Constructive coal morphological parameter inversion method and system based on high-definition video

Abstract

The invention provides a tectonic coal morphological parameter inversion method and system based on a high-definition video, and relates to the technical field of coal mine safety, and the method comprises the steps: collecting a video image of underground tectonic coal; the collected video images are transmitted in a wireless or wired mode; performing denoising, enhancement and segmentation preprocessing on the transmitted video image, and extracting morphological feature parameters of the tectonic coal; establishing a prominent mechanism inversion model based on the morphological characteristic parameters; and displaying a highlighting mechanism inversion result according to the highlighting mechanism inversion model. According to the invention, non-contact and high-precision dynamic observation of the form of the underground tectonic coal is realized through high-definition video image real-time acquisition, intelligent preprocessing and feature extraction technologies, and the defects of low efficiency and high risk of traditional manual sampling are overcome.

Description

Technical Field

[0001] The present invention relates to the technical field of coal mine safety, and particularly to a method and system for inverting the morphological parameters of tectonic coal based on high-definition video. Background Art

[0002] The existence of tectonic coal is one of the important factors causing coal and gas outbursts. Production practice shows that the vast majority of coal and gas outbursts in the world occur in geological structure damage zones. When geological anomalies affecting production are formed or exist in coal seams (such as faults, karst collapse columns, coal seam scouring zones, etc.), the change of local stress field caused by the change of boundary conditions will cause changes in the development scale, mechanical properties, spatial occurrence, combination form, etc. of coal seam joints. Analyzing and studying the development characteristics and change laws of coal seam joints can help predict the hidden geological structures in front of the driving roadway or in the working face, so as to take measures in advance to ensure the normal progress of roadway driving engineering and coal mining operations.

[0003] Traditional methods for observing the morphology of tectonic coal mainly rely on manual underground sampling and laboratory analysis, and have the following defects:

[0004] (1) Low efficiency: The manual sampling period is long and cannot meet the requirements of real-time monitoring.

[0005] (2) High risk: The underground operation environment is complex, and there are potential safety hazards such as gas explosion and roof collapse.

[0006] (3) Data limitation: Laboratory analysis can only obtain the static parameters of local samples and is difficult to comprehensively reflect the dynamic changes of coal seams.

[0007] (4) Non-real-time: It is impossible to give real-time warnings about the hidden geological structures in front of the driving roadway or in the working face.

[0008] In response to the above problems, remote sensing or ordinary camera equipment has been tried in the prior art for monitoring. However, due to insufficient image resolution, poor environmental adaptability and imperfect data processing algorithms, there are still problems such as incomplete parameter extraction and low accuracy of inversion models. Summary of the Invention

[0009] In order to overcome the deficiencies of the prior art, the purpose of the present invention is to provide a method and system for inverting the morphological parameters of tectonic coal based on high-definition video, which realizes non-contact and high-precision dynamic observation of the morphology of underground tectonic coal and overcomes the defects of low efficiency and high risk of traditional manual sampling.

[0010] To achieve the above object, the present invention provides the following solutions:

[0011] A method for inverting the morphological parameters of tectonic coal based on high-definition video, comprising:

[0012] Collect video images of in - situ tectonic coal;

[0013] Wirelessly or wiredly transmit the collected video images;

[0014] Perform denoising, enhancement and segmentation pre - processing on the transmitted video images, and extract the morphological feature parameters of tectonic coal;

[0015] Establish an inversion model of outburst mechanism based on the morphological feature parameters;

[0016] Display the inversion result of outburst mechanism according to the outburst mechanism inversion model.

[0017] Preferably, the morphological feature parameters of tectonic coal include: joint parameters and coal - rock layer parameters; the joint parameters include: joint width, length, direction angle, density and spatial distribution law; the coal - rock layer parameters include: dip angle of coal - rock layer and coal seam thickness.

