An intelligent hidden danger investigation system and method for coal mine safety accidents

Abstract

The present application relates to the technical field of industrial data analysis, in particular to a kind of intelligent hidden danger investigation system and method of coal mine safety accident, comprising: the camera on monitoring trolley is used to collect coal mine image;In the coal mine image, dust area is segmented with lighting lamp as reference area, and the gray scale performance of dust area in light irradiation direction is taken as reference point dust concentration;The change rate difference of multiple reference point dust concentration is used to judge whether multiple reference points are in the same dust diffusion environment;If in the same dust diffusion environment, estimate current environmental dust concentration according to reference point dust concentration, if not in the same dust diffusion environment, move monitoring trolley to make reference point located in the same diffusion environment;Using the estimated current environmental dust concentration, intelligent hidden danger investigation of coal mine safety accident is carried out.The present application effectively improves the flexibility of coal mine monitoring and hidden danger investigation effect.

Description

Technical Field

[0001] This application relates to the field of industrial data analysis technology, specifically to an intelligent hazard investigation system and method for coal mine safety accidents. Background Technology

[0002] In the coal mining industry, safety has always been a crucial issue concerning the lives of miners and the safety of mine production. During underground operations, coal dust, especially coal dust, rock dust, and other dust generated during the mining process, has become a major hidden danger to coal mine safety. Coal dust not only harms the health of miners but can also trigger coal mine explosions. Especially in the confined space of underground mines, the hazards of explosions are amplified, posing a significant risk to the lives of workers and the safety of equipment. How to effectively monitor, investigate, and control dust hazards in coal mines has become an important issue in coal mine safety management.

[0003] Traditional dust monitoring in mines typically uses sensors, which are usually located in fixed positions. However, the complex underground environment makes fixed-position sensors inflexible and increases the workload for maintenance. Mobile monitoring devices are also used, which monitor dust at different locations by pre-setting a trajectory. However, the changing underground environment makes traditional mobile monitoring devices, which are generally single-point monitoring devices, unsuitable for the diverse underground environment and thus unable to effectively detect potential safety hazards. Summary of the Invention

[0004] This invention provides an intelligent hazard investigation system and method for coal mine safety accidents to solve existing problems.

[0005] The intelligent hazard detection system and method for coal mine safety accidents of the present invention adopts the following technical solution:

[0006] One embodiment of the present invention provides an intelligent method for identifying potential hazards in coal mine safety accidents, the method comprising the following steps:

[0007] The monitoring vehicle is used to acquire real-time images of the current coal mine environment.

[0008] In a coal mine image, the location of the lighting lamp is used as a reference point to segment out the dust area; the grayscale distribution of the dust area in the direction of the light illumination is used as the dust concentration reference point.

[0009] Draw a line connecting the reference point with the highest dust concentration to the reference points of other reference points, minimizing the distance between all reference points and the line connecting the reference points; the dust concentrations of all reference points passing through the line connecting the reference points form a dust concentration sequence; calculate the difference D in the rate of change of dust concentration in the dust concentration sequence; if the difference D is greater than a preset threshold, it is determined that the dust concentration sequences are not in the same dust diffusion environment; if D is less than or equal to the preset threshold, it is determined that the dust concentration sequences are in the same dust diffusion environment;

[0010] When the dust concentration sequence is not in the same dust diffusion environment, the mobile monitoring vehicle continues until the dust concentration sequence at the current moment is in the same dust diffusion environment, and the average value of the dust concentration sequence is taken as the dust concentration G of the current environment.

[0011] The obtained dust concentration G in the current environment is used to assess potential safety hazards and to identify potential hazards for coal mine safety accidents.

[0012] In a coal mine image, the dust region is segmented using the location of the lighting fixtures as a reference point. The specific steps involved are as follows:

[0013] In the coal mine image, the lighting area is segmented out, and the center of the tangent circle within the lighting area is used as the reference point;

[0014] A circular region with a preset radius centered at a reference point is defined, and the area within this circular region excluding the lighting area is designated as the reference region with the best concentration performance.

[0015] Threshold segmentation is performed within the reference area to obtain multiple dust regions.

[0016] Using the grayscale distribution of the dust area along the direction of light illumination as a reference point for dust concentration, the specific steps include the following:

[0017] A1: Within the reference area, the direction of the straight line passing through the reference point and having the maximum average gray value is taken as the direction of light illumination;

[0018] A2: The feature representation range is a pre-defined fan-shaped area within the reference area with the direction of light illumination as the axis.

[0019] A3: Based on the gray level and gray level changes within the feature representation range, determine the dust concentration at the reference point, wherein the dust concentration is positively correlated with the gray level and gray level changes.

