent missed detection. In a time period, if any algorithm of the target detection or video action sequence detection recognizes a drill rod, the drill rod will be counted. Because the two algorithms have different recognition methods, the target detection focuses on identifying a certain state of the drill rod and the personnel, and the video action sequence detection is a process of identifying the personnel operating the rod withdrawal. The two algorithms have different detection focuses, so that the two algorithms complement each other and obtain the accuracy rate closest to the actual situation. Finally, the number of rod withdrawals in the current time period is obtained, and then the number of rod withdrawals in each time period is combined to calculate the final total number of drill rod withdrawals.

[0066] 7. Online detection process

[0067] The online detection of anti-bumping and pressure relief process refers to installing a fixed camera in the anti-bumping and pressure relief area and connecting the camera to the AI ​​application platform. By configuring the equipment name, drilling location, designed drilling depth, construction start time and construction personnel, an online drilling process is created. The on-site environment can be monitored in real time through a PC or Android client. When the sign is raised to start the anti-bumping and pressure relief drilling, click the start button to decode and infer the real-time camera RTSP. When the construction is completed as required, click the end button to stop the algorithm model reasoning calculation, and compare the result of the reasoning calculation with the specified design depth. If the number of drills identified by the algorithm model exceeds the specified design depth, the anti-bumping and pressure relief process is returned as qualified, otherwise it is returned as unqualified, and the relevant alarm information is uploaded to the AI ​​application platform for secondary review, and the final result is manually confirmed.

[0068] 8. Offline detection process

[0069] The offline detection drilling function is similar to the online detection function, but the video acquisition method is different. The video of the anti-bumping and pressure-relief drilling process is recorded underground, and the video is stored online for reasoning. The offline video is manually uploaded and added through the AI ​​application platform, and the algorithm model reasoning is performed. The subsequent steps and alarm review content are the same as online reasoning.

[0070] 9. The equipment

[0071] A coal mine anti-blowout pressure relief drill rod identification algorithm, equipment and system, including image acquisition equipment, communication transmission equipment, data processing equipment, data storage equipment, etc.

[0072] The image acquisition equipment is an intrinsically safe camera in coal mines, which can collect high-definition images of the coal conveyor belt during operation. The surface of the camera needs to be made of dust-proof material to avoid the adhesion of dust in the mine, which will affect the image quality.

[0073] Communication transmission equipment includes equipment required for wireless transmission methods such as 5G and industrial wifi, as well as equipment required for wired transmission methods such as USB3.0 and Ethernet.

[0074] The edge processing device uses an AI edge computing server, which has a certain amount of computing power and memory to complete the YOLO-based image target detection task.

[0075] The data storage device is an SD card or a hard disk with a large storage capacity, which is used to store images and videos identified as drill rods.

[0076] The present invention covers any substitution, modification, equivalent method and scheme made on the essence and scope of the present invention. In order to make the public have a thorough understanding of the present invention, specific details are described in detail in the following preferred embodiments of the present invention, but those skilled in the art can fully understand the present invention without the description of these details. In addition, in order to avoid unnecessary confusion about the essence of the present invention, well-known methods, processes, procedures, components and circuits are not described in detail.

[0077] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A coal mine anti-blowout pressure relief drill rod identification algorithm and system, including image acquisition equipment, communication transmission equipment, data processing equipment and data storage equipment, characterized in that: The image acquisition device is an intrinsically safe camera in the coal mine, which collects relevant images and video data of the drilling rig, drill rod and operators during the anti-bumping and pressure relief process. The image acquisition device is installed about 2 meters above the drilling rig, and can clearly capture the movement status of the drilling rig and drill rod and the operation of the operator; The communication transmission equipment uses 5G, industrial WiFi or wired communication to transmit the collected image and video data to the data processing equipment in real time; The data processing equipment uses the target detection algorithm and the video action sequence detection algorithm to analyze the collected images and video data. The target detection algorithm is based on the artificial intelligence visual model to identify the status of the drill rod, drilling rig and personnel in the image, and judges the number of drill rods withdrawn during the anti-blowout pressure relief process by analyzing the connection and disconnection status of the drilling rig and the movement of the personnel holding the drill rod. The video action sequence detection algorithm is based on the continuous video analysis of the time dimension to detect the withdrawal behavior and the movement status of the drilling rig during the anti-blowout pressure relief process, and dynamically count the number of drill rods during the entire withdrawal process. The data storage device is used to store the identified drill pipe images, videos and calculation results, and upload the data to the AI ​​application platform for further analysis and recording.

2. A coal mine anti-bumping and pressure relief drill rod identification algorithm and system according to claim 1, characterized in that: The video acquisition resolution of the image acquisition device is 1920×1080 and above, the frame rate is 25 frames per second, and the total bit rate is 4096kbps and above, which can clearly capture the movement status of the drill pipe and the drilling rig in a complex underground environment.

3. The coal mine anti-bumping and pressure relief drill rod identification algorithm and system according to claim 1, characterized in that: The data processing device uses a target detection algorithm to identify the drill rod held by the person, the operating status of the drilling rig, and the connection and disconnection status of the drilling rig and the drill rod, and calculates the number of drill rods withdrawn in real time in the image.

4. The coal mine anti-bumping and pressure relief drill rod identification algorithm and system according to claim 1, characterized in that: The video action sequence detection algorithm adopts the ViViT model and is based on the Transformer self-attention mechanism to detect the motion state of the drilling rig, the drill rod removal action and the drill withdrawal behavior in continuous time segments during the anti-bumping and pressure relief process, and counts the number of each drill withdrawal process.

5. The coal mine anti-bumping and pressure relief drill rod identification algorithm and system according to claim 1, characterized in that: The data processing device comprehensively determines the number of drill rod withdrawals by fusing the detection results of the target detection algorithm and the video action sequence detection algorithm, and performs correction by combining the results of the two algorithms when missed detection or false detection occurs.

6. A coal mine anti-bumping and pressure relief drill rod identification algorithm and system according to claim 1, characterized in that: The system utilizes dynamic detection of the movement state of the drilling rig and the connection and disconnection state between the drill rod and the drilling rig, and combines an algorithm to calculate the movement direction of the center point of the drill rod to further confirm the number of drill rods to be withdrawn.

7. A coal mine anti-bumping and pressure relief drill rod identification algorithm and system according to claim 1, characterized in that: The system generates anti-bumping pressure relief management records by real-time detection of images and videos during the anti-bumping pressure relief drilling process, and generates alarm information when the number of drill rods does not meet the design requirements and uploads it to the AI ​​application platform for review.

8. The coal mine anti-bumping and pressure relief drill rod identification algorithm and system according to claim 1, characterized in that: The algorithm in the system is combined with image preprocessing operations, including angle rotation, image enhancement and other methods, to improve the accuracy of image recognition under complex lighting conditions during the anti-impact pressure relief process.

9. The coal mine anti-bumping and pressure relief drill rod identification algorithm and system according to claim 1, characterized in that: The system has two detection modes: online detection and offline detection. The online detection monitors the drill rod withdrawal process through real-time video, and the offline detection analyzes the pre-recorded video.

10. The coal mine anti-bumping and pressure relief drill rod identification algorithm and system according to claim 1, characterized in that: The system's AI edge computing server has high computing power and large memory, and can process image and video data in real time to ensure the accuracy and timeliness of detection of the number of drill rods withdrawn.

Images