[0018] Preferably, performing denoising, enhancement and segmentation pre - processing on the transmitted video images, and extracting the morphological feature parameters of tectonic coal, includes:

[0019] Use the adaptive Gaussian filtering algorithm to suppress the noise of the video images to eliminate the uneven illumination and dust interference in the underground environment, and obtain the filtered image;

[0020] Enhance the details of the filtered image through histogram equalization and local contrast stretching techniques to highlight the joint and coal - rock layer boundary features of tectonic coal, and obtain the enhanced image;

[0021] Based on the improved Otsu multi - threshold segmentation algorithm, divide the enhanced image into coal body area, joint area and background area;

[0022] Use the Canny edge detection algorithm to extract the joint contours of the joint area, and combine morphological closing operation to eliminate small fractures to calculate the joint width, length and direction angle;

[0023] Detect the joint linear features of the joint contours through Hough transform, and combine the density clustering algorithm to statistically analyze the joint density and spatial distribution law;

[0024] Based on the coal body area, use the least - squares fitting method to calculate the dip angle of the coal - rock layer, and invert the coal seam thickness through pixel calibration technology.

[0025] Preferably, the calculation formula of the adaptive standard deviation in the adaptive Gaussian filtering algorithm is as follows:

[0026]

[0027] Among them, σ(x,y) is the adaptive standard deviation, where x and y represent the coordinate values of the x-axis and y-axis of the current pixel in the video image; σ base is the preset basic standard deviation, I(x,y) is the gray value of the pixel point (x,y), I max is the maximum gray value within the local window, γ is the illumination compensation coefficient to control the adjustment range of uneven illumination, and the value range of γ is 0.5 to 1.5; is the local gradient amplitude to reflect the dust interference intensity, where G x and G y are the Sobel convolution results in the horizontal and vertical directions respectively; is the maximum gradient amplitude within the local window, η is the dust interference suppression coefficient, and the value range of the dust interference suppression coefficient is 0.3 to 0.8.

[0028] Preferably, the calculation formula of the improved Otsu multi-threshold segmentation algorithm is:

[0029]

[0030] Among them, t1 and t2 are both segmentation thresholds to divide the enhanced image into three categories: coal body area, joint area, and background area; is the optimization objective function, and maximizing the value of the optimization objective function is used to determine the optimal threshold; ω i is the probability of the i-th type of pixel, N i is the number of pixels of the i-th type, N is the total number of pixels; μ i is the average gray value of the i-th type, h(k) is the gray histogram; μ T is the total average gray value of the image, L is the number of gray levels; λ is the preset continuity constraint weight coefficient, C i is the regional continuity constraint term, the stronger; β is the gray distribution weight coefficient, W i is the gray distribution weight factor,

[0031] Preferably, an outburst mechanism inversion model is established based on the morphological characteristic parameters, including:

[0032] Performing multi-source data fusion on the morphological characteristic parameters of the tectonic coal, the preset geomechanical parameters, and the gas parameters, and eliminating the dimension difference through normalization processing to generate a standardized input data set;

[0033] Based on the random forest feature importance evaluation algorithm, screen the key parameters in the standardized input dataset that have a significant impact on outburst danger; the key parameters include joint density, coal seam and rock stratum dip angle, and gas pressure;

[0034] Use the analytic hierarchy process combined with expert experience to assign dynamic weights to the key parameters; the weight values of the dynamic weights are adjusted according to the real-time underground working conditions;

[0035] Establish a hybrid prediction model based on support vector machine and long short-term memory network;

[0036] Use the historical outburst accident data to train the hybrid prediction model, and evaluate the model accuracy through five-fold cross-validation, optimize the hyperparameters, and obtain the trained outburst mechanism inversion model;

[0037] Input the key parameters and the dynamic weights into the outburst mechanism inversion model to output the outburst mechanism inversion result.

[0038] Preferably, the outburst mechanism inversion result includes: outburst danger level and probability prediction result.