[0020] The specific process of drawing a line connecting the reference point with the highest dust concentration to other reference points is as follows:

[0021] In the coal mine image, select the reference point with the highest dust concentration as the starting point, and draw preliminary lines from the starting point to other reference points. Select the preliminary line with the smallest distance from all reference points as the reference point line.

[0022] The dust concentration sequence is formed by connecting all reference points through the aforementioned reference points, and the specific steps involved are as follows:

[0023] B1: In the coal mine image, starting from the line connecting the reference points, extend in a direction perpendicular to the line connecting the reference points, with an extension width of r, to obtain a rectangular bounding box with a width of 2r, where r is a preset value;

[0024] B2: Use all reference points within the rectangular bounding box as reference points for the lines connecting these reference points;

[0025] B3: Arrange all reference points according to their projection positions on the line connecting the reference points to obtain a dust concentration sequence composed of the dust concentrations of the reference points.

[0026] The specific calculation process for the difference D in the rate of change of dust concentration in the dust concentration sequence is as follows:

[0027] First, calculate the rate of change of all dust concentrations in the dust concentration sequence. Then, cluster the rate of change of all dust concentrations in the sequence to obtain the cluster centers. The mean of the differences between the rate of change of all dust concentrations and the cluster centers is taken as the difference D of the rate of change.

[0028] The specific calculation process for calculating the rate of change of all dust concentrations in the dust concentration series is as follows:

[0029] A polynomial fit was performed on all dust concentrations in the dust concentration sequence, and the slope of the tangent line of the fitted curve corresponding to each fitted point was taken as the rate of change of dust concentration at the reference point corresponding to the fitted point.

[0030] The specific process of the mobile monitoring vehicle is as follows:

[0031] The direction of movement of the monitoring trolley is preset, and the movement distance L of the monitoring trolley is determined by the change of the dust concentration sequence. The movement distance L is positively correlated with the change of the dust concentration sequence.

[0032] The specific process for assessing safety hazards based on the obtained current environmental dust concentration G is as follows:

[0033] A preset concentration threshold G_th is set. When the current ambient dust concentration G is greater than the concentration threshold G_th, the current ambient dust concentration is determined to be an ambient dust concentration that poses a safety hazard.

[0034] Another embodiment of the present invention provides an intelligent hazard investigation system for coal mine safety accidents, including a sensor module, a computing module, a main control module, and a communication module. The data collected by the sensor module includes coal mine images and the position of the monitoring trolley. The data collected by the sensor module is sent to the computing module through the communication module. The computer program in the computing module calculates the moving direction and moving distance of the monitoring trolley and the ambient dust concentration based on the data collected by the sensors. The computing module sends the moving direction and moving distance of the monitoring trolley to the main control module through the communication module. The main control module is used to control the movement of the monitoring trolley. The computer program implements the steps of the intelligent hazard investigation method for coal mine safety accidents.

[0035] The beneficial effects of the technical solution of the present invention are as follows: First, the use of a mobile monitoring trolley can monitor different locations in the coal mine, overcoming the tendency of fixed sensors to miss dynamic areas, while avoiding long-term maintenance of a large number of sensors, effectively reducing operating costs; at the same time, by using the coal mine lighting lamps as reference points, dust concentrations at multiple locations in an image can be obtained, enabling a single sampling device to obtain multiple sampling data, improving sampling efficiency and reducing sampling costs; then, by utilizing the movement of the monitoring trolley, multiple reference points in the image are located in the same dust diffusion environment, making it easier to estimate the environmental dust concentration through multiple sampling points. Attached Figure Description

[0036] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0037] Figure 1 This is a flowchart illustrating the steps of an intelligent method for identifying potential safety hazards in coal mines, as described in this invention. Detailed Implementation

[0038] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of an intelligent hazard investigation system and method for coal mine safety accidents proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0039] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0040] The following description, in conjunction with the accompanying drawings, details the specific scheme of the intelligent hidden danger investigation system and method for coal mine safety accidents provided by this invention.

[0041] Please see Figure 1 The diagram illustrates a flowchart of an intelligent hazard detection method for coal mine safety accidents according to an embodiment of the present invention. The method includes the following steps:

[0042] S001. Deploy the monitoring trolley components and set relevant parameters, while simultaneously using a camera to collect images of the coal mine.

[0043] In coal mines, the long-term high dust concentration makes the monitoring efficiency of general sensors poor. However, there are usually lights in the mine, and the dust is more obvious under the light. Therefore, a monitoring trolley with a camera is used to acquire coal mine images. Dust is identified in the captured coal mine images, and the dust concentration is estimated to achieve the purpose of identifying safety hazards.

[0044] It should be noted that coal mine images are obtained by capturing images from a camera mounted on a monitoring trolley. Three consecutive frames of coal mine images captured by the camera are considered as a single monitoring data point, with a time interval of 0.5 seconds between consecutive captures. The coal mine images undergo grayscale processing, a well-known technique.

