Method for identifying cutting interference state of roller of coal mining machine based on infrared image

Abstract

The invention discloses a coal cutter roller cutting interference state identification method based on an infrared image, and the method employs an infrared camera to shoot the surrounding environment, effectively prevents the recognition precision from being limited by complex illumination environments, such as underground low illumination, high dust, and strong water mist, and guarantees the stability and definition of a target contour. Then analyzing and processing the infrared image through a deep learning network, determining four corner points of the hydraulic support face guard in the area from the infrared image by adopting a Shii-Tomasi corner point detection algorithm, further accurately calculating the spatial attitude of the hydraulic support face guard, and constructing a three-dimensional trajectory model of the coal mining machine roller; the Kalman filtering is adopted to obtain the predicted poses of the two, and then the cutting interference state is judged through early warning grading, so that the response speed and the safety guarantee capability of the fully-mechanized coal mining system are remarkably improved, the dynamic prediction and real-time response of the cutting interference state are realized, and the efficiency and the precision of cutting interference state recognition are ensured.

Description

Technical Field

[0001] This invention belongs to the field of coal mine safety production and intelligent fully mechanized mining equipment, and particularly relates to a method for identifying the interference state of coal mining machine drum cutting based on infrared images. Background Technology

[0002] As the core operational area of ​​coal production, the safety of the coordinated operation of equipment in the fully mechanized mining face directly affects the efficiency of coal mine production and personnel safety. In the "three-machine" system of fully mechanized mining, the coal shearing machine is responsible for cutting the coal face, while the hydraulic support is used to support the roof and coordinate with the coal mining process to complete the support movement. To achieve automated and efficient coal mining, the spatial relationship between the coal shearing machine drum and the hydraulic support sidewall plates must be strictly controlled. If the two are in the same spatial position at the same time, cutting interference will occur, which will cause equipment damage and may even lead to serious safety accidents, ultimately affecting the stable operation of the entire working face.

[0003] However, due to problems such as coal dust, water mist interference, and strong local glare in underground fully mechanized mining faces, traditional visual recognition methods struggle to acquire clear images in complex environments. This results in insufficient accuracy in recognizing the spatial pose of hydraulic support sidewalls and coal mining machine drums, failing to meet the requirements for real-time performance and accuracy. Furthermore, even when existing methods acquire data using non-visible light, subsequent recognition methods can roughly determine the relative position and shape of targets. However, limitations in image quality and algorithms make it difficult to integrate angle information to construct a complete three-dimensional spatial relationship model. Especially during the dynamic changes in the rotation angle of the coal mining machine drum, it is impossible to accurately capture the dynamic pose evolution characteristics between equipment. Ultimately, current methods are unable to achieve dynamic prediction and real-time response to cutting interference states, thus restricting the safe operation efficiency and intelligence level of fully mechanized mining automation systems. Summary of the Invention

[0004] To address the problems existing in the prior art, this invention provides a method for identifying the interference state of a coal mining machine drum cutting based on infrared images. The method uses infrared detection to acquire infrared images, reducing the impact of the environment on image acquisition. Through a specific recognition algorithm, it can accurately capture the dynamic pose evolution characteristics between devices, thereby achieving dynamic prediction and real-time response to the cutting interference state.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is: a method for identifying the interference state of a coal mining machine drum cutting based on infrared images, comprising the following steps: Step 1: Use an infrared imager to continuously photograph the operating area of ​​the coal mining machine drum and the hydraulic support side plate to obtain the corresponding infrared images.

[0006] Step 2: Use a deep neural network to segment the hydraulic support side plate area from the infrared image obtained in Step 1.

[0007] Step 3: The Shi-Tomasi corner detection algorithm is used to analyze the gradient changes in the hydraulic support side plate area obtained in Step 2. Combined with the minimum eigenvalue criterion, points with significant edge changes are found in the image, and finally the four corner points of the hydraulic support side plate in the area are determined.

[0008] Step 4: Calibrate the infrared imager to obtain its intrinsic parameter matrix and distortion coefficients.

[0009] Step 5: Use the PNP algorithm to establish the projection relationship of 2D-3D point pairs. Combine the intrinsic parameter matrix and distortion coefficients from Step 4 with the four corner points determined in Step 3 to obtain the initial pose of the hydraulic support side plate in the infrared imager coordinate system.

