Intelligent early warning system and method for coal mine safety accidents

Abstract

The invention relates to the technical field of coal mine safety early warning, in particular to an intelligent early warning system and method for coal mine safety accidents. The method comprises the following steps: acquiring an underground monitoring image of a coal mine; dust point locations belonging to high-frequency characteristics are extracted; performing graying processing and division to obtain image areas, and determining a gray discrete index of each image area according to numerical distribution discreteness of gray values of all pixels in each image area; analyzing the dust quality, and determining a quality correction index according to any height and a gray discrete index of an image area below the height; correcting the gray discrete index to obtain a corrected discrete value; and determining a possibility index by combining the number of the dust point locations in each image area and the corrected discrete value, and performing early warning judgment according to the numerical value of the possibility index of the adjacent acquisition time frames. According to the invention, the reliability of overall intelligent hidden danger checking can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of coal mine safety warning, and particularly relates to an intelligent warning system and method for coal mine safety accidents. Background Art

[0002] During the mining process inside a coal mine shaft, mechanical coal breaking and transportation processes are accompanied by the generation of some dust, which floats in the air. The dust hazard inside the mine is an important factor. Dust can easily cause respiratory diseases. When dust prevention management is not in place and the concentration of dust in the space reaches a certain level, if the equipment ages or malfunctions and generates sparks at this time, a flash explosion is extremely likely to occur, causing casualties and property losses. In the prior art, there is a method of analyzing the gray difference between the collected image and the standard image, and then analyzing the influence of dust for early warning. Due to the large space and complex ventilation conditions, the air flow is unstable, and the dust concentrations at different heights and different positions are inconsistent, making it difficult for a single monitoring point to accurately represent the entire dust situation, and the reliability of dust analysis based only on image gray change is poor; therefore, it is necessary to effectively identify the dust in the mine and conduct intelligent hidden danger investigation. Summary of the Invention

[0003] In order to solve the technical problem in the related art that due to the different influences of dust at different heights and inconsistent concentrations at different positions, the reliability of dust hidden danger investigation based only on image gray change is poor, the present invention provides an intelligent warning system and method for coal mine safety accidents, and the specific technical solutions adopted are as follows: The present invention proposes an intelligent warning system and method for coal mine safety accidents. The method includes: Obtain the coal mine underground monitoring images on each acquisition time frame; based on Gaussian filtering, obtain the high-frequency information and low-frequency information of the overall coal mine underground monitoring images, and extract the dust points belonging to the high-frequency features; Perform gray-scale processing on each coal mine underground monitoring image and evenly divide it to obtain different image regions. According to the numerical distribution discreteness of the gray values of all pixel points in each image region, determine the gray discreteness index of each image region; Analyze the dust quality. According to any height and the gray discreteness index of the image region below it, determine the quality correction index of each image region; combine the quality correction index to correct the gray discreteness index to obtain the corrected discrete value; Combine the number of dust points belonging to each image region and the corrected discrete value to determine the possibility index of dust hazard occurrence under the corresponding acquisition time frame, and perform early warning judgment according to the numerical values of the possibility index in adjacent acquisition time frames.

[0004] Further, obtaining the high-frequency information and low-frequency information of the overall monitoring image in the coal mine underground based on Gaussian filtering, and extracting the dust points belonging to the high-frequency features, including: Performing smoothing processing on the monitoring image in the coal mine underground based on Gaussian filtering as the low-frequency image; Calculating the ratio of the brightness values of the monitoring image in the coal mine underground and the low-frequency image at the same point as the high-frequency feature index at the corresponding point; Determining the dust points according to the numerical value of the high-frequency feature index.

[0005] Further, the determining the dust points according to the numerical value of the high-frequency feature index includes: Taking the points where the high-frequency feature index is greater than the preset index threshold as the dust points under the high-frequency feature.

[0006] Further, the determining the gray-scale discrete index of each image region according to the numerical distribution discreteness of the gray-scale values of all pixel points in each image region includes: Calculating the standard deviation of the gray-scale values of all pixel points in the same image region as the gray-scale discrete index.

[0007] Further, the determining the quality correction index of each image region according to any height and the gray-scale discrete index of the image region below it includes: Taking any point height as the target height and all point heights below the target height as the candidate heights; Calculating the sum value of the gray-scale discrete indexes under the target height and all candidate heights, and performing normalization processing as the quality correction index.

[0008] Further, the correcting the gray-scale discrete index by combining the quality correction index to obtain a corrected discrete value includes: Calculating the product value of the quality correction index and the gray-scale discrete index of the image region as the corrected discrete value.

