Coal quality dynamic detection method and system based on image recognition mining face

Abstract

The present invention relates to the technical field of coal quality identification and monitoring, and discloses a method and system for dynamic detection of coal quality based on image identification of mining face, wherein the method comprises: in the process of mining operation by a coal mining machine, obtaining images of the first, second and third regions from top to bottom of the mining face; extracting bedding features from the first region image and the second region image to obtain a bedding line distribution map; calculating the maximum interlayer spacing and the ordinate variance based on the bedding line distribution map; determining the uniformity of the coal seam based on the maximum interlayer spacing; determining the flatness of the coal seam based on the ordinate variance; obtaining the quality of the coal seam based on the uniformity of the coal seam and the flatness of the coal seam; extracting geometric features from the third region image and generating a bright spot distribution map; calculating the average area of ​​the bright spots based on the bright spot distribution map; determining the degree of crushing of the coal rock according to the average area of ​​the bright spots, and then determining the quality of the coal rock. The present invention can analyze the quality of the coal seam and the quality of the coal rock.

Description

Technical Field

[0001] The present invention relates to the technical field of coal quality identification and monitoring, and in particular to a method and system for dynamic detection of coal quality based on image recognition of mining faces. Background Art

[0002] During coal mining operations, coal quality changes dynamically with mining. Even for the same coalfield, the distribution of coal quality can be uneven. Therefore, in order to formulate corresponding mining work plans for different coal qualities, implement corresponding safety management measures, select appropriate raw coal processing methods, and improve production efficiency; at the same time, make preliminary judgments on the mined coal quality in a timely manner to reduce resource waste. The dynamic coal quality detection method and system based on image recognition mining face is an efficient and effective means to save manpower and material resources.

[0003] However, the following problems exist in coal quality testing: Whether it is on-site testing or laboratory testing, it is necessary to sample the mined coal and then process and analyze the samples. The test results are lagging, which makes it impossible to adjust the mining and treatment plans in a timely manner.

[0004] Traditional testing methods require staff to be on-site or in the laboratory to perform operations, which is cumbersome and requires education and training of personnel, which is costly.

[0005] Traditional coal quality testing loses a lot of in-situ first-hand information, such as bedding distribution information and broken coal geometry information. It is unable to characterize the uniformity, flatness, and original geometric characteristics of the coal seams in the mining area, thus wasting the opportunity to judge the coal quality in the first place.

[0006] The current sampling and testing method can lead to accidental test results, which are prone to deviations and make it difficult to determine the coal seam area that the tested samples can represent. Summary of the invention

[0007] In order to solve the deficiencies of the prior art, the present invention provides a method and system for dynamic detection of coal quality based on image recognition of mining face; the detection content includes coal seam quality and coal rock quality, and based on image intelligent recognition technology, the bedding distribution characteristics and geometric characteristics of the coal seam of the mining face are obtained while the mining operation is in progress, so that the coal seam uniformity, flatness, coal particle size and other information can be obtained on site at the first time, and the quality of the coal seam and coal rock can be analyzed. Bedding distribution characteristics include layer uniformity and layer flatness; geometric characteristics include the size of the bright spot of coal falling at the bottom.

[0008] On the one hand, a method for dynamic detection of coal quality based on image recognition of mining faces is provided, including: during the mining operation of a coal mining machine, obtaining images of the first area, the second area and the third area of ​​the mining face from top to bottom; extracting bedding features from the images of the first area and the second area to obtain a bedding line distribution map; based on the bedding line distribution map, calculating the maximum interlayer spacing and the ordinate variance; based on the maximum interlayer spacing, determining the uniformity of the coal seam; based on the ordinate variance, determining the flatness of the coal seam; based on the uniformity of the coal seam and the flatness of the coal seam, obtaining the quality of the coal seam; extracting geometric features from the image of the third area and generating a bright spot distribution map; based on the bright spot distribution map, calculating the average area of ​​the bright spots; and determining the degree of coal rock crushing according to the average area of ​​the bright spots, thereby determining the quality of the coal rock.