[0010] Step 6: Use an optimization algorithm to iteratively optimize the initial pose of the hydraulic support side guard plate to obtain the optimized pose of the hydraulic support side guard plate.

[0011] Step 7: Convert the optimized hydraulic support side plate pose to the world coordinate system, and at the same time obtain the pose of the coal mining machine drum in the world coordinate system.

[0012] Step 8: Set early warning classification criteria to identify and judge whether the hydraulic support side plate and the coal mining machine drum will cause cutting interference. Then, use Kalman filtering to predict the pose of the hydraulic support side plate and the coal mining machine drum, and identify and judge whether the two will cause cutting interference in the future based on the predicted pose.

[0013] Furthermore, in step one, the acquired infrared image undergoes denoising processing, specifically as follows: First, the infrared image is decomposed into multiple levels using wavelet transform to separate low-frequency subject information from high-frequency noise components; then, the wavelet coefficients are thresholded using linear minimum mean square error estimation, and the threshold function is optimized by combining inter-scale and intra-scale correlations to remove Gaussian noise while preserving edge texture; next, an adaptive morphological weighting algorithm is used to handle verification noise and background blurring issues, removing isolated noise points through opening and closing operations, while adjusting the size of the structuring element to avoid damaging the target area. For stripe noise caused by uneven sensor response, an L2,1 norm optimization function is finally constructed based on a total variational model, and solved iteratively using the alternating direction multiplier method (ADMM) to effectively separate periodic noise and restore image details, obtaining the denoised infrared image.

[0014] Furthermore, step two specifically involves: first, constructing an infrared image dataset of the hydraulic support side panel; using the LabelImg tool to annotate the ground truth bounding boxes; and improving the model's generalization ability through data augmentation. A deep neural network model based on the YOLOv8 architecture is trained, adjusting the input layer size and network parameters to adapt to the characteristics of infrared images. A CIoU loss function is designed to optimize the bounding box regression accuracy. After training, the denoised infrared image is input into the model, outputting the bounding box coordinates of the hydraulic support side panel, marking the approximate location of the hydraulic support side panel in the infrared image, and delineating the hydraulic support side panel region, which is called the Region of Interest (ROI). Finally, the IoU metric is used to verify the overlap between the ROI and the ground truth bounding boxes to ensure positioning accuracy.

[0015] Furthermore, step three specifically involves: first, adjusting the size and standard deviation σ of the Gaussian kernel to control the smoothness of gradient calculation and balance noise suppression and edge preservation; second, setting a corner response threshold Rthreshold, retaining only pixels with R ≥ Rthreshold as candidate corners; finally, introducing a non-maximum suppression step to remove redundant corners with small response values ​​in the neighborhood, ensuring that each significant corner is retained only once; ultimately, the four corners of the hydraulic support side plate in the region are determined. In the detection of the four corners of the hydraulic support side plate, non-maximum suppression can effectively eliminate false corners caused by edge burrs or local noise, improving detection accuracy.

[0016] Furthermore, in step four, an infrared calibration plate is used to calibrate the infrared imager to obtain its intrinsic parameter matrix and distortion coefficients, and to obtain the rotation matrix R and translation vector t of the infrared imager relative to the world coordinate system.

[0017] Further, step five specifically involves: establishing the projection relationship between 2D and 3D point pairs using the PNP algorithm, calculating the initial pose of the infrared imager relative to the hydraulic support side plate, including the rotation matrix R and the translation vector t; then establishing the projection relationship between the four corner points of the hydraulic support side plate in three-dimensional space and the four corner points determined by the two-dimensional infrared image through the intrinsic parameter matrix K of the infrared imager, and obtaining the six-degree-of-freedom pose of the hydraulic support side plate in the coordinate system of the infrared imager by solving the initial rotation matrix R or converting the rotation matrix R into Euler angles (α, β, γ) and the translation vector t = (x, y, z).

[0018] Furthermore, in step seven, after obtaining the poses of the coal mining machine drum and the hydraulic support side plate in the world coordinate system, the hydraulic support side plate is simplified into a rectangle or a combination of multiple planes, and the drum is simplified into a cylinder. Then, based on the simplified shapes, the relative attitude angle Δθ and the minimum spatial distance d between the hydraulic support side plate and the coal mining machine drum are calculated. min .