[0045] Monitoring vehicle: Features a designed mechanical structure, and includes appropriate sensor modules, main control and computing modules, and communication modules. Specific component descriptions are as follows:

[0046] (1) Monitoring cart components and parameter descriptions:

[0047]

[0048] (2) Sensors for acquiring data and parameter descriptions:

[0049]

[0050] (3) Description of components and parameters for main control computing:

[0051]

[0052] (4) Parameter description of the communication module:

[0053]

[0054] S002. In the coal mine image, the dust area is segmented with the position of the lighting lamp as a reference point, and the dust concentration at the reference point is determined in the obtained dust area based on the direction of the lighting lamp.

[0055] The overall environment in mines is generally dark, and the coal mine images collected contain a large number of dark areas. Furthermore, the lighting in mines is uneven, and high-contrast areas are easily formed around the lighting fixtures, making imaging relatively easy. Under the lighting, the contrast of dust particles is more obvious, making it easier to identify dust areas in coal mine images, which is helpful for estimating the concentration of environmental dust.

[0056] Therefore, the lighting area in the coal mine image has good prerequisites for dust identification. So in the actual monitoring process, the lighting area needs to be identified in the coal mine image first for subsequent dust identification and dust concentration estimation.

[0057] As one example, the specific process of segmenting the lighting area with high brightness in a coal mine image is as follows:

[0058] (1) The coal mine image is segmented into multiple regions using the watershed algorithm.

[0059] The watershed algorithm is an existing technology, and its implementation process will not be described in detail in this embodiment.

[0060] (2) Calculate the mean gray value of each region.

[0061] (3) Cluster the gray mean values ​​of the multiple regions obtained to obtain several clusters. Specifically, the mean drift clustering method is adopted. The mean drift clustering method is an existing basis, and its implementation process will not be described in detail in this embodiment.

[0062] (4) In the mean-shift clustering results, select the area corresponding to the cluster with the largest gray-scale mean as the lighting area.

[0063] As a supplementary embodiment, further screening of the lighting area is required, specifically:

[0064] In the mean-shift clustering results, the region corresponding to the cluster with the largest gray-scale mean is selected, and the roundness of each selected region is calculated. In this embodiment, the standard deviation method is used to calculate the roundness. The standard deviation method is an existing technology and will not be described in detail in this embodiment. Then, a roundness threshold is preset, and the region with a roundness greater than the roundness threshold is taken as the lighting area. The roundness threshold can be set to 0.8 in this embodiment, and can be modified in other embodiments.

[0065] In some other embodiments, neural networks are also used to identify the lighting area. The specific process is as follows:

[0066] (1) Data tagging:

[0067] Collect 3,000 images of underground coal mine lighting and mark the bounding boxes; the resolution of the coal mine images is uniformly 1280×720.

[0068] (2) Model training:

[0069] The pre-trained YOLOv5s model was used, with the input size adjusted to 640×640, and the optimizer was SGD (initial learning rate 0.01, momentum 0.937); the loss function was the cross-entropy loss function.

[0070] The training period is 300 epochs, the batch size is 16, and the data augmentation strategies include random rotation (±15°), brightness jitter (±20%), and adding dust fogging effect.

[0071] (3) Identification of lighting areas:

[0072] The captured coal mine images are input into the trained model to identify the lighting areas in the coal mine images.

[0073] Furthermore, after identifying the lighting areas in the coal mine image, it is necessary to determine which specific areas of dust are visible under the illumination of the lights. It is known that the lighting areas identified in the above process are only the areas where the lights are located, not the areas illuminated by the lights. Therefore, the lighting areas are not necessarily the optimal areas for dust concentration. Therefore, it is also necessary to determine an area where dust concentration is optimal, using the lighting areas as a reference; this is called the reference area.

[0074] The obtained reference area with the best concentration performance has the best illumination and obvious contrast. In the reference area with the best concentration performance, the dust has a higher grayscale performance due to the illumination. At this time, based on the high grayscale performance of the dust, the dust area is segmented in the reference area.

[0075] As one embodiment, the process of dividing the dust area within the reference area is as follows:

[0076] (1) For a lighting area, calculate the center of its corresponding inscribed circle and use the obtained center as a reference point. The calculation of the inscribed circle is an existing technology and will not be described in detail in this embodiment.

[0077] (2) Based on the reference point, a circular area is expanded outward, with an expansion radius of x = k × m, where m is the farthest distance from the reference point to the boundary of the corresponding lighting area, and k is a scaling factor. In this embodiment, k = 1 can be set, while in other embodiments, it can be set as needed. After excluding the lighting area, the expanded circular area is the reference area with the best concentration performance corresponding to the current reference point.

[0078] (3) Perform adaptive threshold segmentation within the reference area to obtain the dust area. Adaptive threshold segmentation is an existing technology and will not be described in detail in this embodiment.