[0039] A high-definition video-based coal structure morphology parameter inversion system, comprising:

[0040] A video acquisition module for acquiring video images of underground coal structures;

[0041] A data transmission module for wirelessly or wiredly transmitting the acquired video images;

[0042] A data processing module for performing denoising, enhancement, and segmentation preprocessing on the transmitted video images, and extracting coal structure morphology characteristic parameters;

[0043] An inversion model module for establishing an outburst mechanism inversion model based on the morphology characteristic parameters;

[0044] A result display module for displaying the outburst mechanism inversion result according to the mechanism inversion model.

[0045] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0046] The present invention provides a method and system for inverting the morphological parameters of tectonic coal based on high-definition video. The method includes: collecting video images of underground tectonic coal; wirelessly or wiredly transmitting the collected video images; performing preprocessing such as denoising, enhancement, and segmentation on the transmitted video images, and extracting the morphological characteristic parameters of tectonic coal; establishing an outburst mechanism inversion model based on the morphological characteristic parameters; and displaying the outburst mechanism inversion result according to the outburst mechanism inversion model. Through real-time collection, intelligent preprocessing, and feature extraction technologies of high-definition video images, the present invention realizes non-contact and high-precision dynamic observation of the morphology of underground tectonic coal, overcomes the defects of low efficiency and high danger of traditional manual sampling; the outburst mechanism inversion model constructed based on multi-source data fusion and hybrid modeling methods can comprehensively quantify joint characteristics, coal and rock parameters, and gas dynamic data, significantly improving the accuracy and real-time performance of outburst danger assessment, providing a scientific basis for coal mine safety early warning, and effectively reducing the risk of coal and gas outburst accidents. Its technical solution combines engineering practicability and innovation, filling the technical gap in real-time intelligent monitoring and mechanism inversion of tectonic coal. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0048] Figure 1 It is a flowchart of the method provided by an embodiment of the present invention;

[0049] Figure 2 It is a flowchart of image processing and parameter extraction provided by an embodiment of the present invention;

[0050] Figure 3 It is a flowchart of establishing an outburst mechanism inversion model provided by an embodiment of the present invention;

[0051] Figure 4 It is a schematic structural diagram of the system provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0052] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0053] The object of the present invention is to provide a method and system for inverting the morphological parameters of tectonic coal based on high-definition video, which realizes non-contact and high-precision dynamic observation of the morphology of underground tectonic coal, and overcomes the defects of low efficiency and high danger of traditional manual sampling.

[0054] To make the above objects, features and advantages of the present invention more obvious and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0055] Figure 1 It is a flowchart of the method provided by the embodiment of the present invention. As Figure 1 shown, the present invention provides a method for inverting the morphological parameters of tectonic coal based on high-definition video, including:

[0056] Step 100: Collect video images of underground tectonic coal;

[0057] Step 200: Transmit the collected video images wirelessly or wiredly;

[0058] Step 300: Perform preprocessing of denoising, enhancement and segmentation on the transmitted video images, and extract the morphological characteristic parameters of tectonic coal;

[0059] Step 400: Establish an inversion model of outburst mechanism based on the morphological characteristic parameters;

[0060] Step 500: Display the inversion result of the outburst mechanism according to the inversion model of the outburst mechanism.

[0061] Specifically, step 100 realizes the collection of video images of tectonic coal through an explosion-proof high-definition camera system deployed in underground roadways or working faces. In this embodiment, a mine intrinsically safe explosion-proof camera (resolution ≥ 1920×1080@30fps) is selected, equipped with a high-brightness LED fill light (illuminance ≥ 500lux) and a programmable pan-tilt (supporting ±90° pitch angle and 360° horizontal rotation) to ensure stable imaging in low-light and high-dust environments; in this embodiment, camera nodes are arranged at intervals of 50-100m along the coal seam strike, and the camera is controlled by the pan-tilt to scan the exposed surface of tectonic coal from multiple angles (including the roof, floor and side walls), and the GPS / Beidou positioning coordinates and acquisition timestamps are recorded synchronously; and the wide dynamic range (WDR) imaging technology is used to suppress local overexposure / underexposure, and the real-time dust concentration sensor data is embedded to dynamically adjust the fill light intensity and focal length; before transmission, frame-level verification is performed on the video stream, and blurred frames caused by vibration or occlusion are discarded (judged based on the Laplacian gradient value threshold).