[0009] On the other hand, a dynamic coal quality detection system for a mining face based on image recognition is provided, comprising: an acquisition module, which is configured to: acquire images of a first area, a second area and a third area of ​​the mining face from top to bottom during mining operations by a coal mining machine; a coal seam quality determination module, which is configured to: extract bedding features from the first area image and the second area image to obtain a bedding line distribution map; calculate the maximum interlayer spacing and the ordinate variance based on the bedding line distribution map; determine the uniformity of the coal seam based on the maximum interlayer spacing; determine the flatness of the coal seam based on the ordinate variance; obtain the quality of the coal seam based on the uniformity of the coal seam and the flatness of the coal seam; a coal rock quality determination module, which is configured to: extract geometric features from the third area image and generate a bright spot distribution map; calculate the average area of ​​the bright spots based on the bright spot distribution map; determine the degree of coal rock crushing based on the average area of ​​the bright spots, and then determine the quality of the coal rock.

[0010] The above technical solution has the following advantages or beneficial effects: through real-time judgment of coal seam quality, the coal seam quality detection results can provide a reference for adjusting the mining power and forward speed of the coal mining machine in a timely manner, avoiding safety accidents caused by loosening of coal and falling blocks due to unevenness and changes in uniformity of the coal seam.

[0011] Through real-time judgment of coal and rock quality, the crushing status of coal and rock at the mining face can be obtained. Excessive crushing will increase energy consumption and accelerate the wear of the cutter head. The bright spot distribution detection results can provide a reference for the selection and replacement of cutter head models, avoiding production accidents such as coal block jams and accumulation of broken coal at the mining face.

[0012] Through this method and system, the manpower cost of on-site detection and monitoring tasks in coal mining can be reduced, the detection technology threshold can be lowered, and the personnel training and operating economic costs can be reduced.

[0013] The method and system realize real-time detection of the entire mining process, with high detection frequency and sufficient evaluation samples, which reduces the randomness of the detection results and improves the reliability of the detection results. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] 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.

[0015] Figure 1 This is a flow chart of the method of embodiment 1.

[0016] Figure 2 This is a schematic diagram of the mining process of the coal mining machine in Example 1.

[0017] Among them, 1, the first area, 2, the second area, 3, the third area. DETAILED DESCRIPTION

[0018] 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.

[0019] Embodiment 1, as Figure 1 As shown, this embodiment provides a method for dynamic detection of coal quality based on image recognition of mining faces, including: S101: during the mining operation of the coal mining machine, images of the first area 1, the second area 2 and the third area 3 from top to bottom of the mining face are obtained; S102: bedding features are extracted from the first area image and the second area image to obtain a bedding line distribution map; based on the bedding line distribution map, the maximum interlayer spacing and the ordinate variance are calculated; based on the maximum interlayer spacing, the uniformity of the coal seam is determined; based on the ordinate variance, the flatness of the coal seam is determined; based on the coal seam uniformity and the coal seam flatness, the quality of the coal seam is obtained; S103: the third area image is binarized and segmented to obtain a binary image, and the obtained binary image is used as a bright spot distribution map; based on the bright spot distribution map, the average area of ​​the bright spots is calculated; according to the average area of ​​the bright spots, the degree of coal rock fragmentation is determined, and then the quality of the coal rock is determined.

[0020] Furthermore, if Figure 2As shown, S101: during the mining operation of the coal mining machine, images of the first area, the second area and the third area of ​​the mining face from top to bottom are obtained, including: installing a camera and three strip light sources on the side of the coal mining machine; the center points of the three strip light sources are on a vertical line; the three strip light sources are, from top to bottom, the first strip light source, the second strip light source and the third strip light source; the illuminated area of ​​the mining face by the first strip light source is the first area; the image of the first area is captured by the camera to obtain an image of the first area; the illuminated area of ​​the mining face by the second strip light source is the second area; the image of the second area is captured by the camera to obtain an image of the second area; the illuminated area of ​​the mining face by the third strip light source is the third area; the image of the third area is captured by the camera to obtain an image of the third area.