[0019] Furthermore, step eight specifically involves: setting a safe distance d between the hydraulic support side guard plate and the coal mining machine drum. safe The safety angle between the hydraulic support side guard plate and the coal mining machine drum is θ. safe The warning distance between the hydraulic support side guard plate and the coal mining machine drum is d. warn The warning angle between the hydraulic support side guard plate and the coal mining machine drum is θ. warn ;d warn and θ warn The value is determined based on the actual working conditions; the warning level is divided into safety levels: d min >d safe And Δθ < θ safe Warning level: d warn <d min ≤d safe or θ safe ≤Δθ<θ warn Interference level: d min ≤d warn Or Δθ≥θ warn The relative attitude angle Δθ and minimum spatial distance d obtained in step seven are combined. min Substituting the aforementioned warning levels, the cutting interference state between the hydraulic support sidewall and the coal mining machine drum under the current condition is obtained; then, Kalman filtering is used to obtain the predicted poses of both, thereby obtaining the predicted relative attitude angle Δθ and the minimum spatial distance d. min Substituting the above warning levels, we can obtain the cutting interference state between the hydraulic support side plate and the coal mining machine drum under the predicted conditions.

[0020] Compared with the prior art, the present invention has the following advantages: 1. This invention uses an infrared camera to capture images of the surrounding environment. Taking advantage of the fact that infrared images are not limited by visible light conditions, it overcomes the limitations on recognition accuracy caused by complex lighting environments such as low illumination, high dust, and strong water mist underground, ensuring the stability and clarity of the target outline.

[0021] 2. After obtaining infrared images, this invention combines deep learning networks with infrared image analysis, and uses the Shi-Tomasi corner detection algorithm to determine the four corner points of the hydraulic support side plate in the region from the infrared images. This allows for precise calculation of the spatial attitude of the hydraulic support side plate and the construction of a three-dimensional trajectory model of the coal mining machine drum. This enables high-precision judgment and dynamic prediction of the spatial relationship between the two. In this process, Kalman filtering is used to obtain the predicted poses of both, and the cutting interference state is determined through early warning level classification. This significantly improves the response speed and safety assurance capabilities of the fully mechanized mining system, achieving dynamic prediction and real-time response to the cutting interference state, and ensuring the efficiency and accuracy of cutting interference state identification. Attached Figure Description

[0022] Figure 1 This is an overall flowchart of an embodiment of the present invention.

[0023] Figure 2 This is a schematic diagram of multi-level truncation interference identification in an embodiment of the present invention.

[0024] Figure 3 This is a schematic diagram of the recognition results in an embodiment of the present invention. Detailed Implementation

[0025] The present invention will be further described below.

[0026] like Figure 1 As shown, the present invention includes the following steps: S1. An infrared imager is used to continuously capture images of the operating area of ​​the coal mining machine drum and hydraulic support sidewalls, obtaining corresponding infrared images. The acquired infrared images are then denoised. Specifically: First, wavelet transform is used to decompose the infrared images into multiple levels, separating low-frequency main information from high-frequency noise components. Next, linear minimum mean square error estimation is used to threshold the wavelet coefficients, and the threshold function is optimized by combining inter-scale and intra-scale correlations to remove Gaussian noise while preserving edge texture. Then, an adaptive morphological weighting algorithm is used to handle verification noise and background blurring. Isolated noise points are removed through opening and closing operations, while the size of the structuring element is adjusted to avoid damaging the target area. For stripe noise caused by uneven response of the infrared imager, an L2,1 norm optimization function is constructed based on a total variational model. This function is then iteratively solved using the alternating direction multiplier method (ADMM) to effectively separate periodic noise and restore image details, obtaining the denoised infrared image.

[0027] S2. A deep neural network is used to segment the hydraulic support sidewall region from the infrared images obtained in Step 1. Specifically, the following steps are taken: First, an infrared image dataset of the hydraulic support sidewall is constructed. The LabelImg tool is used to annotate the ground truth bounding boxes, and data augmentation is used to improve the model's generalization ability. A deep neural network model based on the YOLOv8 architecture is trained, and the input layer size and network parameters are adjusted to adapt to the characteristics of infrared images. A CIoU loss function is designed to optimize the bounding box regression accuracy. After training, the denoised infrared image is input into the model, and the bounding box coordinates of the hydraulic support sidewall are output, marking the approximate position of the hydraulic support sidewall in the infrared image and segmenting the hydraulic support sidewall region, which is called the Region of Interest (ROI). Finally, the overlap between the ROI and the ground truth bounding boxes is verified using the IoU metric to ensure positioning accuracy.

