Intelligent monitoring and judging method and system for temporary support of blasting driving working face

Abstract

The invention relates to the technical field of coal mine safety management, and discloses an intelligent monitoring and judging method and system for temporary support of a blasting driving working face, the system comprises a monitoring camera, a data processing unit, a judging module and an alarm device, the data processing unit comprises a data acquisition module, a data storage module, a data processing module and a communication module, a monitoring camera, a data processing unit, a judgment module and an alarm device are integrated, safety supervision intelligence of temporary support of a methane sensor and a blasting tunneling face of a fully-mechanized coal mining face air return corner under a coal mine is achieved through a monitoring video level, and a traditional safety production supervision mode is changed into a mode that a video line is manually switched to watch a monitoring video; manual intervention in safety production supervision is reduced, safety production supervision is intelligentized and standardized, the efficiency of finding unsafe factors is improved, accidents are reduced, underground production is indirectly supported, production efficiency is improved, and a large amount of intangible economic benefits are brought.

Description

Technical Field

[0001] The present invention relates to the technical field of coal mine safety management, and in particular to an intelligent monitoring and judgment method and system for real-time support of a blasting working surface. Background Art

[0002] In the process of coal mining, temporary support of blasting working face is an important part to ensure the safety of workers. Traditional temporary support monitoring methods mainly rely on manual inspection, which has problems such as low efficiency and susceptibility to human factors. With the improvement of the intelligent level of coal mines, higher requirements are put forward for temporary support monitoring technology.

[0003] In the prior art, although there are some monitoring systems based on video surveillance, these systems still have deficiencies in recognition accuracy, real-time performance, and adaptability to complex environments. For example, some systems have difficulty accurately identifying target objects under complex backgrounds and lighting conditions, resulting in reduced reliability of monitoring results. In addition, existing systems are prone to delays when processing large amounts of data and cannot meet the needs of real-time monitoring.

[0004] To this end, we proposed an intelligent monitoring and judgment method and system for real-time support in blasting work. Summary of the invention

[0005] The present invention mainly solves the technical problems existing in the above-mentioned prior art and provides a method and system for intelligent monitoring and judgment of real-time support of blasting working surface.

[0006] In order to achieve the above-mentioned purpose, the present invention adopts the following technical scheme: an intelligent monitoring and judgment system for temporary support of blasting working face, including a monitoring camera, a data processing unit, a judgment module and an alarm device. The monitoring camera is used to monitor the scope of temporary support of blasting working face. The data processing unit includes a data acquisition module, a data storage module, a data processing module and a communication module, which are used to collect, store, process and transmit the picture data taken by the monitoring camera.

[0007] Preferably, the judgment module includes a judgment algorithm module and a threshold setting module, which are used to judge the safety status of the support structure according to a preset threshold and judgment algorithm.

[0008] Preferably, the alarm device comprises an audible and visual alarm module and a remote alarm module, which is used to send out audible and visual alarm signals when a potential safety hazard is detected, and to send alarm information to relevant personnel via remote communication.

[0009] A method for intelligent monitoring and judging the temporary support of a blasting excavation work surface, comprising the above-mentioned intelligent monitoring and judging system for the temporary support of a blasting excavation work surface, specifically comprises the following steps:

[0010] Step 1: Identification of knocking and asking for the top: First, determine whether there is someone doing the knocking action in the current picture. In the process of identifying the knocking action, the main methods are human body tracking, posture analysis and posture prediction. At the same time, determine whether there is a stick-like object in the picture moving synchronously with the person to enhance the accuracy of the algorithm. If the knocking and asking for the top is not identified, an alarm message is issued through the alarm device;

[0011] Step 2: Human body tracking: One frame of video takes about 0.03 seconds. Two images 10 frames apart are taken as the criterion for judging the motion trend. That is, the motion trend of the target object is judged every 0.3 seconds. The motion information can be obtained by connecting the information of each time point in series.

