Method and system for measuring visual state of face guard based on deep learning

Abstract

The invention provides a face guard visual state measurement method and system based on deep learning, and the system comprises an image collection unit and a visual detection system. The image acquisition unit comprises a first image acquisition unit arranged below the first hydraulic support, a light source arranged below the second hydraulic support and a second image acquisition unit arranged below the third hydraulic support; a side protecting plate is arranged on the second hydraulic support; the light source directly faces the face guard and is used for measuring the light intensity and the light irradiation angle of the face guard; the first image acquisition unit and the second image acquisition unit shoot the side protection plate in a left-right adjacent side cross shooting mode to obtain a first acquisition image and a second acquisition image; and the visual detection system is configured to receive the first acquisition image and the second acquisition image, transmit the first acquisition image and the second acquisition image to a deep learning model for feature extraction, monitor the state of the face guard based on the extracted features, judge whether the working state of the face guard conforms to a preset state or not, and perform early warning if the working state does not conform to the preset state.

Description

Technical Field

[0001] The present invention belongs to the technical field of visual state measurement of a guard plate, and in particular, relates to a method and system for measuring the visual state of a guard plate based on deep learning. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] The guard plate is the key part of the hydraulic support to support the coal wall. Its main function is to prevent the coal wall from breaking off. When the coal mining machine drum cuts the coal wall, when the drum passes by, the guard plate on the coal wall should be retracted in time to avoid cutting interference. Therefore, there are strict requirements for the working status detection of the hydraulic support guard plate.

[0004] At present, contact sensors are mainly used for detecting the status of guard plates in underground comprehensive mining faces. For example, patent application number 201310087316.4, the invention name is hydraulic support guard plate control device and control method thereof; this application installs an inclination sensor and a pressure sensor on each guard plate by connecting guard plate sensors in series, and detects and controls the status of each guard plate through sensors.

[0005] The above-mentioned contact sensors have the problems of complex layout and high installation and maintenance costs, as well as difficulties in signal transmission and poor consistency of multi-sensor data fusion. In addition, the vibration and weak signal of underground coal mining machines lead to difficulties in data transmission and low measurement accuracy, making it difficult to achieve full-time and space multi-target synchronous measurement.

[0006] In addition, there are existing technologies that use neural network models to measure the telescopic angle of the guard plate, for example: 202111596613.2, the patent name is: Guard plate telescopic angle measurement method and device based on image sequence, which can realize real-time measurement and visualization of the telescopic angle of the guard plate during movement. It is of great practical significance to realize intelligent control of hydraulic supports, prevent the occurrence of coal wall slabs on the working face and collision and interference between the guard plate and the coal machine drum, and reduce the labor intensity of personnel.

[0007] In the prior art, there are methods of measuring the guard plates using algorithms. For example, patent CN111173510A uses an inertial navigation device, inclination and displacement sensors, and image analysis methods to form a cutting interference detection system. This system has multiple data coupling errors, and the contact sensor measurement is affected by the vibration of the equipment, resulting in large measurement errors; patent CN111810155B uses a visual measurement solution, and the single camera sensor decision-making warning has a high false alarm and missed alarm rate due to the influence of dust and light underground; patent CN114511799B uses two deep learning network models to measure the support angle of the guard plate. The model data is large, the training is complex, and the real-time performance is poor. Summary of the invention

[0008] In order to overcome the above-mentioned deficiencies in the prior art, the present invention provides a method for measuring the visual state of a guard plate based on deep learning, which has high accuracy in detecting the working state of the guard plate.

[0009] To achieve the above objectives, one or more embodiments of the present invention provide the following technical solutions:

[0010] In a first aspect, a deep learning-based visual state measurement system for a side guard is disclosed, comprising: an image acquisition unit and a visual detection system, wherein the image acquisition unit transmits the acquired image to the visual detection system;

[0011] The image acquisition unit includes: a first image acquisition unit arranged below the first hydraulic support, a light source arranged below the second hydraulic support, and a second image acquisition unit arranged below the third hydraulic support;

[0012] The second hydraulic support is provided with a side guard plate;

[0013] The light source is facing the guard plate and is used to provide light intensity and light irradiation angle for measuring the guard plate;

[0014] The first image acquisition unit and the second image acquisition unit photograph the support plate by cross-photographing the left and right adjacent sides to obtain the first acquired image and the second acquired image;

[0015] The visual inspection system is configured to: receive the first acquired image and the second acquired image, and transmit them to the deep learning model for feature extraction, monitor the status of the guard plate based on the extracted features, determine whether its working status is consistent with the preset status, and issue an early warning if it is not consistent.