[0062] Exemplarily, in this embodiment, 4K-level high-definition videos covering the entire area of tectonic coal can be obtained, and the image texture clarity meets the recognition requirements of joint width ≥ 1mm, providing a high-quality data source for subsequent analysis.

[0063] Specifically, step 200 of this embodiment realizes the highly reliable transmission of video images by constructing a multi-mode redundant transmission network. Specifically, a mine explosion-proof and intrinsically safe gigabit fiber switch and a Mesh self-organizing wireless base station are used for hybrid networking. Among them, the fiber-optic transmission backbone network is deployed in the main roadway (transmission rate ≥ 1 Gbps, latency < 10 ms), and the wireless Mesh nodes cover mobile areas such as the mining face (supporting the IEEE 802.11ac protocol, single-hop transmission distance ≥ 200 m). During the transmission process, an adaptive bitrate adjustment technology is implemented (based on the H.265 coding standard, dynamically adjusting the bitrate from 2 Mbps to 20 Mbps according to the network bandwidth), and a forward error correction (FEC) algorithm and a retransmission mechanism are embedded (the packet loss rate compensation threshold is set to 5%). At the same time, the AES-256 encryption protocol is used to ensure data security. All transmission devices have passed the MA / KA mine product safety certification to ensure the real-time, complete, and safe transmission of video data in the complex underground electromagnetic environment and gas hazardous areas, meeting the requirements of the subsequent processing module for image continuity and timeliness (the end-to-end transmission latency is controlled within 500 ms).

[0064] Preferably, the morphological characteristic parameters of the tectonic coal include: joint parameters and coal seam and rock layer parameters; the joint parameters include: joint width, length, direction angle, density, and spatial distribution law; the coal seam and rock layer parameters include: coal seam and rock layer dip angle and coal seam thickness.

[0065] As Figure 3 shown, this embodiment performs denoising, enhancement, and segmentation preprocessing on the transmitted video images, and extracts the morphological characteristic parameters of the tectonic coal, including:

[0066] An adaptive Gaussian filtering algorithm is used to suppress the noise of the video image to eliminate uneven illumination and dust interference in the underground environment, and a filtered image is obtained;

[0067] The details of the filtered image are enhanced through histogram equalization and local contrast stretching techniques to highlight the joint and coal seam and rock layer boundary features of the tectonic coal, and an enhanced image is obtained;

[0068] Based on an improved Otsu multi-threshold segmentation algorithm, the enhanced image is divided into a coal body region, a joint region, and a background region;

[0069] The Canny edge detection algorithm is used to extract the joint contours of the joint region, and morphological closing operations are combined to eliminate small fractures to calculate the joint width, length, and direction angle;

[0070] The joint linear features of the joint contours are detected through the Hough transform, and the density clustering algorithm is combined to statistically analyze the joint density and spatial distribution law;

[0071] Based on the coal body area, the least squares fitting method is used to calculate the dip angle of the coal and rock strata, and the pixel calibration technology is used to invert the coal seam thickness.

[0072] Optionally, in this embodiment, the adaptive Gaussian filtering algorithm is first used to suppress the noise of the downhole video image. Specifically, the filtering intensity is dynamically adjusted according to the gray value and gradient information of the local area of the image: the smoothing is enhanced in areas with insufficient light or dust coverage to eliminate noise, while the filtering intensity is reduced in the joint edge areas with clear texture to retain details, ensuring that the filtered image after preprocessing not only suppresses interference but also does not lose key features. Subsequently, histogram equalization and local contrast stretching techniques are used to enhance the filtered image. The overall contrast is improved through global gray value redistribution, and at the same time, the local area is dynamically adjusted in brightness by combining a sliding window, with emphasis on strengthening the gray difference between the coal and rock strata interfaces and joint fractures, making the coal seam structure boundary in the enhanced image clearly distinguishable.