[0079] Furthermore, within the reference area corresponding to the above process, due to the high contrast of the lighting area underground, areas with poor dust representation may also be segmented as dust areas. In other words, these segmented dust areas lack sufficient representation of the current environmental dust concentration. Mine lighting typically has a direction of illumination. Due to the competition between direct light and dust obstruction in the main illumination direction, the grayscale peak appears in the area with the largest dust concentration gradient, indicating the diffusion front. That is, dust has the greatest representation in the direction of the lighting illumination.

[0080] Therefore, in order to obtain the current dust performance of the environment, it is necessary to calculate the dust concentration in the direction of the lighting.

[0081] As one example, the dust concentration calculation process is as follows:

[0082] (1) Within the reference area, draw line segments that connect the edges of the reference area starting from the reference point. Select the line segment with the largest average gray value as the direction of light illumination. The gray value of the line segment is the gray value of the pixel point in the reference area through which the line segment passes.

[0083] (2) Within the reference area, a sector with a preset central angle of θ0 and a central axis that is the direction of light illumination is obtained. The obtained sector is the feature representation range. In this embodiment, the central angle θ0 can be set to 15 degrees. Other embodiments can be set by themselves.

[0084] (3) Based on the average gray level and gray level changes within the characteristic range, both overall concentration and local abrupt changes are captured to adapt to the scenario where underground dust exhibits both "static accumulation" and "dynamic diffusion," ultimately obtaining the dust concentration GD at the current reference point:

[0085] GD = P0 × DP

[0086] In the formula, P0 represents the mean gray value of the current feature representation range, and DP represents the absolute value of the difference between the mean gray value of the coal mine image corresponding to the current feature representation range at the current time and the coal mine image corresponding to the previous time.

[0087] In some other embodiments, the dust concentration GD at the current reference point is calculated using the following formula:

[0088] GD = a1 × P0 + a2 × DP

[0089] In the formula, P0 represents the mean gray value of the current feature representation range, DP represents the mean gray value difference between the current feature representation range and the corresponding coal mine image at the current time and the previous time, and a1 and a2 are weight coefficients, which can be set to a1 = 0.3 and a2 = 0.7.

[0090] At this point, the dust concentration at the reference point corresponding to the location of the lighting fixture is obtained.

[0091] S003. Determine whether the reference point is located in the same diffusion environment, and move the monitoring trolley to estimate the current environmental dust concentration based on the dust concentration of the reference point in the same diffusion environment.

[0092] The above process uses the location of the lighting fixtures as reference points to segment dust areas in the coal mine image, and determines the dust concentration at each reference point based on the dust distribution under the lighting. The dust concentration determined at this point represents the coal mine image of a single area. However, a single area cannot represent the current environmental dust concentration. Therefore, it is necessary to use the dust concentrations of multiple reference points to estimate the overall current environmental dust concentration, thereby enabling the safety investigation of dust hazards.

[0093] Furthermore, when estimating the current environmental dust concentration using dust concentrations at multiple reference points, the dust concentrations at these reference points reflect the effects of dust expansion in the corresponding space. Since underground mine spaces are often tunnels, lighting fixtures are typically arranged relatively evenly within them. Therefore, it is sufficient to select several reference points with relatively high dust concentrations and construct a dust concentration sequence from these selected reference points to estimate the current environmental dust concentration.

[0094] As one example, the process of determining the dust concentration sequence is as follows:

[0095] (1) Among the reference points in the coal mine image, select the reference point with the highest dust concentration as the starting point, and make preliminary connections with other reference points from the starting point.

[0096] (2) In each preliminary connection, calculate the sum of the distances from all reference points to the preliminary connection, and select the preliminary connection with the smallest sum of distances as the reference point connection.

[0097] (3) The reference points located on the line connecting the reference points are used as reference points passing through the line connecting the reference points. The arrangement method is that the distance between the reference points located on the line connecting the reference points and the starting point is from near to far.

[0098] (4) Analyzing the concentration change along the diffusion path (connecting directions) from near to far can capture the gradient characteristics of dust diffusion. Therefore, starting from the starting point, reference points that pass through the reference point are arranged in order of distance from the starting point. The arranged reference points form a sequence, and the corresponding dust concentrations form a dust concentration sequence.

[0099] As another embodiment, the process of determining the dust concentration sequence is as follows:

[0100] (1) Among the reference points in the coal mine image, select the reference point with the highest dust concentration as the starting point, and make preliminary connections with other reference points from the starting point.

[0101] (2) In each preliminary connection, calculate the sum of the distances from all reference points to the preliminary connection, and select the preliminary connection with the smallest sum of distances as the reference point connection.