[0016] As a further technical solution, the first hydraulic support, the second hydraulic support and the third hydraulic support are arranged in sequence and have the same structure, all of which include a bottom plate and a top plate, a connecting column is arranged between the bottom plate and the top plate, and a guard plate is arranged at the suspended end of the top plate of the second hydraulic support.

[0017] As a further technical solution, the visual inspection system includes a big data management and control platform, a computer and a cloud server;

[0018] The big data management and control platform stores the first collected image and the second collected image of the side guard plate and the status information of the side guard plate;

[0019] The computer and the cloud server both communicate with the big data management and control platform.

[0020] In a second aspect, a method for measuring the visual state of a guard plate based on deep learning is disclosed, comprising:

[0021] The guard plate is photographed by cross-shooting the left and right adjacent sides to obtain a first acquisition image and a second acquisition image;

[0022] The first acquired image and the second acquired image are transmitted to the deep learning model for processing to obtain the working state of the side guard plate;

[0023] The obtained working states of the side guard plate are weighted to determine the final working state of the side guard plate and judge whether there is any abnormality in the working state of the side guard plate.

[0024] As a further technical solution, the first acquired image and the second acquired image are transmitted to a deep learning model for processing, including:

[0025] Performing grayscale processing on the first acquired image and the second acquired image respectively to obtain a first side guard plate grayscale image and a second side guard plate grayscale image;

[0026] Based on the grayscale image of the first side guard plate, the gap between the right side of the side guard plate and the coal wall and the gap between the top beam of the hydraulic support are identified and judged to obtain the working state of the side guard plate;

[0027] Based on the grayscale image of the first side guard plate, the gap between the left side of the side guard plate and the coal wall and the gap between the top beam of the hydraulic support are identified and judged to obtain the working status of the side guard plate.

[0028] As a further technical solution, six types of data information, including the identified left and right working states of the guard plate, the gap between the guard plate and the coal wall, and the gap between the guard plate and the top beam of the hydraulic support, are integrated and processed to finally determine the actual working state of the guard plate according to the weight formula.

[0029] As a further technical solution, the specific calculation formula is:

[0030]

[0031] Among them, R, X i∈(-1, 0, 1), -1 means no, 1 means yes, 0 means no target is detected, and X1, X2, and X3 in the weight formula represent the detection results of the gap between the guard plate and the coal wall, the detection results of the guard plate and the top beam of the hydraulic support, and the detection results of the working characteristics of the guard plate, respectively.

[0032] As a further technical solution, the training process of the deep learning model is:

[0033] Establishing a guard plate state detection dataset based on the first collected image and the second collected image, and performing enhancement processing on the guard plate state detection dataset to obtain a deep learning dataset;

[0034] The deep learning data set is used to train a deep learning model for detection to obtain a trained deep learning model.

[0035] One or more of the above technical solutions have the following beneficial effects:

[0036] The technical solution of the present invention detects and distinguishes three different working states of the guard plate (including supporting state, retracted state, and running state), uses a positive light source to illuminate the guard plate to improve the quality of photographed images, and uses cameras on adjacent two sides to detect the working state of the guard plate. By weighted processing the detection results of the double-station camera, the working state detection accuracy of the guard plate is high.

[0037] The technical solution of the present invention adopts an end-to-end full visual detection solution, inputs the same-frame guard plate images taken by a dual-station camera, that is, images of the same guard plate taken by two cameras at the same time, and outputs the guard plate working status information. The detection efficiency is high and the real-time performance is strong. In addition, the working status perception of the guard plates on the entire comprehensive mining face at the same time is highly coordinated, is less affected by the vibration of the working environment, and has good robustness.

[0038] The technical solution of the present invention adopts the method of visual sensors, the arrangement of sensor points is simple, the total number of arrangements is small, the installation and maintenance are simple, and the detection cost is low.