[0028] S3. The Shi-Tomasi corner detection algorithm is used to analyze the gradient changes in the hydraulic support sideplate region obtained in step two. Combined with the minimum eigenvalue criterion, points with significant edge changes are searched in the image to ultimately determine the four corner points of the hydraulic support sideplate in the region. Specifically: First, the smoothness of the gradient calculation is controlled by adjusting the size and standard deviation σ of the Gaussian kernel, balancing noise suppression and edge preservation. Second, a corner response threshold Rthreshold is set, retaining only pixels with R ≥ Rthreshold as candidate corner points. Finally, a non-maximum suppression step is introduced to remove redundant corner points with small response values ​​in the neighborhood, ensuring that each significant corner point is retained only once. This ultimately determines the four corner points of the hydraulic support sideplate in the region. In the detection of the four corner points of the hydraulic support sideplate, non-maximum suppression effectively eliminates false corner points caused by edge burrs or local noise, improving detection accuracy.

[0029] S4. Calibrate the infrared imager to obtain its intrinsic parameter matrix and distortion coefficients. Specifically, determine the geometric parameters (i.e., intrinsic and extrinsic parameters) of the infrared imager through experiments and calculations, thereby establishing a mapping relationship between pixel coordinates in the image and three-dimensional world coordinates. The intrinsic parameters describe the physical characteristics of the infrared imager itself. Optical distortion of the infrared imager lens can cause barrel or pincushion distortion at the image edges. After calibration, the true geometric shape of the image can be restored through distortion correction, including the focal length (f). x f y ), principal point coordinates (c x c yThe extrinsic parameters, such as pixel size and distortion coefficients, describe the position and orientation of the infrared imager in three-dimensional space and are necessary parameters for converting image coordinates to world coordinates. These include the rotation matrix R and translation vector t of the infrared imager relative to the world coordinate system. The distortion parameters obtained through calibration can be used to correct image distortion, making the image closer to an ideal perspective projection. The calibration results directly affect the correspondence between points in three-dimensional space and points in the image.

[0030] S5. Using the PNP algorithm, establish the projection relationship between 2D and 3D point pairs, and calculate the initial pose of the infrared imager relative to the hydraulic support guard plate, including the rotation matrix R and translation vector t. Then, establish the projection relationship between the four corner points of the hydraulic support guard plate in 3D space and the four corner points determined by the 2D infrared image through the intrinsic parameter matrix K of the infrared imager. By solving for the initial rotation matrix R or converting the rotation matrix R into Euler angles (α, β, γ) and the translation vector t = (x, y, z), obtain the six-degree-of-freedom pose of the hydraulic support guard plate in the infrared imager coordinate system, as shown below. Figure 3 As shown.

[0031] S6. Due to infrared image noise, feature point detection errors, or calibration deviations, the initial pose may contain significant deviations. The position of the corner points of the support plate in the image may be offset by pixel coordinates due to dust interference, or lens distortion may not be completely eliminated during infrared imager calibration. All of these factors can lead to inaccurate initial estimates of the rotation matrix R and translation vector t. To improve accuracy, the Levenberg-Marquardt (LM) optimization algorithm is used to iteratively optimize the initial pose of the hydraulic support support plate. The goal is achieved by minimizing the reprojection error E of all feature points, thus obtaining the optimized pose of the hydraulic support support plate. S7. In the interference judgment of the cutting, the pose of the hydraulic support side plate and the motion trajectory of the coal mining machine drum need to be unified to the world coordinate system to eliminate the differences between different coordinate systems. The transformation relationship between the infrared imager and the coal mining machine body coordinate system is obtained through calibration, and the pose of the hydraulic support side plate is transformed from the infrared imager coordinate system to the world coordinate system. The position of the coal mining machine drum can be obtained in real time by the inertial measurement unit (IMU) and sensors on the rocker arm of the coal mining machine. Its attitude is obtained by integrating the three-axis angular velocity and acceleration of the IMU. If the outline of the coal mining machine drum is captured by the infrared imager, the rotation matrix R and translation vector t of the coal mining machine drum are calculated by combining the PnP method with the feature point extraction of the infrared image. Then, the hydraulic support side plate is simplified into a rectangle or a combination of multiple planes, while the drum is simplified into a cylinder. Then, based on the simplified shape, the relative attitude angle Δθ and the minimum spatial distance d between the hydraulic support side plate and the coal mining machine drum are calculated. min .