[0073] Furthermore, based on the improved Otsu multi-threshold segmentation algorithm, the enhanced image is divided into three types of regions: coal body, joints, and background. The improved algorithm introduces regional connectivity constraints and within-class gray variance weight factors, giving priority to ensuring the integrity and continuity of the coal body area and suppressing mis-segmentation caused by dust noise. The segmentation result is optimized for edge smoothness through morphological erosion and dilation. For the segmented joint area, the Canny edge detection algorithm is used to extract the joint contour, combined with morphological closing operations to fill small fractures and connect discontinuous edges, and the average width, extension length, and main direction angle of the joints are calculated based on the pixel coordinate system; at the same time, the Hough transform is used to identify the linear trend of the joints, and the DBSCAN density clustering algorithm is combined to analyze the spatial distribution density and aggregation law of the joint groups.

[0074] Even further, in this embodiment, finally, based on the binary mask of the coal body area, the least squares method is used to fit the straight line equation of the coal and rock strata interface, and the spatial occurrence of the coal and rock strata is output through the dip angle calculation model; the coal seam thickness is inverted through the pixel calibration technology. Calibration objects with known sizes are pre-laid downhole, the mapping relationship between image pixels and actual sizes is established, and the true thickness value is calculated in combination with the vertical projection length of the coal body area, realizing the accurate conversion from image features to engineering parameters.

[0075] Preferably, the calculation formula for the adaptive standard deviation in the adaptive Gaussian filtering algorithm is as follows:

[0076]

[0077] where σ(x,y) is the adaptive standard deviation, and x and y respectively represent the x-axis and y-axis coordinate values of the current pixel in the video image; σ base is the preset basic standard deviation, I(x,y) is the gray value of the pixel point (x,y), I maxis the maximum gray value within the local window, γ is the illumination compensation coefficient to control the adjustment range of uneven illumination, and the value range of γ is 0.5 to 1.5; is the local gradient amplitude to reflect the intensity of dust interference, wherein, G x and G y are the Sobel convolution results in the horizontal and vertical directions respectively; is the maximum gradient amplitude within the local window, η is the dust interference suppression coefficient, and the value range of the dust interference suppression coefficient is 0.3 to 0.8.

[0078] Exemplarily, the typical value of the basic standard deviation is 1.0 to 2.0; it determines the overall smoothing intensity of the filtering; the basic standard deviation is pre-calibrated according to the typical dust particle size and noise level underground. This embodiment can solve the problems of uneven illumination and dust interference underground at the same time, which is superior to the traditional fixed σ Gaussian filtering and single illumination compensation method; the physical meaning of the parameters is clear, supports on-site rapid calibration, and meets the requirements of engineering practicability.

[0079] Preferably, the calculation formula of the improved Otsu multi-threshold segmentation algorithm is:

[0080]

[0081] wherein, both t1 and t2 are segmentation thresholds to divide the enhanced image into three categories: coal body area, joint area and background area; is the optimization objective function, and maximizing the value of the optimization objective function is used to determine the optimal threshold; ω i is the probability of the i-th type of pixel, N i is the number of pixels of the i-th type, N is the total number of pixels; μ i is the average gray value of the i-th type, h(k) is the gray histogram; μ T is the total average gray value of the image, L is the number of gray levels; λ is the preset continuity constraint weight coefficient, C i is the regional continuity constraint term, the stronger; β is the gray distribution weight coefficient, W i is the gray distribution weight factor,

[0082] Exemplarily, the number of gray levels in this embodiment can take the value of 256. The typical value of the continuity constraint weight coefficient in the underground scene can be 0.2 to 0.5. The gray distribution weight coefficient is used to suppress background interference and is dynamically adjusted according to the dust concentration (typical value: 0.1 to 0.3).