[0039] The technical solution of the present invention saves the detection result information and image and video information to the cloud, realizes remote monitoring and alarm through a deep learning model platform, can conveniently and timely divide the guard plate status, and facilitates data storage, detection result statistics and hydraulic support control.

[0040] The technical solution of the present invention adopts a dual-camera to collect a multi-view status detection solution for a guard plate, comprehensively decides the current operating status of the guard plate through multiple positions and multiple evaluation criteria, and adopts an end-to-end guard plate status detection algorithm, with high real-time detection.

[0041] Advantages of additional aspects of the present invention will be given in part in the following description, and in part will become obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0043] Figure 1 It is a schematic diagram of the structure of an embodiment of the present invention;

[0044] Figure 2 This is a schematic diagram of cloud control according to an embodiment of the present invention;

[0045] Figure 3 is a flow chart of a detection method according to an embodiment of the present invention;

[0046] Figure 4 This is a flowchart of image processing according to an embodiment of the present invention;

[0047] Figure 5 This is a schematic diagram of the image acquisition structure of an embodiment of the present invention;

[0048] Figure 6 This is a flow chart of the deep learning algorithm of an embodiment of the present invention;

[0049] Figure 7 This is a schematic diagram of a gap in an embodiment of the present invention;

[0050] In the figure, 1, the first hydraulic support; 2, the second hydraulic support; 3, the third hydraulic support; 4, the side guard plate; 5, the left-facing work station camera; 6, the right-facing work station camera; 7, the positive light source, 8, the first gap, 9, the top beam gap. DETAILED DESCRIPTION

[0051] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.

[0052] It should be noted that the terms used herein are for describing specific embodiments only and are not intended to be limiting of exemplary embodiments according to the present invention.

[0053] In the absence of conflict, the embodiments of the present invention and the features of the embodiments may be combined with each other.

[0054] The existing technology uses contact sensors to measure the guard plates, but the following problems exist: the detection accuracy is easily affected by the fully-mechanized mining face environment, resulting in inaccurate data and poor robustness; data transmission in the underground working environment is difficult and has poor real-time performance; multi-sensor data is difficult to achieve full-time and multi-target guard plate status detection; sensor installation and maintenance are difficult and costly; multi-type sensor data synchronization and data processing are cumbersome, and end-to-end status detection cannot be achieved.

[0055] In addition, due to the vibration caused by the operation of underground comprehensive mining equipment and the equipment displacement caused by the operation of the equipment, the contact sensor measurement noise is large, the measurement error is large and it is easy to be damaged, and the sensor installation and maintenance are difficult; because the comprehensive mining face is a narrow and long tunnel, the number of hydraulic supports ranges from 50 to 200, and there are multiple contact sensors on the same support, collecting data transmission and fusion are difficult and there are cumulative errors; the process of further judging the operating status of the hydraulic support guard plate through sensor detection data is complicated and the real-time performance is limited.

[0056] Embodiment 1

[0057] See attached Figure 1 , Figure 5 As shown, this embodiment discloses a visual state measurement system of a guard plate based on deep learning, comprising a first hydraulic support 1, a second hydraulic support 2, a third hydraulic support 3, a guard plate 4, a left-facing station camera 5, a right-facing station camera 6, and a positive light source 7. The guard plate 4 is a target to be detected, and the positive light source 7 irradiates light on the guard plate 4, which mainly serves to improve the clarity of the captured image.

[0058] In this implementation example, the left-facing station camera is set at the lower left of the hydraulic support; the right-facing station camera is set at the lower right of the hydraulic support; the positive light source is set at the lower right of the hydraulic support to provide lighting for the camera shooting. The side station cameras shoot the guard plates of the adjacent hydraulic support, and the station cameras configured at the two end supports shoot the guard plates of the current support.

[0059] The field of view of the left-facing workstation camera 5 on the hydraulic support 3 is the guard plate 4 on the hydraulic support 2, and the field of view of the right-facing workstation camera 6 on the hydraulic support 1 is the guard plate 4 on the hydraulic support 2. The guard plate 4 is photographed by cross-shooting on the left and right adjacent sides.