[0021] The strip light sources are arranged vertically at a certain interval. The bottom area covers the coal accumulation area. The illuminated strip areas are numbered as the first area, the second area and the third area from top to bottom. When the coal mining machine moves, the camera shoots the mining face at a set frequency and updates the processed image in real time. Otherwise, it continues to display the image obtained from the most recent shooting.

[0022] like Figure 2 As shown in the figure, a camera and a vertical strip light source are installed on the side of the coal mining machine to shoot the strip coal seam area illuminated by the light source in the direction of the coal mining machine. When installing the camera and light source, they should be kept facing the same target position and can move with the coal mining machine. The camera shoots the mining face in the direction of the coal mining machine at a set frequency. To ensure the timeliness of the detection results, the shooting frequency should not be less than 2 times per second. Figure 2 As shown, the conveying direction of the conveyor belt is opposite to the forward direction of the coal mining machine.

[0023] A camera and a strip light source are installed on the side of the coal mining machine to shoot the strip coal seam area illuminated by the light source in the forward direction of the coal mining machine; in this embodiment, the coal mining machine is a fully mechanized working face coal mining machine, and the coal mining mechanical arm extends to the side and front, and the illumination light sources corresponding to the first and second areas are installed on the mechanical arm that controls the lifting and lowering of the coal mining drum, and the camera and the illumination light source corresponding to the third area are installed on the fixed surface of the rotating shaft that does not rise and fall with the coal mining drum. Thus, the first and second areas always correspond to the coal rock area directly behind the coal mining drum, and the third area is always the area covering the bottom coal. On the whole, the three areas are kept in a vertical direction, and the area photographed by the camera always covers the first, second and third areas. When the coal mining machine moves, it automatically shoots the mining face at a set frequency, and updates the processed image in real time, otherwise it continues to display the image obtained by the most recent shooting and processing.

[0024] Furthermore, after acquiring the images of the first area, the second area, and the third area of ​​the mining surface from top to bottom in S101, the method further includes: preprocessing the images of the first area and the second area; the preprocessing of the images of the first area and the second area includes: background noise reduction and contrast enhancement. The background noise reduction is performed by using a mean filter for noise reduction.

[0025] It should be understood that since the coal mining machine is accompanied by mechanical vibration, this embodiment uses a mean filter to eliminate electronic noise during image transmission, and determines the size of the filter kernel while ensuring clear image boundaries, thereby ensuring the integrity of the bedding line after processing.

[0026] Among them, contrast enhancement splits the image into three channels: R, G, and B, selects the channel with the strongest contrast, and then uses histogram equalization to enhance the image contrast for the channel with the strongest contrast.

[0027] It should be understood that in order to enhance the bedding recognition in the image, the image is split into three channels, R, G, and B, and the channel with the strongest contrast of the original image is calculated and selected, and the image under this channel is selected separately for histogram equalization, thereby further enhancing the image contrast and highlighting the bedding boundary.

[0028] Furthermore, the step S102 extracts bedding features from the first region image and the second region image to obtain a bedding line distribution map, including edge detection and curve completion.

[0029] Among them, edge detection is to use the Sobel operator to detect the edge of the image. Take the grayscale images of the first and second regions after contrast enhancement, introduce the Sobel operator to detect the edge of the image. The Sobel operator uses two The matrices are used to calculate the gradients in the horizontal and vertical directions respectively, so as to detect the edges and achieve the effect that the output image only presents the layer lines.

[0030] Among them, curve completion is that the trained bedding line completion model uses the trained bedding line completion model to complete the bedding curves of the image after edge detection to obtain a continuous bedding line distribution map, and the bedding line completion model is implemented using a U-Net network; the trained bedding line completion model is to construct a training set, which is a coal seam image of a coal mine with known bedding lines; the training set is input into the bedding line completion model to train the model, and when the total loss function value of the model no longer decreases, the training is stopped to obtain the trained bedding line completion model.