[0102] (3) Starting from the line connecting the reference points, expand outwards to both sides with a preset expansion width of r, obtaining a rectangular bounding box with a width of 2r. Use all reference points within the rectangular bounding box as reference points for the line connecting the reference points. In this embodiment, r = 0.2m can be set. CFD simulation shows that when the downhole wind speed is 1.0m / s, the standard deviation of dust lateral diffusion σ = 0.18m requires a width of 2r = 4σ ≈ 0.72m to cover 95% of the particles. The actual value is rounded down to 2r = 0.4m, i.e., r = 0.2m. Other embodiments can be modified accordingly.

[0103] (4) Starting from the starting point, the reference points that pass through the reference points are arranged in order of distance from the starting point to the farthest point. The arranged reference points form a sequence, and the corresponding dust concentrations form a dust concentration sequence.

[0104] As another embodiment, the process of determining the dust concentration sequence is as follows:

[0105] (1) The positions of all reference points in the coal mine image are fitted with straight lines, and the straight line fitting method is the least squares method.

[0106] (2) Among all the fitted lines, select the line with the smallest root mean square error as the reference point and connect them.

[0107] (3) Set a preset distance threshold d0, select reference points whose distance to the line connecting the reference points is less than d0 as reference points passing through the line connecting the reference points, project the positions of the reference points passing through the line connecting the reference points onto the line connecting the reference points, obtain the projected position of each reference point passing through the line connecting the reference points, obtain a reference point sequence according to the projected position, and the dust concentration corresponding to the reference point sequence constitutes a dust concentration sequence. In this embodiment, d0 can be set to 0.2m, and other embodiments can be modified accordingly.

[0108] As a supplementary embodiment, noise removal is required for the dust concentration sequence. The specific method is as follows:

[0109] If the concentration of a selected reference point differs from the dust concentration of the preceding and following reference points by more than three times the standard deviation, it is determined to be noise, and the mean of adjacent points is used instead.

[0110] The obtained dust concentration sequence is used to estimate the current environmental dust concentration. Changes in this sequence primarily reflect variations in dust concentration at different locations within the mine, and also indicate dust diffusion in the current environment. Because there are relationships between dust concentrations at different locations within the same dust diffusion area, these relationships can be used to estimate the concentration at different locations, and consequently, to estimate the current environmental dust concentration.

[0111] Furthermore, because analyzing concentration changes along the direction of dust diffusion from near to far can capture the gradient characteristics of dust diffusion, the relationship between the obtained dust concentration sequences can be used to analyze whether the dust concentration sequences are located in the same dust diffusion region. The greater the difference in the rate of change of dust concentrations in the dust concentration sequence, the more likely the current environmental dust concentration is to be estimated. If the dust concentration sequences are in the same dust diffusion region, there is a relationship between the concentrations at reference points, so the current environmental dust concentration can be estimated using the dust concentration sequence; that is, the current environmental dust concentration is estimable. Conversely, the current environmental dust concentration cannot be estimated using the dust concentration sequence; that is, the current environmental dust concentration is not estimable.

[0112] As one embodiment, the specific process of calculating the difference in the rate of change of dust concentration between dust concentrations in a dust concentration sequence and determining whether the current environmental dust concentration is estimable is as follows:

[0113] (1) Perform polynomial fitting on the dust concentration sequence to obtain the fitting curve, calculate the slope of the tangent line on the fitting curve for each dust concentration in the dust concentration sequence, and use it as the rate of change of a single dust concentration, wherein the fitting method is least squares fitting.

[0114] (2) Cluster the rate of change to obtain the cluster center. In this embodiment, the K-means clustering algorithm is used and K=1 is set.

[0115] (3) Calculate the absolute value of the difference between the cluster center and all cluster points respectively, and obtain the mean of the difference as the difference D of the rate of change.

[0116] (4) A preset difference threshold M is set. When the difference in the rate of change D is less than the difference threshold M, it is determined that the current environmental dust concentration can be estimated. Otherwise, it cannot be estimated. In this embodiment, the difference threshold M is set to 20.

[0117] As another embodiment, the specific process for calculating the difference in the rate of change of dust concentrations in the dust concentration sequence is as follows:

[0118] (1) Given that the process of determining dust concentration corresponds to a reference point line, first calculate the straight-line distance from the reference point to the reference point line for all dust concentrations in the dust concentration sequence.

[0119] (2) The reciprocal of the obtained straight-line distance is used as the fitting weight coefficient of the corresponding dust concentration. Polynomial fitting is performed to obtain the fitting curve. The slope of the tangent line of each dust concentration in the dust concentration sequence on the fitting curve is calculated as the rate of change of a single dust concentration. The fitting method is least squares fitting.

[0120] (3) Cluster the rate of change to obtain the cluster center. In this embodiment, the K-means clustering algorithm is used and K=1 is set.

[0121] (4) Calculate the difference between the cluster center and all cluster points respectively, and use the reciprocal of the straight distance as the summation weight to obtain the weighted mean of the dust concentration sequence as the difference D of the rate of change.