[0012] Step 3: Posture recognition: Use background difference method and morphological algorithm to extract the target skeleton, use the method based on human body proportion to preliminarily obtain the human body situation, and then use the algorithm based on motion trend to obtain accurate human body information;

[0013] Step 4: Posture prediction: To enhance recognition accuracy and improve computing performance, a posture prediction algorithm is used for auxiliary analysis. The posture prediction algorithm assumes several movement directions based on the movement trends of people and objects. If the movement trends meet the assumptions, the possible set of current movements is obtained to reduce the final recognition range.

[0014] Step 5: Front beam identification: Use image compensation, de-jittering and video frame interpolation to optimize video quality, and then use model matching, pre-judgment screening and image de-duplication technology to identify the front beam;

[0015] Step 6: Front beam recognition model matching: Sample the video to obtain a single frame image, perform target detection reasoning through the ResNet or Inception V4 image recognition model, obtain a rectangular detection frame that may be the front beam, and then match the content in the rectangular detection frame with the front beam model to analyze and calculate whether the content in the rectangular detection frame is the front beam;

[0016] Step 7: Deduplication of front beam recognition images: Count the front beams recognized in the video frame, remove the duplicate recognition areas, and obtain the actual number of front beams. If the number of current beams is large, the number of front beams can be directly estimated through CSRNet.

[0017] Step 8: Backboard connection identification: During the backboard connection operation, since the color, outline, size and other appearance factors of the backboard and installation tools are easily integrated with the background and cannot be directly extracted from the video screen, some targets of interest are first framed by model matching, and then the framed targets are tracked in combination with subsequent images. Invalid targets are eliminated according to the trajectory, and finally the path prediction method is used to confirm whether the target object is the backboard and installation tool, and then combined with human body recognition to determine whether to perform the backboard connection operation;

[0018] Step 9: Anchor drill rig identification: After the back plate is connected to the top, the anchor bolts are started to be driven and the anchor net is fixed to the top. Since the anchor bolts are thin, they cannot be seen clearly in the video, but the anchor drill rig is obviously different from other machines. Therefore, the working status of the anchor drill rig is used to determine whether the anchor bolting operation is currently being performed. The anchor drill rig is mainly identified by the model matching method, and the working status of the anchor drill rig is identified and analyzed by motion recognition and motion prediction technology;

[0019] Step 10: Matching the anchor drill identification model: Due to the special shape of the anchor drill, the image recognition model of ResNet or Inception V4 is also used to identify the anchor drill. Since the anchor drill is very likely to be blocked when working, the entire shape cannot be seen. Therefore, when building the algorithm model, the anchor drill is divided into multiple parts to improve the recognition accuracy.

[0020] Step 11: Anchor drill identification motion trajectory analysis: Due to occlusion, the content matched by the model may not be an anchor drill. Therefore, a motion trajectory analysis algorithm is used to remove non-anchor drill content. If the matched content is doing a straight line reciprocating motion, it can be determined that the content is an anchor drill.

[0021] Step 12: Prediction of anchor drill identification and motion: The algorithm pre-sets the path of the anchor drill when it is working. While obtaining the path, it pre-assumes the possible position of the drill in the next frame of the image. If the assumption is correct, the confidence value is increased according to the strategy. When the confidence value exceeds 95%, it is determined that the matching content is an anchor drill and is in working condition.

[0022] Step 13: Logical implementation: The algorithm starts with knocking on the side and asking about the top. When the front beam appears on the screen, it is determined that knocking on the side and asking about the top are completed, and the temporary support installation process begins. When the anchor drilling rig starts working, it is determined that the temporary support installation is completed and the algorithm analysis is completed.