[0032] S8. Set early warning classification standards; the safe distance between the hydraulic support side guard plate and the coal mining machine drum is d. safe The safety angle between the hydraulic support side guard plate and the coal mining machine drum is θ. safe The warning distance between the hydraulic support side guard plate and the coal mining machine drum is d. warn The warning angle between the hydraulic support side guard plate and the coal mining machine drum is θ. warn In this embodiment, d warn The value is d safe 1 / 5 of the value, θ warn The value is θ safe Twice the value; the warning level is divided into safety levels: d min >d safe And Δθ < θ safe This embodiment emits green light; Warning level: d warn <d min ≤d safe or θ safe ≤Δθ<θ warn This embodiment emits yellow light; interference order: d min ≤d warn Or Δθ≥θ warn In this embodiment, red light is emitted; the relative attitude angle Δθ and minimum spatial distance d obtained in step seven are used. min Substituting the above warning levels, we can obtain the cutting interference state between the hydraulic support side plate and the coal mining machine drum under the current conditions.

[0033] like Figure 2 As shown, the design of the prevention mechanism needs to be based on the predicted output of the Kalman filter rather than the original frame data to avoid misjudgment caused by transient noise. When the motion trajectory of the hydraulic support side plate suddenly changes due to the speed adjustment of the coal mining machine drum, the pose estimation of the original frame may have a brief deviation. However, the Kalman filter, after smoothing, outputs a more continuous predicted trajectory, which can identify potential interference several seconds in advance. Therefore, the Kalman filter is used to obtain the predicted poses of both, and then the predicted relative attitude angle Δθ and minimum spatial distance d are obtained. min Substituting the above warning levels, we can obtain the cutting interference state between the hydraulic support side plate and the coal mining machine drum under the predicted conditions.

[0034] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for identifying the interference state of a coal mining machine drum cutting based on infrared images, characterized in that, Includes the following steps: Step 1: Use an infrared imager to continuously photograph the operating area of ​​the coal mining machine drum and hydraulic support side guard plate to obtain corresponding infrared images; Step 2: Use a deep neural network to segment the hydraulic support side plate area from the infrared image obtained in Step 1; Step 3: The Shi-Tomasi corner detection algorithm is used to analyze the gradient change of the hydraulic support side plate area obtained in Step 2. Combined with the minimum eigenvalue criterion, points with significant edge changes are found in the image, and finally the four corner points of the hydraulic support side plate in the area are determined. Step 4: Calibrate the infrared imager to obtain its intrinsic parameter matrix and distortion coefficients; Step 5: Use the PNP algorithm to establish the projection relationship of 2D-3D point pairs. Combine the intrinsic parameter matrix and distortion coefficients from Step 4 with the four corner points determined in Step 3 to obtain the initial pose of the hydraulic support side plate in the infrared imager coordinate system. Step 6: Use an optimization algorithm to iteratively optimize the initial pose of the hydraulic support side guard plate to obtain the optimized pose of the hydraulic support side guard plate. Step 7: Convert the optimized hydraulic support side plate pose to the world coordinate system, and at the same time obtain the pose of the coal mining machine drum in the world coordinate system; Step 8: Set early warning classification criteria to identify and judge whether the hydraulic support side plate and the coal mining machine drum will cause cutting interference. Then, use Kalman filtering to predict the pose of the hydraulic support side plate and the coal mining machine drum, and identify and judge whether the two will cause cutting interference in the future based on the predicted pose.