[0083] Such as Figure 4As shown, an inversion model of outburst mechanism is established based on the morphological characteristic parameters, including:

[0084] Fuse the morphological characteristic parameters of tectonic coal, the preset geomechanical parameters and gas parameters, and eliminate the dimensional difference through normalization to generate a standardized input data set;

[0085] Based on the random forest feature importance evaluation algorithm, screen the key parameters in the standardized input data set that have a significant impact on outburst danger; the key parameters include joint density, coal seam and rock layer dip angle, and gas pressure;

[0086] Use the analytic hierarchy process combined with expert experience to assign dynamic weights to the key parameters; the weight values of the dynamic weights are adjusted according to the real-time underground working conditions;

[0087] Establish a hybrid prediction model based on support vector machine and long short-term memory network;

[0088] Use historical outburst accident data to train the hybrid prediction model, evaluate the model accuracy through five-fold cross-validation, optimize the hyperparameters, and obtain the trained outburst mechanism inversion model;

[0089] Input the key parameters and the dynamic weights into the outburst mechanism inversion model to output the outburst mechanism inversion results. The outburst mechanism inversion results include: outburst danger level and probability prediction results.

[0090] Specifically, in this embodiment, the morphological characteristic parameters of tectonic coal are first integrated with the preset geomechanical parameters and gas parameters through multi-source data fusion technology. The principal component analysis method is used to reduce the dimension and extract features of heterogeneous data. At the same time, the maximum-minimum normalization method is used to eliminate the dimensional differences between different parameters, generating a standardized input data set with a unified dimension to ensure that the contribution degrees of various parameters in the model are comparable. Based on the random forest algorithm, the feature importance of the standardized data set is evaluated. By calculating the decrease in the Gini index when each parameter is used for node division in the decision tree, the influence degree of the parameter on the outburst risk is quantified, and three key parameters, namely joint density, coal seam inclination, and gas pressure, are selected as the core input variables of the model. These parameters can most significantly characterize the stability of the coal body structure and the gas occurrence state. The analytic hierarchy process is used to construct a dynamic weight allocation mechanism for key parameters. The initial weights of each parameter are determined by constructing a judgment matrix and calculating the eigenvector. At the same time, a real-time working condition feedback adjustment factor is introduced. When an abnormal increase in gas concentration or a sudden change in coal and rock stress is detected, the weight ratio of the corresponding parameter is automatically adjusted, enabling the model to dynamically respond to changes in the underground environment. A hybrid prediction model architecture of support vector machine and long short-term memory network is constructed. The support vector machine is responsible for processing the classification of static structural feature parameters and uses a radial basis kernel function to achieve nonlinear mapping. The long short-term memory network is used to analyze the temporal variation laws of parameters such as gas pressure and in-situ stress, and captures the dynamic evolution characteristics of outburst risk through memory units. The outputs of the two models are integrated and predicted through a weighted fusion layer. The historical outburst accident case data is used to train the hybrid model. The five-fold cross-validation method is used to evaluate the model performance. The penalty factor and kernel function parameters of the support vector machine and the number of layers and neurons of the long short-term memory network are optimized through a grid search algorithm to ensure that the prediction accuracy of the model on the test set reaches more than 85%, and the F1 score is not less than 0.8. The key parameters and their dynamic weights obtained in real time are input into the trained outburst mechanism inversion model. The model outputs include the outburst risk level (divided into levels 1-5) and the occurrence probability (0%-100%). At the same time, a risk heat map is generated to intuitively display the distribution of dangerous areas. When the predicted level reaches level 3 or above, an audible and visual alarm is automatically triggered and prevention and control measure suggestions are pushed to the underground intelligent terminal, realizing a closed-loop management from data collection to risk warning.