[0031] Affected by uncertain factors such as coal dust and light obstruction, the output image after edge detection is discontinuous and missing. To solve this problem, a dataset is constructed using images of shallow and surface coal mines that have been safely excavated and have clear bedding, and a bedding line completion model is trained based on the U-Net network architecture.

[0032] It should be noted that the interlayer spacing calculation at the discontinuity of the bedding line will result in extreme values, and the curvature of the bedding line at the discontinuity cannot be calculated, so it must be completed. However, since the blank distance at the discontinuity is very short relative to the entire bedding line, and the shape of a single bedding line is irrelevant to the burial depth of the coal seam, shallow and surface coal mine coal seam images are used to construct the data set. The images used to construct the data set should also undergo the above-mentioned background noise reduction, contrast enhancement, and edge detection processing.

[0033] The training set used by the trained bedding line completion model is a dataset constructed using images of shallow and surface coal mines that have been safely excavated and have complete bedding lines. When constructing the dataset, the initial image with complete bedding lines is used as the label image data. The bedding lines of the initial image are randomly removed, that is, the background pixel matrix is ​​evenly arranged in the image to randomly cover the bedding line fragments in the initial image. The image after the bedding line fragments are randomly removed is used as the sample image data. In this embodiment, the dataset is divided into a training set and a test set in a ratio of 8:2.

[0034] The trained layer line completion model is implemented based on the U-Net network architecture. The training set is input into the network in batches for forward propagation. The Dice Loss, a loss function that measures sample similarity, is used as the loss function to calculate the difference between the predicted value and the true value. Then, backpropagation is performed based on the calculated gradient to optimize the internal parameters.

[0035] When using this bedding line completion model, the image size needs to be adjusted to be consistent with the image used when training the model. Since the image input to the model is a binary image output after edge detection, no normalization is performed here. The input feature data is the coal seam bedding line image containing discontinuities, and the output feature data is the complete coal seam bedding line image after completion, which is the bedding line distribution map.

[0036] Further, the step S102: calculating the maximum interlayer distance and the ordinate variance based on the bedding line distribution diagram includes: assuming that two adjacent bedding lines are Layer by layer cable management and Layer by layer cable management , in Layer by layer cable management Randomly select points, along Draw a point A vertical line, The vertical line and Layer by layer cable management Intersect intersection points. Layer by layer cable management Previous Dot pixels The coordinates are , No. Layer by layer cable management Upper and lower Dot pixels Pixels on the same vertical line The coordinates are , No. Layer by layer cable management With Layer by layer cable management The interlayer spacing The formula is: ; The maximum inter-layer spacing refers to the maximum value of all inter-layer spacings in the same area image.

[0037] Furthermore, the S102: determining the uniformity of the coal seam based on the maximum value of the interlayer spacing, includes: when the maximum interlayer spacing is greater than two-thirds of the excavation surface height, it indicates that the coal seam has little stratification within the unit thickness and uniform properties; when the maximum interlayer spacing is greater than one-third of the excavation surface height and is less than or equal to two-thirds of the excavation surface height, it indicates that the coal seam has obvious stratification within the unit thickness and uneven properties; when the maximum interlayer spacing is less than one-third of the excavation surface height, it indicates that the coal seam has serious stratification within the unit thickness and extremely uneven properties.

[0038] It should be noted that the judgment of coal seam uniformity here refers to the change in properties between different coal deposits in a certain area on a macro scale. And the evaluation of coal seam uniformity cannot rely on the number of bedding lines in a certain area. For example, when the thickness of some layers in a certain area is large while the thickness of other layers is extremely small, the evaluation based on the number of bedding lines will result in a significantly lower uniformity.

[0039] Furthermore, the formula of S102: vertical coordinate variance is: ;in, is the ordinate variance, is the average value of the ordinates of all points on the bedding line, that is, the ordinate of the point on the central axis, is the ordinate of any point on the bedding line, is the number of points on the bedding line.