[0060] See attached Figure 2 As shown, the visual inspection system also includes a big data management and control platform, a computer and a cloud server;

[0061] The data control platform is used to uniformly transmit and store the downhole visual measurement results, and the model and data processing are performed in the computer;

[0062] The big data management and control platform stores the first collected image and the second collected image of the side guard plate and the status information of the side guard plate;

[0063] Both computers and cloud servers communicate with the big data management and control platform.

[0064] Among them, the computer will alarm and shut down in time when the side station camera captures the abnormal working status of the guard plate.

[0065] When the above system is working, see the attached Figure 3 As shown, by adjusting the light source and the camera field of view, the corresponding first captured image and the second captured image are obtained. The computer stores a trained deep learning model for detecting the working status of the guard plate and determining whether its working status is consistent with the preset status. If it is consistent, the information is uploaded to the cloud platform, otherwise, an alarm is issued.

[0066] Embodiment 2

[0067] This embodiment discloses a visual inspection method for side guard plates based on a deep learning model, and the steps are as follows:

[0068] Step 1: Under the action of light source, the guard plate to be inspected is photographed by the workstation cameras on the guard plates on both sides adjacent to each other to obtain the current working state image of the guard plate;

[0069] Step 2: Process the image collected in step 1 to detect the working status of the side guard plate;

[0070] Step 3: Perform weighted processing on the detection results of the workstation cameras on the side guard plates on both sides to determine the final working state of the side guard plates;

[0071] Step 4: Determine whether there is any abnormality in the working state of the side guard plate according to the weighted detection result of step 3;

[0072] Step 5: Store the collected images and the obtained status information in the big data management and control platform, and perform remote real-time monitoring, maintenance and parameter adjustment through the cloud server.

[0073] In this implementation example, the image processing process of the collected image in step 2 is as follows:

[0074] The image obtained in step 1 is processed by grayscale to perform feature screening and remove redundant information to obtain a grayscale image of the guard plate;

[0075] After obtaining the grayscale image of the guard plate, the ResNet deep learning model is used to extract the feature information in the image.

[0076] The feature map obtained in step 2 is subjected to feature analysis, feature fusion and feature detection, including feature extraction through image convolution layer and semantic compression pooling layer, decoding compression information of the output feature map through upsampling operation, linking feature maps of different scales to realize context feature fusion, using CNN and sliding anchor box method to detect targets, and using NMS operator to remove duplicate detection targets.

[0077] The final detection results obtained in step three are classified and evaluated using a weighted algorithm.

[0078] For more details, see Figure 4 , 6 , the visual inspection system of this embodiment is used to detect the state of the hydraulic support guard plate, wherein the training of the deep learning model and the detection process of the hydraulic support guard plate state are as follows:

[0079] S1. Adjust the light intensity and irradiation angle;

[0080] Adjust the light intensity and light irradiation angle of the positive light source 7 to ensure that the captured image does not have reflection, overexposure, glare and halo problems, and ensure that the captured image is clear and the status characteristics of the side guard plate 4 are obvious.

[0081] S2, adjust the camera shooting field of view;

[0082] See attached Figure 7 As shown, adjust the right-facing workstation camera 6 on the hydraulic support 1 so that its shooting field of view includes all parts of the guard plate 4, the first gap 8 between the left side of the guard plate 4 and the coal wall 7, and the gap 9 between the left side of the guard plate 4 and the top beam of the hydraulic support 2.

[0083] Adjust the left-facing workstation camera 5 on the hydraulic support 3 so that its shooting field of view includes all parts of the guard plate 4, the gap between the right side of the guard plate 4 and the coal wall, and the gap between the right side of the guard plate 4 and the top beam of the hydraulic support 2.

[0084] After adjusting the light intensity and the illumination angle and adjusting the camera shooting field of view, image acquisition is performed to obtain the acquired images, a data set is constructed based on the acquired images, and a model is trained based on the constructed data set to obtain a trained model;

[0085] S3. Check the working status of the side guard

[0086] The detected image is obtained again and input into the trained model. The trained model uses the state detection algorithm of deep learning to extract features and detect the gaps in the image and the working state of the guard plate. The purpose of detecting the gap is to comprehensively judge the working state of the guard plate. The left-facing station camera 5 judges the working characteristics of the guard plate 4, and comprehensively identifies the gap between the right side of the guard plate 4 and the coal wall and the gap between the top beam of the hydraulic support 2. The right-facing station camera 6 judges the working characteristics of the guard plate 4, and comprehensively identifies the gap between the left side of the guard plate 4 and the coal wall and the gap between the top beam of the hydraulic support 2.