[0091] Corresponding to the above method, as Figure 4 shown, this embodiment also provides a system for inverting the morphological parameters of tectonic coal based on high-definition video, including:

[0092] A video acquisition module for acquiring video images of underground tectonic coal;

[0093] A data transmission module for wirelessly or wiredly transmitting the acquired video images;

[0094] A data processing module for preprocessing the transmitted video images by denoising, enhancing, and segmenting, and extracting the morphological feature parameters of tectonic coal;

[0095] An inversion model module for establishing an outburst mechanism inversion model based on the morphological feature parameters;

[0096] A result display module for displaying the outburst mechanism inversion result according to the mechanism inversion model.

[0097] The beneficial effects of the present invention are as follows:

[0098] (1) Through the real-time acquisition and intelligent analysis technology of high-definition video images, the present invention realizes non-contact and automated observation of the morphology of tectonic coal, overcomes the defects of low efficiency and large errors in traditional manual sampling, and significantly improves the monitoring accuracy and timeliness of the danger of coal and gas outbursts.

[0099] (2) The present invention adopts adaptive filtering, improved segmentation algorithms, and multi-source data fusion technology to effectively suppress environmental interferences such as underground dust and uneven illumination, ensure the accuracy of the extraction of morphological feature parameters, and provide a reliable data basis for outburst mechanism analysis.

[0100] (3) Based on the hybrid machine learning model (SVM + LSTM) and the dynamic weight allocation mechanism, the model can respond to the changes in underground working conditions in real time, output quantitative danger levels and probability predictions, and provide a scientific decision-making basis for coal mine safety production.

[0101] (4) The present invention is designed to be compatible with existing coal mine monitoring equipment, supports embedded deployment and real-time feedback, can be directly applied to complex underground environments, significantly reduces the risk of coal and gas outburst accidents, and has important engineering promotion value.

[0102] In this specification, each embodiment is described in a progressive manner. The key point of each embodiment is to illustrate the differences from other embodiments. The same or similar parts among the embodiments can be referred to each other. For the system disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple, and the relevant parts can be referred to the description of the method part.

[0103] Specific examples are used in this article to elaborate on the principles and implementation methods of the present invention. The descriptions of the above embodiments are only used to help understand the method and its core idea of the present invention; at the same time, for those of ordinary skill in the art, according to the idea of the present invention, there will be changes in the specific implementation methods and application scopes. In summary, the content of this specification should not be construed as a limitation to the present invention.

Claims

1. A method for inverting the morphological parameters of tectonic coal based on high-definition video, characterized in that Including: Collecting video images of underground tectonic coal; Wirelessly or wiredly transmitting the collected video images; Performing denoising, enhancement, and segmentation preprocessing on the transmitted video images, and extracting morphological feature parameters of tectonic coal; Establishing an inversion model of outburst mechanism based on the morphological feature parameters; Displaying the inversion result of outburst mechanism according to the inversion model of outburst mechanism; 2. The method for inverting coal morphological parameters based on high-definition video according to claim 1, wherein The morphological feature parameters of the tectonic coal include: joint parameters and coal seam and rock stratum parameters; the joint parameters include: joint width, length, direction angle, density, and spatial distribution law; the coal seam and rock stratum parameters include: dip angle of coal seam and rock stratum and coal seam thickness.

3. The method for inverting coal morphology parameters based on high-definition video according to claim 2, wherein Performing denoising, enhancement, and segmentation preprocessing on the transmitted video images, and extracting morphological feature parameters of tectonic coal, including: Using an adaptive Gaussian filtering algorithm to suppress noise of the video images to eliminate uneven illumination and dust interference in the underground environment, and obtaining a filtered image; Enhancing details of the filtered image through histogram equalization and local contrast stretching techniques to highlight the joint and coal seam and rock stratum boundary features of tectonic coal, and obtaining an enhanced image; Based on an improved Otsu multi-threshold segmentation algorithm, dividing the enhanced image into coal body region, joint region, and background region; Using a Canny edge detection algorithm to extract joint contours of the joint region, and combining morphological closing operation to eliminate small fractures to calculate joint width, length, and direction angle; Detecting joint linear features of the joint contours through Hough transform, and combining density clustering algorithm to statistically analyze joint density and spatial distribution law; Based on the coal body region, using the least squares fitting method to calculate the dip angle of coal seam and rock stratum, and inversely calculating the coal seam thickness through pixel calibration technology.