[0009] Further, the determining the possibility index of dust hazard occurrence in the corresponding acquisition time frame by combining the number of dust points and the corrected discrete value in each image region includes: Calculating the product value of the number of dust points and the corrected discrete value in the image region, and performing normalization processing as the possibility index.

[0010] Further, the performing early warning judgment according to the numerical value of the possibility index in adjacent acquisition time frames includes: Determining the image regions where the possibility indexes of all image regions in the same acquisition frame are greater than the preset possibility threshold as the early warning regions; Determine whether to issue an early warning based on the area change of the warning area within the current and the previous acquisition time frames, and the mean value of the possibility indicators of all warning areas within the current acquisition time frame.

[0011] Further, the determining whether to issue an early warning based on the area change of the warning area within the current and the previous acquisition time frames, and the mean value of the possibility indicators of all warning areas within the current acquisition time frame includes: Calculate the area difference of the warning area between the current acquisition time frame and the previous one, and perform normalization processing as the area influence index; Normalize the product of the mean value of the possibility indicators of all warning areas and the area influence index as the warning judgment value; When the warning judgment value is greater than the preset warning threshold, determine to perform early warning processing; otherwise, determine not to perform early warning processing.

[0012] On the other hand, an intelligent early warning system for coal mine safety accidents is also provided. The system includes a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, the steps of the method described in any one of the foregoing are implemented.

[0013] The present invention has the following beneficial effects: In the embodiment of the present invention, by acquiring the coal mine underground monitoring images on each acquisition time frame; and according to the high-frequency characteristics of the dust, analyzing the high-frequency information and low-frequency information to obtain the dust points; then, combining the distribution dispersion index of the dust in different blocks, as well as the quality characteristics and lifting characteristics of the dust itself, performing the distribution analysis of the dust to obtain the corrected dispersion value. The corrected dispersion value can effectively integrate the quality characteristics and the dust lifting characteristics, improving the reliability of the dust aggregation analysis; after that, using the number of dust points as the density analysis feature, combining with the corrected dispersion value, determining the possibility index of the occurrence of dust hazards, so that the analysis of the possibility index can effectively combine multi-dimensional data characteristics, avoiding inaccurate results caused by only analyzing the gray-scale characteristics. Through the numerical change of the possibility index of adjacent acquisition time frames, performing early warning judgment, combining the change characteristics of adjacent acquisition time frames, so that the early warning judgment combines dynamic change analysis, improving the reliability of the overall intelligent hidden danger investigation. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings 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.

[0015] Figure 1 The flowchart of an intelligent early warning method for coal mine safety accidents provided by an embodiment of the present invention. Specific embodiments

[0016] In order to further elaborate on the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following combines the drawings and preferred embodiments to detail the specific embodiments, structures, features and effects of an intelligent early warning 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. In addition, the specific features, structures or characteristics in one or more embodiments can be combined in any suitable form.

[0017] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which the present invention belongs.

[0018] The following specifically describes the specific solution of an intelligent early warning method for coal mine safety accidents provided by the present invention with reference to the drawings.

[0019] Please refer to Figure 1 , which shows the flowchart of an intelligent early warning method for coal mine safety accidents provided by an embodiment of the present invention. The method includes: S101: Obtain the monitoring images of the coal mine underground at each acquisition time frame; based on Gaussian filtering, obtain the high-frequency information and low-frequency information of the overall monitoring images of the coal mine underground, and extract the dust points belonging to the high-frequency features.

[0020] Since mining operations are carried out inside the mine, dust is often produced. The dust hazard in the mine is an important factor. Dust is likely to cause respiratory diseases. When the dust prevention management is not in place and the dust concentration in the space reaches a certain level, if the equipment is aging or a malfunction occurs and a spark is generated, a flash explosion is extremely likely to occur, causing casualties and property losses. Therefore, it is necessary to effectively identify the dust in the mine and conduct intelligent hidden danger investigation.

[0021] In the related art, by analyzing the gray difference between the collected image and the standard image, the influence of dust is analyzed for early warning. In this way, since the dust concentration is inconsistent at different heights and positions, and the reliability of analyzing dust only based on the gray change of the image is poor, further optimization analysis is required.

[0022] In the embodiments of the present invention, high-definition cameras are installed in the mine to capture images of different time frames. Specifically, the monitoring images of the coal mine underground can be obtained at a speed of 1 frame per second or 1 frame per 10 seconds for periodic analysis. Then, the time frame of each captured monitoring image of the coal mine underground can be referred to as the acquisition time frame. To improve the comparison accuracy, it is necessary to ensure that during the entire shooting process, the position of the camera and the elements in the object to be photographed do not change, and it is also necessary to ensure that the brightness and darkness of the captured images remain consistent.