[0023] The present invention provides a method and system for intelligent monitoring and judgment of temporary support for blasting work faces. It has the following beneficial effects:

[0024] 1. This method and system for intelligent monitoring and judgment of temporary support for blasting work faces integrates monitoring cameras, data processing units, judgment modules and alarm devices, and realizes intelligent safety supervision of methane sensors at the return air corners of fully mechanized mining faces in underground coal mines and temporary support for blasting work faces through monitoring video levels. It transforms the traditional safety production supervision mode of manually switching video lines to watch monitoring videos and manually taking screenshots and saving them after anomalies are found, into system automatic processing. During use, it is only necessary to configure the algorithm and parameters into the system, and the system automatically allocates algorithm services to each camera video stream and automatically starts calculation and analysis. The discovery of anomalies and the fixation of evidence are fully programmed, and the judgment of video content will not be affected by factors such as experience and concentration. It reduces manual intervention in safety production supervision, makes safety production supervision intelligent and standardized, improves the efficiency of discovering unsafe factors, reduces accidents, indirectly supports underground production, improves production efficiency, and brings a large number of intangible economic benefits.

[0025] 2. This method and system of intelligent monitoring and judgment of real-time support for blasting work has greatly improved the safety management efficiency by monitoring the underground production process and giving real-time alarms. It has good economic and social benefits. Through the implementation of this project, the intelligent diversified development capabilities of coal mines have been further improved, redundant workloads have been reduced, and the intelligence of coal mine control and information management has reached a new level, taking a solid step towards the development of smart mines.

[0026] 3. This method and system for intelligent monitoring and judgment of real-time support for blasting work can effectively reduce safety accidents caused by human negligence or untimely response in practical applications, thereby ensuring the safety of miners' lives. The system can also record and store monitoring data in real time, providing a detailed basis for subsequent analysis and improvement, and further improving the level of safety management in coal mines.

[0027] 4. This method and system for intelligent monitoring and judgment of real-time support for blasting work adopts advanced image recognition technology and artificial intelligence algorithms to achieve accurate recognition and judgment of complex underground environments. The system can automatically identify and track target objects, such as front exploration beams, back plates, anchor drills, etc., to ensure the continuity and accuracy of monitoring. The system also has the ability of self-learning and optimization. With the accumulation of data and iteration of algorithms, the recognition accuracy and response speed will continue to improve.

[0028] 5. This method and system for intelligent monitoring and judgment of real-time support for blasting working faces is not only applicable to blasting working faces, but can also be extended to other coal mine working faces and even to the safety monitoring field of other industries. Its versatility and scalability provide new safety supervision ideas for coal mines and other industries, and promote the advancement of safety technology in the entire industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 It is a system module diagram of the present invention;

[0030] Figure 2 The figure is a flow chart of the method of the present invention. DETAILED DESCRIPTION

[0031] Embodiment 1: An intelligent monitoring and judgment system for the real-time support of a blasting working face, such as Figure 1 As shown, it includes monitoring cameras, data processing units, judgment modules and alarm devices. The monitoring cameras are used to monitor the scope of temporary support for blasting work faces. By integrating monitoring cameras, data processing units, judgment modules and alarm devices, the intelligent safety supervision of return air corner methane sensors and temporary support for blasting work faces in underground coal mines is realized through the monitoring video level. The traditional safety production supervision mode, which is manually switched to watch monitoring videos and manually take screenshots and save them after abnormalities are found, is transformed into system automatic processing. During use, it is only necessary to configure the algorithm and parameters into the system. The system automatically allocates algorithm services to each camera video stream and automatically starts calculation and analysis. The discovery of abnormalities and the fixation of evidence are fully programmed, and the judgment of video content will not be affected by factors such as experience and concentration. Manual intervention in safety production supervision is reduced, making safety production supervision intelligent and standardized, improving the efficiency of discovering unsafe factors, reducing accidents, indirectly supporting underground production, improving production efficiency, and bringing a large number of intangible economic benefits.

[0032] Embodiment 2: Based on Embodiment 1, Figure 1 As shown, the data processing unit includes a data acquisition module, a data storage module, a data processing module and a communication module, which are used to collect, store, process and transmit the image data taken by the monitoring camera. By monitoring the underground production process and real-time alarm, it has a great effect on improving the safety management efficiency and has good economic and social benefits. Through the implementation of this project, the intelligent and diversified development capabilities of coal mines have been further improved, redundant workloads have been reduced, and the intelligence of coal mine control and information management has reached a new level, taking a solid step towards the development of smart mines.