[0122] Furthermore, when the dust concentration sequence is not in the same dust diffusion area, i.e. the current environmental dust concentration is unpredictable, it is necessary to adjust the position of the monitoring vehicle so that the reference point in the coal mine image captured by the monitoring vehicle is in the same dust diffusion area, thereby accurately estimating the current environmental dust concentration.

[0123] The adjustment of the monitoring trolley's position mainly involves its direction of movement and the distance it travels. The direction of movement is primarily aimed at obtaining areas with better concentration performance, higher dust concentrations, and greater hazard factors.

[0124] As one embodiment, the method for determining the movement direction of the monitoring vehicle is as follows:

[0125] The monitoring vehicle is equipped with a positioning and navigation system. Its positioning module can load a preset map or build a map in real time using SLAM (Cartographer algorithm) and determine the current position of the monitoring vehicle on the map. In addition, the monitoring vehicle sets a fixed monitoring route before monitoring. In this embodiment, the monitoring route is set manually. At this time, the next direction of movement of the monitoring vehicle at the current position is determined based on the set monitoring route.

[0126] It should be noted that the more drastic the fluctuations in the concentration sequence (the worse the environmental consistency), the greater the distance the monitoring vehicle needs to travel to approach the stable region. Therefore, the travel distance of the monitoring vehicle needs to be determined based on the changes in the dust concentration sequence to get closer to the dust source and facilitate the estimation of the current environmental dust concentration.

[0127] The movement of the trolley is monitored by the STM32 sending movement commands through the CAN bus, the PID controller adjusting the motor speed, and the encoder feeding back the actual movement distance, with an error compensation accuracy of ±0.05m.

[0128] As one embodiment, the method for determining the distance traveled by the monitoring vehicle is as follows:

[0129] Calculate the variance of all dust concentrations in the dust concentration sequence, and determine the moving distance L of the monitoring vehicle based on the variance:

[0130] L=α×σ1

[0131] In the formula, σ1 represents the variance of all dust concentrations in the dust concentration sequence, and α is a preset first proportionality coefficient. In this embodiment, α can be set to 0.5, and other embodiments can be set by themselves.

[0132] Results obtained by the method provided in this embodiment under an artificially simulated dusty environment:

[0133] Conditions: Wind speed 1.2 m / s, concentration gradient 20-80 mg / m³ 3 .

[0134] Results: When σ1 = 2.4, the moving distance L = 1.2, and the concentration estimation error after moving decreased from 25% to 8%; when α = 0.5, the average number of moves was 3.2 to cover the hazard area, which reduced the monitoring time by 42% compared with a fixed step length (such as L = 0.3 mL = 0.3 m).

[0135] As another embodiment, the method for determining the distance traveled by the monitoring vehicle is as follows:

[0136] Among all reference points obtained in the coal mine image, the reference point with the highest dust concentration is selected as the target point. Then, a straight line passing through the reference point and parallel to the horizontal axis of the coal mine image is drawn to divide the coal mine image into two regions. The number of reference points ma in the region above the straight line and the number of reference points mb in the region below the straight line are counted. Based on the number ma and the number mb, the moving distance L of the monitoring trolley is determined as follows:

[0137] L = β × Sigmoid(ma - mb)

[0138] In the formula, ma and mb represent the number of reference points in the region above the straight line and the number of reference points in the region below the straight line in the coal mine image, respectively, and β is a preset second proportional coefficient. In this embodiment, β can be set to 5, and other embodiments can be set by themselves.

[0139] After the monitoring vehicle moves once, it corresponds to a new monitoring location. Dust concentration monitoring is then performed again at the new location, and the following steps are continued:

[0140] (S1) Take images of the coal mine, see step S001 for details;

[0141] (S2) In the coal mine image, determine the dust concentration at the reference point, see step S002 for details;

[0142] (S3) Determine whether the current environmental dust concentration can be estimated, see step S003 for details;

[0143] (S4) If the concentration cannot be estimated, move the monitoring vehicle and execute steps (S1), (S2), (S3), and (S4); if the concentration can be estimated, calculate the current ambient dust concentration.

[0144] Furthermore, when it is determined that the environmental dust concentration at the location of the monitoring vehicle is estimable, that is, when multiple reference points corresponding to the dust concentration sequence are located in the same dust diffusion environment, the dust concentrations between the dust concentration sequences are correlated, that is, the dust concentrations between the dust concentration sequences are estimable. Therefore, the current environmental dust concentration can be estimated by using the overall dust concentration of the dust concentration sequence.

[0145] As one example, the specific process for determining the current dust concentration G in the environment is as follows:

[0146] Given the specific concentration values ​​of the dust concentration sequence, calculate the mean of all concentration values ​​in the dust concentration sequence. The obtained mean is taken as the current environmental concentration G.