[0040] The ordinate variance refers to the variance of the ordinate of a point on the bedding line relative to the ordinate of a point on the centerline of the bedding line.

[0041] Furthermore, the S102: determining the flatness of the coal seam based on the ordinate variance includes: for a bedding line, the larger the ordinate variance is, the greater the degree of vertical dispersion of the points on it is, and the more uneven it is.

[0042] Furthermore, after S102 and before S103, the method further includes: preprocessing the third region image; the preprocessing the third region image includes: background noise reduction and contrast enhancement.

[0043] Wherein, the background noise reduction is performed by using a mean filter.

[0044] Among them, contrast enhancement splits the image into three channels: R, G, and B, selects the channel with the strongest contrast, and then uses histogram equalization to enhance the image contrast for the channel with the strongest contrast.

[0045] Further, the S103: performing binary segmentation on the third region image to obtain a binary image, and using the obtained binary image as a bright spot distribution map, includes: wherein, in the binary segmentation, the Otsu algorithm is used when selecting the segmentation threshold, so that the bright spot is clearest when the threshold is used for segmentation. Since the bright spot is the specular reflection of the coal particles to the light, it has a special metallic luster, so a smaller gray value is selected as the segmentation threshold during the binary segmentation. When the bright spot is not clear, consider lowering the segmentation threshold, so as to further accurately distinguish the bright spot.

[0046] After the binary segmentation, the bright spot part appears white highlight, and the rest appears black. The bright spots are distributed in the black background as white blocks, so the binary image is used as the bright spot distribution map.

[0047] Furthermore, the step S103: calculating the average area of ​​the bright spots based on the bright spot distribution map includes: first importing the segmented image into the matlab software, calling the bwlabel() function in the matlab software to mark each connected area in the image, then using the unique() function to find the label of the marked area, and using the numel() function to count the number of labels, thereby obtaining the number of connected areas, which is also the number of bright spots. ; Secondly, perform grayscale analysis on the bright spot distribution map, call the imhist() function to count the number of grayscale pixels in the image, and obtain the total pixel area of ​​the bright spot .

[0048] Then calculate the average area of ​​all bright spots , the formula is: .

[0049] Furthermore, the S103: determining the degree of coal rock crushing according to the average area of ​​the bright spots includes: calculating the diameter of the bright spots , with the diameter of the bright spot To indicate the particle size of coal particles , consider the bright spot as a perfect circle, the granularity The calculation formula is: ;when When it is considered to be powdered coal, it means that the coal rock is very fragile; when When it is considered to be slag-like coal, it indicates that the coal rock is fragile; when At that time, it was believed to be lumpy coal, indicating that the coal rock was not easy to break.

[0050] Logically speaking, the particle size obtained from the average area of ​​the bright spots represents the degree of coal rock fragmentation under the current mining operation. The smaller the particle size, the smaller the coal rock is after mechanical disturbance and the smaller the particle size formed. Under certain operating conditions, the particle size of the falling coal can reflect the fragility of the coal rock. The smaller the particle size, the more fragile the coal rock.

[0051] For fragile coal rock, its hardness is low, its toughness is poor, and its strength is low. These characteristics can reflect some characteristics of coal rock: from a structural point of view, the fragility of coal rock indicates that its internal structure is loose, porous and fissured; from a compositional point of view, the fragility of coal rock indicates that it has a high water content and a low organic matter content, because there are stable CH and CC molecular bonds in the dense organic matter, and the structure is stable, and the presence of water will weaken this stable structure. From a functional point of view, the fragile mechanical properties represent the difficulty of mining operations, high safety risks, and high support costs. In addition, the higher water content and lower organic matter content will significantly reduce the calorific value of coal and affect its combustion performance.