[0033] Embodiment 3: Based on Embodiment 1 and Embodiment 2, Figure 1 As shown, the judgment module includes a judgment algorithm module and a threshold setting module, which are used to judge the safety status of the support structure according to the preset threshold and judgment algorithm. In practical applications, it can effectively reduce safety accidents caused by human negligence or untimely response, thereby ensuring the life safety of miners. The system can also record and store monitoring data in real time, providing a detailed basis for subsequent analysis and improvement, and further improving the level of safety management in coal mines.

[0034] Embodiment 4: Based on Embodiment 1, Embodiment 2 and Embodiment 3, Figure 1 As shown, the alarm device includes an audible and visual alarm module and a remote alarm module, which are used to send out audible and visual alarm signals when potential safety hazards are detected, and send alarm information to relevant personnel through remote communication. By adopting advanced image recognition technology and artificial intelligence algorithms, accurate recognition and judgment of complex underground environments are achieved. The system can automatically identify and track target objects, such as front exploration beams, back plates, anchor drilling rigs, etc., to ensure the continuity and accuracy of monitoring. The system also has the ability of self-learning and optimization. With the accumulation of data and iteration of algorithms, the recognition accuracy and response speed will continue to improve.

[0035] Embodiment 5: Based on Embodiment 1, Embodiment 2, Embodiment 3 and Embodiment 4, Figure 2As shown, a method for intelligent monitoring and judgment of temporary support for blasting work faces, including the above-mentioned intelligent monitoring and judgment system for temporary support for blasting work faces, specifically includes the following steps: Step 1: Identification of knocking on the side and asking for the top: First, it is to determine whether there is a person performing a knocking action in the current picture. In the process of identifying the knocking action, the methods of human body tracking, posture analysis and posture prediction are mainly adopted. At the same time, it is judged whether there is a stick-like object in the picture and the person are moving synchronously to enhance the accuracy of the algorithm. If the knocking on the side and asking for the top are not identified, an alarm message is issued through an alarm device; Step 2: Human body tracking: The time of one frame of video is about 0.03 seconds, and two images separated by 10 frames are taken as the judgment standard of motion trend, that is, the motion trend of the target object is judged every 0.3 seconds, and the information of each time point is connected in series to obtain the motion information; Step 3: Posture Posture recognition: Use background difference method and morphological algorithm to extract the target skeleton, use the method based on human body proportion to obtain the human body situation preliminarily, and then use the algorithm based on motion trend to obtain accurate human body information; Step 4: Posture prediction: In order to enhance recognition accuracy and improve computing performance, use posture prediction algorithm for auxiliary analysis. The posture prediction algorithm assumes several movement directions according to the movement trend of people and objects. If the movement trend meets the assumption, the possible set of current movement is obtained to reduce the final recognition range; Step 5: Front beam recognition: Use picture compensation, de-shaking and video interpolation to optimize video quality, and then use model matching, prediction screening, and image deduplication technology to identify the front beam; Step 6: Front beam recognition model matching: Sample the video to obtain a single frame image, and use ResNet or Inception V4's image recognition model performs target detection reasoning to obtain a rectangular detection frame that may be a front probe beam, and then analyzes and calculates whether the content in the rectangular detection frame is a front probe beam by matching the content in the rectangular detection frame with the front probe beam model; Step 7: Front probe beam recognition image deduplication: Count the front probe beams identified in the video frame, remove repeated recognition areas, and obtain the actual number of front probe beams. If the current number of probe beams is large, the number of front probe beams can be directly estimated through CSRNet; Step 8: Backboard top connection recognition: During the backboard top connection operation, since the color, outline and size of the backboard and installation tools are easily integrated with the background and cannot be directly extracted from the video screen, some targets of interest are first framed by model matching, and then the framed targets are tracked in combination with subsequent images. Invalid targets are eliminated according to the trajectory, and finally the path prediction method is used to confirm whether the target is a backboard and an installation tool, and then combined with human body posture recognition to determine whether to perform the backboard top connection operation;Step 9: Anchor drill rig identification: After the back plate is connected to the top, the anchor bolts are started and the anchor net is fixed to the top. Since the anchor bolts are thin, they cannot be seen clearly in the video, but the anchor drill rig is obviously different from other machines. Therefore, the working status of the anchor drill rig is used to determine whether the anchor bolting operation is currently in progress. The anchor drill rig is mainly identified by the model matching method. The working status of the anchor drill rig is identified and analyzed by motion recognition and motion prediction technology; Step 10: Anchor drill rig identification model matching: Due to the special shape of the anchor drill rig, ResNet or Inception is also used V4's image recognition model is used to identify the anchor drill. Since the anchor drill is very likely to be blocked when working, its entire appearance cannot be seen. Therefore, the anchor drill is divided into multiple parts when building the algorithm model to improve the recognition accuracy. Step 11: Anchor drill identification motion trajectory analysis: Due to occlusion, the content matched by the model may not be the anchor drill. For this reason, the motion trajectory analysis algorithm is used to remove non-anchor drill content. If the matched content is doing a straight line reciprocating motion, it can be determined that the content is an anchor drill. Step 12: Anchor drill identification motion prediction: The algorithm pre-sets the path of the anchor drill when it is working, and assumes in advance that in the next frame while obtaining the path The possible position of the drilling rig in the image, if the assumption is correct, the confidence value is increased according to the strategy. When the confidence value exceeds 95%, it is determined that the matching content is an anchor drilling rig and is in working condition; Step 13: Logical implementation: Starting from knocking the side and asking the top as the algorithm starting point, after the front beam appears on the screen, it is judged that the knocking the side and asking the top are finished, and the temporary support installation process is entered. When the anchor drilling rig starts working, it is judged that the temporary support installation is finished and the algorithm analysis is finished. It is not only applicable to blasting working faces, but also can be extended to other coal mining working faces and even the safety monitoring field of other industries. Its versatility and scalability provide new safety supervision ideas for coal mines and other industries, and promote the progress of safety technology in the entire industry. ;