[0087] S4, the six data collected in step 3, namely, the left and right working states, the gap between the coal wall, and the gap between the hydraulic support top beam, are integrated and processed, and the actual working state of the guard plate is finally determined according to the weight formula. In the weight formula, x1, x2, and x3 respectively represent the detection results of the gap between the guard plate 4 and the coal wall, the detection results of the guard plate 4 and the hydraulic support top beam, and the detection results of the working characteristics of the guard plate.

[0088]

[0089] Among them, R, X i ∈(-1, 0, 1), -1 means no, 1 means yes, and 0 means the target is not detected.

[0090] S5. Store the collected images and the obtained working status information in the cloud to analyze whether cutting interference will occur, which is convenient for subsequent status statistical analysis, dangerous condition cause analysis, and historical data query.

[0091] See also Figure 4 The specific process of image processing the collected guard plate working pictures during the establishment of the data set in this embodiment is as follows:

[0092] The camera takes underground images and performs image preprocessing, including random cropping, image color processing, and random rotation of the image setting angle. The first round of image acquisition and processing is realized, and a guard plate status detection data set is established. The data set includes the support state, retraction state, operation state, top beam gap, and coal wall gap of the guard plate. Its main purpose is to train a deep learning model for detection, and then the first round of data set collected is processed by image enhancement. The main operations include: random cropping at a scale of 640*640, random selection of RGB values, arbitrary rotation of the image angle by 15°, manual image annotation after image enhancement, and establishment of a deep learning data set.

[0093] The established deep learning dataset is put into the target detection model YOLOv5s for model training, where the key training parameters are set to the learning rate Learningrate to 0.012, the training round epoch to 500, and the training image input size Batchsize to 64.

[0094] Put the newly collected images of the guard plate working status into the trained deep learning model.

[0095] Specifically, the dataset is loaded into the deep learning model for forward propagation and reverse gradient descent, the model is trained, feature extraction is performed through the image convolution layer and the semantic compression pooling layer, the output feature map is decoded and compressed through an upsampling operation, and feature maps of different scales are linked to achieve contextual feature fusion.

[0096] The deep learning model uses the YOLOv5s model. The YOLOv5s model detection process is as follows:

[0097] The image of the guard plate in working state is put into the feature extraction part of the trained deep learning model. The image enters the feature extraction network in RGB three-channel format. The image feature extraction method of convolution and pooling finally outputs the feature map of 20*20*1024, where 1024 is the number of feature maps finally extracted. The working state features of the guard plate image are fully extracted through the feature extraction network part, and these image features are saved for the next state detection part.

[0098] The output results of the feature extraction part are fused with feature maps of different scales of 80*80, 40*40, and 20*20 and the feature maps to be detected through convolution and upsampling, and put into different model detection parts for target detection, and the duplicate detection results are eliminated through the NMS non-maximum suppression algorithm.

[0099] By fusing feature maps of different scales, the context feature information extracted from the image can be fused, and object detection at different scales can improve detection accuracy. The model directly outputs an image with the status of each position marked on the image.

[0100] Analyze the test results at different angles and put them into the test result weight formula to determine the final working status of the side guard.

[0101] Although the above describes the specific implementation mode of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without creative work are still within the scope of protection of the present invention.

Claims

1. A visual state measurement system for a guard plate based on deep learning, characterized in that: include: An image acquisition unit and a visual inspection system, wherein the image acquisition unit transmits the acquired image to the visual inspection system; The image acquisition unit includes: a first image acquisition unit arranged below the first hydraulic support, a light source arranged below the second hydraulic support, and a second image acquisition unit arranged below the third hydraulic support; The second hydraulic support is provided with a side guard plate; The light source is facing the guard plate and is used to provide light intensity and light irradiation angle for measuring the guard plate; The first image acquisition unit and the second image acquisition unit photograph the support plate by cross-photographing the left and right adjacent sides to obtain the first acquired image and the second acquired image; The visual inspection system is configured to: receive the first acquired image and the second acquired image, and transmit them to the deep learning model for feature extraction, monitor the status of the guard plate based on the extracted features, determine whether its working status is consistent with the preset status, and issue an early warning if it is not consistent.