2. The method for identifying the interference state of a coal mining machine drum cutting based on infrared images according to claim 1, characterized in that, In step one, the acquired infrared image is denoised. Specifically, the infrared image is first decomposed into multiple levels using wavelet transform to separate low-frequency main information from high-frequency noise components. Then, the wavelet coefficients are thresholded using linear minimum mean square error estimation. Next, isolated noise points are removed using opening and closing operations. Finally, an L2,1 norm optimization function is constructed based on the total variational model and solved iteratively using the alternating direction multiplier method to effectively separate periodic noise and restore image details, thus obtaining the denoised infrared image.

3. The method for identifying the interference state of a coal mining machine drum cutting based on infrared images according to claim 1, characterized in that, The second step is as follows: First, construct an infrared image dataset of the hydraulic support side plate, train a deep neural network model with YOLOv8 architecture, adjust the input layer size and network parameters to adapt to the characteristics of infrared images, and after training, input the denoised infrared image into the model, output the bounding box coordinates of the hydraulic support side plate, mark the approximate position of the hydraulic support side plate in the infrared image, and delineate the region of the hydraulic support side plate.

4. The method for identifying the interference state of a coal mining machine drum cutting based on infrared images according to claim 1, characterized in that, Step three specifically involves: first, adjusting the size and standard deviation σ of the Gaussian kernel to control the smoothness of gradient calculation and balance noise suppression and edge preservation; second, setting a corner response threshold Rthreshold to retain only pixels with R≥Rthreshold as candidate corners; finally, introducing a non-maximum suppression step to remove redundant corners with small response values ​​in the neighborhood, ensuring that each significant corner is retained only once; and finally determining the four corners of the hydraulic support side plate in the region.

5. The method for identifying the interference state of a coal mining machine drum cutting based on infrared images according to claim 1, characterized in that, In step four, an infrared calibration plate is used to calibrate the infrared imager, obtain its intrinsic parameter matrix and distortion coefficients, and obtain the rotation matrix R and translation vector t of the infrared imager relative to the world coordinate system.

6. The method for identifying the interference state of a coal mining machine drum cutting based on infrared images according to claim 1, characterized in that, Step five specifically involves: establishing the projection relationship between 2D and 3D point pairs using the PNP algorithm, calculating the initial pose of the infrared imager relative to the hydraulic support guard plate, including the rotation matrix R and the translation vector t; then establishing the projection relationship between the four corner points of the hydraulic support guard plate in three-dimensional space and the four corner points determined by the two-dimensional infrared image through the intrinsic parameter matrix K of the infrared imager, and obtaining the six-degree-of-freedom pose of the hydraulic support guard plate in the coordinate system of the infrared imager by solving the initial rotation matrix R or converting the rotation matrix R into Euler angles (α, β, γ) and the translation vector t = (x, y, z).

7. The method for identifying the interference state of a coal mining machine drum cutting based on infrared images according to claim 1, characterized in that, Step seven involves obtaining the poses of the coal mining machine drum and the hydraulic support side plate in the world coordinate system. The hydraulic support side plate is simplified into a rectangle or a combination of multiple planes, and the drum is simplified into a cylinder. Then, based on the simplified shapes, the relative attitude angle Δθ and the minimum spatial distance d between the hydraulic support side plate and the coal mining machine drum are calculated. min .

8. The method for identifying the interference state of a coal mining machine drum cutting based on infrared images according to claim 7, characterized in that, Step eight specifically involves: setting the safe distance between the hydraulic support side guard plate and the coal mining machine drum as d. safe The safety angle between the hydraulic support side guard plate and the coal mining machine drum is θ. safe The warning distance between the hydraulic support side guard plate and the coal mining machine drum is d. warn The warning angle between the hydraulic support side guard plate and the coal mining machine drum is θ. warn The warning level is classified into a safety level: d min >d safe And Δθ < θ safe Warning level: d warn <d min ≤d safe or θ safe ≤Δθ<θ warn Interference level: d min ≤d warn Or Δθ≥θ warn The relative attitude angle Δθ and minimum spatial distance d obtained in step seven are combined. min Substituting the aforementioned warning levels, the cutting interference state between the hydraulic support sidewall and the coal mining machine drum under the current condition is obtained; then, Kalman filtering is used to obtain the predicted poses of both, thereby obtaining the predicted relative attitude angle Δθ and the minimum spatial distance d. min Substituting the above warning levels, we can obtain the cutting interference state between the hydraulic support side plate and the coal mining machine drum under the predicted conditions.

Images