[0036] The working principle of the present invention is as follows: by real-time monitoring of the underground working environment of the coal mine, various safety hazards are automatically identified and warned. The system first collects real-time video data underground through a high-resolution camera, and then uses image processing technology to pre-process the video data, such as denoising, contrast enhancement, etc., to improve the accuracy of subsequent recognition. On the basis of image preprocessing, the system uses a deep learning model to analyze the video frames frame by frame. Through a well-trained convolutional neural network (CNN) model, the system can identify underground workers, equipment, tools and potential dangerous objects. For example, the system can identify whether the worker is wearing a safety helmet, whether he is working in a dangerous area, whether there is unfixed equipment, etc. In order to improve the accuracy and robustness of recognition, the system also uses a variety of algorithms for cross-validation. For example, when identifying the front exploration beam, the system not only uses model matching technology, but also combines image deduplication and video frame interpolation technology to ensure that the existence of the front exploration beam can be accurately identified even in a complex background. In addition, the system also uses motion trajectory analysis and motion prediction technology to accurately judge the working status of the anchor drilling rig, thereby ensuring that potential safety hazards are discovered in time during underground operations.

[0037] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. An intelligent monitoring and judgment system for the real-time support of a blasting working face, characterized in that: It includes a monitoring camera, a data processing unit, a judgment module and an alarm device. The data processing unit includes a data acquisition module, a data storage module, a data processing module and a communication module.

2. The intelligent monitoring and judgment system for the temporary support of the blasting working face according to claim 1 is characterized by: The judgment module includes a judgment algorithm module and a threshold setting module.

3. The intelligent monitoring and judgment system for the real-time support of the blasting working face according to claim 1 is characterized by: The alarm device comprises an audible and visual alarm module and a remote alarm module.