2. A deep learning-based visual state measurement system for a side guard as claimed in claim 1, characterized in that: The first hydraulic support, the second hydraulic support and the third hydraulic support are arranged in sequence and have the same structure, and all include a bottom plate and a top plate, a connecting column is arranged between the bottom plate and the top plate, and a guard plate is arranged at the suspended end of the top plate of the second hydraulic support.

3. The deep learning-based visual state measurement system for a guard plate as claimed in claim 1, characterized in that: The visual inspection system includes a big data management and control platform, a computer and a cloud server; The big data management and control platform stores the first collected image and the second collected image of the side guard plate and the status information of the side guard plate; The computer stores a trained deep learning model; The computer and the cloud server both communicate with the big data management and control platform.

4. A method for measuring the visual state of a guard plate based on deep learning, characterized in that: include: The guard plate is photographed by cross-shooting the left and right adjacent sides to obtain a first acquisition image and a second acquisition image; The first acquired image and the second acquired image are transmitted to the deep learning model for processing to obtain the working state of the side guard plate; The obtained working states of the side guard plate are weighted to determine the final working state of the side guard plate and judge whether there is any abnormality in the working state of the side guard plate.

5. The method for measuring the visual state of a guard plate based on deep learning as claimed in claim 4, characterized in that: The first acquired image and the second acquired image are transmitted to the deep learning model for processing, including: Performing grayscale processing on the first acquired image and the second acquired image respectively to obtain a first side guard plate grayscale image and a second side guard plate grayscale image; Based on the grayscale image of the first side guard plate, the gap between the right side of the side guard plate and the coal wall and the gap between the top beam of the hydraulic support are identified and judged to obtain the working state of the side guard plate; Based on the grayscale image of the first side guard plate, the gap between the left side of the side guard plate and the coal wall and the gap between the top beam of the hydraulic support are identified and judged to obtain the working status of the side guard plate.

6. The method for measuring the visual state of a guard plate based on deep learning as claimed in claim 4, characterized in that: The six types of data information, including the left and right working states of the guard plate, the gap between the guard plate and the coal wall, and the gap between the guard plate and the top beam of the hydraulic support, are integrated and processed, and the actual working state of the guard plate is finally determined according to the weight formula.

7. The method for measuring the visual state of a guard plate based on deep learning as claimed in claim 4, characterized in that: The specific calculation formula is: Among them, R, X i ∈(-1,0,1), -1 means no, 1 means yes, and 0 means no target is detected. In the weight formula, X1, X2, and X3 represent the detection results of the gap between the guard plate and the coal wall, the detection results of the guard plate and the top beam of the hydraulic support, and the detection results of the working characteristics of the guard plate, respectively.

8. The method for measuring the visual state of a guard plate based on deep learning as claimed in claim 4, characterized in that: The training process of the deep learning model is: Establishing a guard plate state detection dataset based on the first collected image and the second collected image, and performing enhancement processing on the guard plate state detection dataset to obtain a deep learning dataset; The deep learning data set is used to train a deep learning model for detection to obtain a trained deep learning model.

9. The method for measuring the visual state of a guard plate based on deep learning as claimed in claim 4, characterized in that: The first acquired image and the second acquired image include: all parts of the side guard plate, the first gap between the left side of the side guard plate and the coal wall, and the gap between the left side of the side guard plate and the top beam of the hydraulic support.

10. The method for measuring the visual state of a guard plate based on deep learning as claimed in claim 4, characterized in that: The first acquired image is used to judge the working characteristics of the side guard plate, and the gap between the right side of the side guard plate and the coal wall and the gap between the top beam of the hydraulic support are comprehensively identified for judgment; The second acquired image is used to judge the working characteristics of the guard plate, and the judgment is made by comprehensively identifying the gap between the left side of the guard plate and the coal wall and the gap between the top beam of the hydraulic support.

Images