[0052] Embodiment 2: This embodiment provides a dynamic coal quality detection system for mining faces based on image recognition, including: an acquisition module, which is configured to: acquire images of the first area, the second area and the third area from top to bottom of the mining face during the mining operation of the coal mining machine; a coal seam quality determination module, which is configured to: extract bedding features from the first area image and the second area image to obtain a bedding line distribution map; based on the bedding line distribution map, calculate the maximum interlayer spacing and the ordinate variance; based on the maximum interlayer spacing, determine the uniformity of the coal seam; based on the ordinate variance, determine the flatness of the coal seam; based on the coal seam uniformity and the coal seam flatness, obtain the coal seam quality; a coal rock quality determination module, which is configured to: extract geometric features from the third area image and generate a bright spot distribution map; based on the bright spot distribution map, calculate the average area of ​​the bright spots; based on the average area of ​​the bright spots, determine the fragility of the coal rock, and then determine the quality of the coal rock.

[0053] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for dynamic detection of coal quality based on image recognition of mining face, characterized in that: include: During the mining operation of the coal mining machine, images of a first area, a second area and a third area from top to bottom of the mining face are obtained; Extracting bedding features from the first region image and the second region image to obtain a bedding line distribution map; Based on the bedding line distribution map, the maximum interlayer distance and the vertical coordinate variance are calculated; Based on the maximum inter-layer distance, the uniformity of the coal seam is determined; Based on the vertical coordinate variance, the flatness of the coal seam is determined; Based on the coal seam uniformity and the coal seam flatness, the coal seam quality is obtained; Extracting geometric features from the third region image and generating a bright spot distribution map; Based on the bright spot distribution map, the average area of ​​the bright spots is calculated; based on the average area of ​​the bright spots, the degree of coal rock crushing is determined, and then the quality of the coal rock is determined.

2. The method for dynamic detection of coal quality based on image recognition of mining face according to claim 1, characterized in that: During the mining operation of the coal mining machine, images of the first area, the second area and the third area of ​​the mining face from top to bottom are obtained, including: A camera and three strip light sources are installed on the side of the coal mining machine; the center points of the three strip light sources are on a vertical line; the three strip light sources are respectively a first strip light source, a second strip light source and a third strip light source from top to bottom; The first strip-shaped light source illuminates the first area of ​​the mining surface; the camera is used to capture the image of the first area to obtain the image of the first area; the second strip-shaped light source illuminates the second area of ​​the mining surface; the camera is used to capture the image of the second area to obtain the image of the second area; the third strip-shaped light source illuminates the third area of ​​the mining surface; the camera is used to capture the image of the third area to obtain the image of the third area.

3. The method for dynamic detection of coal quality based on image recognition of mining face according to claim 1, characterized in that: After acquiring images of the first area, the second area and the third area from top to bottom of the mining surface, the method further includes: preprocessing the first area image and the second area image; The preprocessing of the first area image and the second area image includes: background noise reduction and contrast enhancement; wherein, the background noise reduction adopts a mean filter to perform noise reduction processing; wherein, the contrast enhancement splits the image into three channels of R, G, and B, selects the channel with the strongest contrast, and then enhances the image contrast by using histogram equalization for the channel with the strongest contrast.

4. The method for dynamic detection of coal quality based on image recognition of mining face according to claim 1, characterized in that: Extracting bedding features from the first region image and the second region image to obtain a bedding line distribution map, including: edge detection and curve completion; Among them, edge detection is to use the Sobel operator to detect the edge of the image; take the grayscale images of the first and second regions after contrast enhancement, introduce the Sobel operator to detect the edge of the image, and the Sobel operator uses two The matrices are used to calculate the gradients in the horizontal and vertical directions respectively, so as to detect the edges and achieve the effect that the output image only presents the layer lines; Among them, curve completion is that the trained bedding line completion model uses the trained bedding line completion model to complete the bedding curves of the image after edge detection to obtain a continuous bedding line distribution map, and the bedding line completion model is implemented using a U-Net network; the trained bedding line completion model is to construct a training set, which is a coal seam image of a coal mine with known bedding lines; the training set is input into the bedding line completion model to train the model, and when the total loss function value of the model no longer decreases, the training is stopped to obtain the trained bedding line completion model.