[0147] In some other embodiments, the accuracy of the current environmental dust concentration estimation is ensured by filtering points in the dust concentration sequence. Specifically, the method is as follows:

[0148] First, a polynomial fitting is performed on the dust concentration sequence to obtain the fitting curve, where the fitting method is the least squares method. Then, the regression difference of each fitting point is calculated. The preset regression difference threshold is 5. Fitting points with regression differences greater than the set regression difference threshold are removed. The remaining fitting points are used as the data source for estimating the current environmental dust concentration. That is, the average dust concentration of the remaining fitting points is taken as the current environmental dust concentration G.

[0149] As another embodiment, considering the influence of the concentration gradient in the dust concentration sequence, the specific process for determining the current environmental dust concentration G is as follows:

[0150] First, calculate the mean concentration G0 of the dust concentration sequence, then calculate the standard deviation σ0 of the concentration gradient of the dust concentration sequence, and further estimate the current environmental dust concentration as follows:

[0151] G = k1 × G0 + k2 × σ0

[0152] Where G0 is the mean concentration of the dust concentration sequence, σ0 is the standard deviation of the concentration gradient of the dust concentration sequence, and k1 and k2 are the preset third proportional coefficient and the preset fourth proportional coefficient, respectively. In this embodiment, k1 = 0.7 and k2 = 0.3 can be set.

[0153] The following are simulation experiments conducted under different dust concentrations, where the unit of dust concentration is mg / m³. 3 The dust concentration estimation results obtained in this embodiment are shown in the table below:

[0154] dust concentration average concentration Gradient Standard Deviation Estimated concentration error(%) 8 105 10.2 8.1 +1.25 15 75 18.5 14.7 -2.0 22 60 25.3 21.9 -0.45

[0155] The experimental results show that the current environmental dust concentration estimation method provided in this embodiment has a relatively accurate estimation effect under different dust concentrations.

[0156] S004. Utilize the estimated current ambient dust concentration to conduct intelligent hazard investigation for coal mine safety accidents.

[0157] After obtaining the current ambient dust concentration, a hazard warning is issued based on this concentration, enabling intelligent hazard identification for safety accidents. Specifically, the first step is to determine the dust concentration at which a potential hazard exists.

[0158] As one example, the specific process for determining the concentration of dust posing a potential hazard is as follows:

[0159] After determining the current ambient dust concentration, the monitoring trolley is kept stationary and three sets of monitoring data are repeatedly collected, with a 10-second time interval between each set. This results in three sets of data representing three different ambient dust concentrations. The average of these three concentrations is calculated as G_f, and then a concentration threshold G_th is set to 10 mg / m³. 3 If the average value G_f is greater than the set concentration threshold G_th, the warning information is uploaded to the control center via 5G / LoRa, and an audible and visual alarm is triggered locally.

[0160] The method provided in this embodiment can effectively avoid misjudgment and unnecessary panic. Compared with traditional fixed sensor monitoring, this method has obvious advantages, as shown in the table below:

[0161] index This method Traditional fixed sensors Accuracy of hazard identification 93.5％ 74.2％ Average number of false alarms per month 0.9 6.3

[0162] Thus, by monitoring coal mine images captured by cameras mounted on a mobile vehicle, the current dust concentration in the environment can be estimated, and intelligent hazard identification for coal mine safety accidents can be achieved.

[0163] Another embodiment of the present invention provides an intelligent hazard investigation system for coal mine safety accidents, the system comprising: a sensor module, a computing module, a main control module, and a communication module.

[0164] The sensor module is used to acquire raw data, including taking pictures of the coal mine with a camera and determining the specific location of the monitoring trolley using a position sensor. The data collected by the sensors is sent to the computing module through the communication module. The computer program in the computing module calculates the direction and distance of movement of the monitoring trolley based on the data collected by the sensors, and calculates the current dust concentration in the environment. The computing module sends the direction and distance of movement of the monitoring trolley to the main control module through the communication module. The main control module is used to control the movement of the monitoring trolley, including controlling the direction and distance of movement of the monitoring trolley.

[0165] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An intelligent hidden danger investigation method for coal mine safety accidents, characterized in that, The method comprises the following steps: Real-time acquisition of the current environment of the coal mine image using a monitoring trolley; In the coal mine image, the position of the illuminating lamp is taken as a reference point, and a dust area is segmented; the gray scale distribution of the dust area in the light irradiation direction is taken as the reference point dust concentration; The reference point with the maximum dust concentration is connected with other reference points to form a reference point connecting line; the reference point with the maximum dust concentration is taken as a starting point, and a preliminary connecting line is drawn from the starting point to each of the other reference points; the reference point connecting line is selected as the preliminary connecting line with the minimum distance to all other reference points; the dust concentrations of all reference points passing through the reference point connecting line form a dust concentration sequence; the difference D of the dust concentration change rate in the dust concentration sequence is calculated; when the difference D of the dust concentration change rate is greater than a preset threshold value, it is determined that the dust concentration sequence is in a different dust diffusion environment; when the difference D of the dust concentration change rate is less than or equal to the preset threshold value, it is determined that the dust concentration sequence is in the same dust diffusion environment; When the dust concentration sequence is in a different dust diffusion environment, the monitoring trolley is moved until the dust concentration sequence at the current time is in the same dust diffusion environment; the average value of the dust concentration sequence is taken as the dust concentration G of the current environment; The obtained dust concentration G of the current environment is subjected to a safety hazard judgment to realize the hidden danger investigation of the coal mine safety accident.