4. The method for inverting coal morphology parameters based on high-definition video according to claim 3, wherein, The calculation formula of the adaptive standard deviation in the adaptive Gaussian filtering algorithm is as follows: Among them, σ(x,y) is the adaptive standard deviation, where x and y respectively represent the coordinate values of the x-axis and y-axis of the current pixel in the video image; σ base is the preset basic standard deviation, I(x,y) is the gray value of the pixel point (x,y), I max is the maximum gray value within the local window, γ is the illumination compensation coefficient to control the adjustment amplitude of uneven illumination, and the value range of γ is 0.5 to 1.5; is the local gradient amplitude to reflect the dust interference intensity, where G x and G y are the Sobel convolution results in the horizontal and vertical directions respectively; is the maximum gradient amplitude within the local window, η is the dust interference suppression coefficient, and the value range of the dust interference suppression coefficient is 0.3 to 0.

8.

5. The method for inverting coal morphological parameters based on high-definition video according to claim 3, wherein The calculation formula of the improved Otsu multi-threshold segmentation algorithm is: Among them, both t1 and t2 are segmentation thresholds to divide the enhanced image into three categories: coal body area, joint area, and background area; To optimize the objective function and maximize the value of the optimized objective function to determine the optimal threshold; ω i is the probability of the i-th type of pixel, N i is the number of pixels of the i-th type, and N is the total number of pixels; μ i is the average gray level of the i-th type, h(k) is the gray-level histogram; μ T is the total average gray level of the image, L is the number of gray levels; λ is the preset continuity constraint weight coefficient, C i is the regional continuity constraint term, β is the gray-level distribution weight coefficient, W i is the gray-level distribution weight factor, 6. The method for inverting coal morphological parameters based on high-definition video according to claim 1, wherein Establishing an inversion model of outburst mechanism based on the morphological feature parameters, including: Performing multi-source data fusion on the morphological feature parameters of the tectonic coal, preset geomechanical parameters, and gas parameters, and eliminating dimension differences through normalization processing to generate a standardized input data set; Based on a random forest feature importance evaluation algorithm, screening key parameters that have a significant impact on outburst danger in the standardized input data set; the key parameters include joint density, dip angle of coal seam and rock stratum, and gas pressure; Using the analytic hierarchy process combined with expert experience to assign dynamic weights to key parameters; the weight values of the dynamic weights are adjusted according to real-time underground working conditions; Establishing a hybrid prediction model based on support vector machine and long short-term memory network; Training the hybrid prediction model using historical outburst accident data, evaluating the model accuracy through five-fold cross-validation, and optimizing hyperparameters to obtain the trained inversion model of outburst mechanism; Inputting the key parameters and the dynamic weights into the inversion model of outburst mechanism to output the inversion result of outburst mechanism.

7. The method for inverting coal morphological parameters based on high-definition video according to claim 6, characterized in that, The inversion result of outburst mechanism includes: outburst danger level and probability prediction result.

8. An inversion system for the morphological parameters of tectonic coal based on high-definition video, characterized in that, Including: A video acquisition module for collecting video images of underground tectonic coal; A data transmission module for wirelessly or wiredly transmitting the collected video images; A data processing module for preprocessing the transmitted video image by denoising, enhancing and segmenting, and extracting the morphological feature parameters of tectonic coal; An inversion model module for establishing an outburst mechanism inversion model based on the morphological feature parameters; A result display module for displaying the outburst mechanism inversion result according to the mechanism inversion model.

Images