4. A method for intelligent monitoring and judgment of the temporary support of a blasting working face, characterized in that: The intelligent monitoring and judgment system for the temporary support of the blasting working face according to any one of claims 1 to 3 specifically comprises the following steps: Step 1: Identification of knocking and asking for the top: First, determine whether there is someone doing the knocking action in the current picture. In the process of identifying the knocking action, the main methods are human body tracking, posture analysis and posture prediction. At the same time, determine whether there is a stick-like object in the picture moving synchronously with the person to enhance the accuracy of the algorithm. If the knocking and asking for the top is not identified, an alarm message is issued through the alarm device; Step 2: Human body tracking: One frame of video takes about 0.03 seconds. Two images 10 frames apart are taken as the criterion for judging the motion trend. That is, the motion trend of the target object is judged every 0.3 seconds. The information of each time point is connected in series to obtain the motion information. Step 3: Posture recognition: Use background difference method and morphological algorithm to extract the target skeleton, use the method based on human body proportion to obtain the human body situation, and then use the algorithm based on motion trend to obtain accurate human body information; Step 4: Posture prediction: To enhance recognition accuracy and improve computing performance, a posture prediction algorithm is used for auxiliary analysis. The posture prediction algorithm assumes several movement directions based on the movement trends of people and objects. If the movement trends meet the assumptions, the possible set of current movements is obtained to reduce the final recognition range. Step 5: Front beam identification: Use image compensation, de-jittering and video frame interpolation to optimize video quality, and then use model matching, pre-judgment screening and image de-duplication technology to identify the front beam; Step 6: Front beam recognition model matching: Sample the video to obtain a single frame image, perform target detection reasoning through the ResNet or Inception V4 image recognition model, obtain a rectangular detection frame that may be the front beam, and then match the content in the rectangular detection frame with the front beam model to analyze and calculate whether the content in the rectangular detection frame is the front beam; Step 7: Deduplication of front beam recognition images: Count the front beams recognized in the video frame, remove the duplicate recognition areas, and obtain the actual number of front beams. If the number of current beams is large, the number of front beams can be directly estimated through CSRNet. Step 8: Backboard connection identification: During the backboard connection operation, since the color, outline, size and other appearance factors of the backboard and installation tools are easily integrated with the background and cannot be directly extracted from the video screen, some targets of interest are first framed by model matching, and then the framed targets are tracked in combination with subsequent images. Invalid targets are eliminated according to the trajectory, and finally the path prediction method is used to confirm whether the target object is the backboard and installation tool, and then combined with human body recognition to determine whether to perform the backboard connection operation; Step 9: Anchor drill rig identification: After the back plate is connected to the top, the anchor bolts are started to be driven and the anchor net is fixed to the top. Since the anchor bolts are thin, they cannot be seen clearly in the video, but the anchor drill rig is obviously different from other machines. Therefore, the working status of the anchor drill rig is used to determine whether the anchor bolting operation is currently being performed. The anchor drill rig is mainly identified by the model matching method, and the working status of the anchor drill rig is identified and analyzed by motion recognition and motion prediction technology; Step 10: Matching the anchor drill identification model: Due to the special shape of the anchor drill, the image recognition model of ResNet or Inception V4 is also used to identify the anchor drill. Since the anchor drill is very likely to be blocked when working, the entire shape cannot be seen. Therefore, when building the algorithm model, the anchor drill is divided into multiple parts to improve the recognition accuracy. Step 11: Anchor drill identification motion trajectory analysis: Due to occlusion, the content matched by the model may not be an anchor drill. Therefore, a motion trajectory analysis algorithm is used to remove non-anchor drill content. If the matched content is doing a straight line reciprocating motion, it can be determined that the content is an anchor drill. Step 12: Prediction of anchor drill identification and motion: The algorithm pre-sets the path of the anchor drill when it is working. While obtaining the path, it pre-assumes the possible position of the drill in the next frame of the image. If the assumption is correct, the confidence value is increased according to the strategy. When the confidence value exceeds 95%, it is determined that the matching content is an anchor drill and is in working condition. Step 13: Logical implementation: The algorithm starts with knocking on the side and asking about the top. When the front beam appears on the screen, it is determined that knocking on the side and asking about the top are completed, and the temporary support installation process begins. When the anchor drilling rig starts working, it is determined that the temporary support installation is completed and the algorithm analysis is completed.

Images