5. The method for dynamic detection of coal quality based on image recognition of mining face according to claim 1, characterized in that: Based on the bedding line distribution map, the maximum interlayer spacing and ordinate variance are calculated, including: Assume that the two adjacent bedding lines are Layer by layer cable management and Layer by layer cable management , in Layer by layer cable management Randomly select points, along Draw a point A vertical line, The vertical line and Layer by layer cable management Intersect intersection points; Suppose the first Layer by layer cable management Previous Dot pixels The coordinates are , No. Layer by layer cable management Upper and lower Dot pixels Pixels on the same vertical line The coordinates are , No. Layer by layer cable management With Layer by layer cable management The interlayer spacing The formula is: ; The maximum inter-layer spacing refers to the maximum value of all inter-layer spacings in the same area image.

6. The method for dynamic detection of coal quality based on image recognition of mining face according to claim 1, characterized in that: Based on the maximum interlayer distance, the uniformity of the coal seam is determined, including: When the maximum interlayer distance is greater than two-thirds of the excavation face height, it indicates that the coal seam has few layers per unit thickness and is uniform in nature. When the maximum interlayer distance is greater than one-third of the excavation face height and less than or equal to two-thirds of the excavation face height, it indicates that the coal seam has obvious stratification within the unit thickness and its properties are uneven; When the maximum interlayer distance is less than one-third of the excavation face height, it indicates that the coal seam is severely stratified within unit thickness and its properties are extremely uneven.

7. The method for dynamic detection of coal quality based on image recognition of mining face according to claim 1, characterized in that: The vertical axis variance is: ;in, is the ordinate variance, is the average value of the ordinates of all points on the bedding line, that is, the ordinate of the point on the central axis, is the ordinate of any point on the bedding line, It is the number of points on the bedding line; based on the ordinate variance, the flatness of the coal seam is determined, including: for a bedding line, the larger the ordinate variance, the greater the degree of vertical dispersion of the points on it, and the more uneven it is.

8. The method for dynamic detection of coal quality based on image recognition of mining face according to claim 1, characterized in that: Based on the bright spot distribution map, the average area of ​​the bright spots is calculated, including: Mark each connected area in the image, find the label of the marked area, count the number of labels, and thus get the number of connected areas, which is also the number of bright spots. ; Secondly, the grayscale analysis of the bright spot distribution map is performed, and the number of grayscale pixels in the image is counted to obtain the total pixel area of ​​the bright spot ; Then calculate the average area of ​​all bright spots , the formula is: .

9. The method for dynamic detection of coal quality based on image recognition of mining face according to claim 1, characterized in that: According to the average area of ​​the bright spot, determine the degree of coal rock fragmentation, including: calculate the diameter of the bright spot , with the diameter of the bright spot To indicate the particle size of coal particles , consider the bright spot as a perfect circle, the granularity The calculation formula is: ;when When it is considered to be powdered coal, it means that the coal rock is very fragile; when When it is considered to be slag-like coal, it indicates that the coal rock is fragile; when At that time, it was believed to be lumpy coal, indicating that the coal rock was not easy to break.

10. A coal quality dynamic detection system based on image recognition mining face, characterized in that: include: An acquisition module is configured to: acquire images of a first area, a second area, and a third area from top to bottom of a mining face during a mining operation of a coal mining machine; A coal seam quality determination module is configured to: extract bedding features from the first region image and the second region image to obtain a bedding line distribution map; Based on the bedding line distribution map, the maximum interlayer distance and the vertical coordinate variance are calculated; Based on the maximum inter-layer distance, the uniformity of the coal seam is determined; Based on the ordinate variance, the coal seam flatness is determined; based on the coal seam uniformity and the coal seam flatness, the coal seam quality is obtained; A coal rock quality determination module is configured to: extract geometric features from the third region image and generate a bright spot distribution map; Based on the bright spot distribution map, the average area of ​​the bright spots is calculated; based on the average area of ​​the bright spots, the degree of coal rock crushing is determined, and then the quality of the coal rock is determined.

Images