2. The intelligent hazard investigation method for coal mine safety accidents according to claim 1, characterized in that, The specific steps of taking the position of the illuminating lamp as a reference point in the coal mine image to segment the dust area include the following: In the coal mine image, the illuminating lamp area is segmented, and the center of the circle inside the illuminating lamp area is taken as a reference point; A circular area with a preset radius is drawn with the reference point as the center; the area inside the circular area excluding the illuminating lamp area is recorded as a reference area with the best concentration performance; Threshold segmentation is performed in the reference area to obtain a plurality of dust areas.

3. The intelligent hidden trouble checking method for coal mine safety accidents according to claim 1, characterized in that, The specific steps of taking the gray scale distribution of the dust area in the light irradiation direction as the reference point dust concentration include the following: A1: In the reference area, the direction of the straight line passing through the reference point and having the maximum average gray scale value is taken as the light irradiation direction; A2: A preset sector area in the reference area and taking the light irradiation direction as the axis is taken as a feature performance range; A3: Based on the gray scale and gray scale change in the feature performance range, the dust concentration of the reference point is determined, and the dust concentration is positively correlated with the gray scale and gray scale change, respectively.

4. The intelligent hidden trouble checking method for coal mine safety accidents according to claim 1, characterized in that, The specific steps of forming a dust concentration sequence with the dust concentrations of all reference points passing through the reference point connecting line include the following: B1: In the coal mine image, the reference point connecting line is expanded in a direction perpendicular to the reference point connecting line, and the width of the expansion is r, to obtain a rectangular bounding box with a width of 2r, where r is a preset value; B2: All reference points in the rectangular bounding box are taken as reference points passing through the reference point connecting line; B3: All reference points are arranged according to the projection positions of the reference points on the reference point connecting line to obtain a dust concentration sequence composed of the dust concentrations of the reference points.

5. The intelligent hidden trouble checking method for coal mine safety accidents according to claim 1, characterized in that, The specific calculation process of the difference D of the dust concentration change rate in the dust concentration sequence is as follows: First, the rate of change of all dust concentrations in the dust concentration sequence is calculated, the rate of change of all dust concentrations in the sequence is clustered, the cluster center is obtained, and the average of the difference between the rate of change of all dust concentrations and the cluster center is taken as the difference D of the rate of change.

6. The intelligent hidden trouble checking method for coal mine safety accidents according to claim 5, characterized in that, The specific calculation process of calculating the rate of change of all dust concentrations in the dust concentration sequence is as follows: The dust concentration sequence is fitted by a polynomial, and the slope of the tangent line of the fitting curve corresponding to each fitting point is taken as the dust concentration rate of change of the reference point corresponding to the fitting point.

7. The intelligent hidden trouble checking method for coal mine safety accidents according to claim 1, characterized in that, The specific process of the moving monitoring trolley is as follows: The moving direction of the monitoring trolley is pre-set, and the moving distance L of the monitoring trolley is determined by the change of the dust concentration sequence, and the moving distance L and the change of the dust concentration sequence are in a positive correlation. 8.The intelligent hidden danger investigation method for coal mine safety accidents according to claim 1, characterized in that, The specific process of the safety hazard judgment of the obtained current environmental dust concentration G is as follows: Pre-set concentration threshold When the current ambient dust concentration G is greater than the concentration threshold , it is determined that the current ambient dust concentration is an ambient dust concentration with safety hazards.

9. An intelligent hidden trouble investigation system for coal mine safety accidents, comprising a sensor module, a calculation module, a main control module and a communication module, characterized in that, The data collected by the sensor module includes coal mine images and the position of the monitoring trolley; the data collected by the sensor module is sent to the computing module through the communication module, the computer program in the computing module calculates the moving direction, moving distance and environmental dust concentration of the monitoring trolley according to the data collected by the sensor, the computing module sends the moving direction and moving distance of the monitoring trolley to the main control module through the communication module, and the main control module is used to control the movement of the monitoring trolley, and the computer program realizes the steps of the intelligent hidden danger investigation method of coal mine safety accidents in any one of claims 1-8 when executed.

Images