[0023] In the embodiments of the present invention, due to the strong regularity of the dust distribution in the mine tunnel, specifically, in a long and narrow tunnel, dust may accumulate in local areas, resulting in an increase in dust concentration, while a large-space mine tunnel helps to dilute the dust distribution; smaller dust particles will suspend in the air for a longer time, and larger particles are more likely to settle and usually concentrate in places closer to the source; by analyzing the number of dust corresponding pixel points and the relationship between pixel points in some local areas of the entire image, the specific situation of the dust distribution in this area can be obtained.

[0024] To achieve an ideal effect, it is necessary to first separate the dust part from the mine tunnel part, that is, to obtain the specific positions of the dust, which are called dust positions in the embodiments of the present invention. Since dust appears as fine particles in the image and shows high-frequency characteristics, in the embodiments of the present invention, the dust positions are determined through high-frequency feature analysis.

[0025] Further, in some embodiments of the present invention, based on Gaussian filtering, the high-frequency information and low-frequency information of the monitoring images of the coal mine underground are obtained, and the dust positions belonging to the high-frequency features are extracted, including: performing smoothing processing on the monitoring images of the coal mine underground based on Gaussian filtering as the low-frequency image; calculating the ratio of the brightness values of the monitoring images of the coal mine underground and the low-frequency image at the same position as the high-frequency feature index at the corresponding position; and determining the dust positions according to the numerical values of the high-frequency feature index.

[0026] Among them, since Gaussian filtering blurs the image through a smoothing operation in the spatial domain, reducing the fast-changing details and thus retaining the general structure of the image, the low-frequency information in the image is obtained by using Gaussian filtering. Performing smoothing processing on the monitoring images of the coal mine underground based on Gaussian filtering as the low-frequency image, and since the dust features are high-frequency information, in the embodiments of the present invention, the ratio of the brightness values of the monitoring images of the coal mine underground and the low-frequency image at the same position can be directly calculated as the high-frequency feature index at the corresponding position.

[0027] Since the larger the value of the high-frequency feature index, the greater the difference between the normal brightness and the low-frequency information brightness at the corresponding point, and it also shows that the high-frequency gray-scale feature is more obvious, which can effectively distinguish white noise in the image, the greater the possibility that it is a dust point. Therefore, in the embodiments of the present invention, the point where the high-frequency feature index is greater than the preset index threshold is used as the dust point under the high-frequency feature.

[0028] Among them, the preset index threshold is the threshold value of the high-frequency feature index. In the embodiments of the present invention, the preset index threshold can be specifically set to 0.8, that is, when the high-frequency feature index is greater than 0.8, the corresponding point is used as the dust point, and when the high-frequency feature index is less than or equal to 0.8, the corresponding point is used as the background point.

[0029] S102: Perform gray-scale processing and average division on each coal mine underground monitoring image to obtain different image regions, and determine the gray-scale dispersion index of each image region according to the numerical distribution discreteness of the gray-scale values of all pixel points in each image region.

[0030] The dust inside the mine is generally composed of tiny particles generated by ore crushing, equipment operation, etc.; usually very fine particulate matter, which appears as floating and dispersed phenomena in the image, so the distribution of dust inside the mine tunnel is irregular, presenting as blurred dots or smoke-like. And the dust is distributed irregularly in different blocks. Therefore, specific dust analysis can be carried out according to the dispersion index of dust distribution.

[0031] Since the distribution positions and states of dust with different diameters in the mine are different, in order to analyze different blocks, the coal mine underground monitoring image can be divided first. In the embodiments of the present invention, the coal mine underground monitoring image can be specifically divided into different image regions. In the embodiments of the present invention, the coal mine underground monitoring image can be directly divided into a fixed number of image regions, or the size of the preset image region can be set, and divided according to the size. For example, the coal mine underground monitoring image is segmented according to the pixel size of 25×25. It should be noted that when any subsequent column or row is less than 25 pixel points during segmentation, pixel points with a gray-scale value of 255 need to be supplemented. This is because this embodiment is for coal mine safety analysis, and the gray-scale value of coal is close to the gray-scale value of 0, so the operation of supplementing pixel points with a gray-scale value of 255 is selected for subsequent analysis.

[0032] Further, in some embodiments of the present invention, determining the gray-scale dispersion index of each image region according to the numerical distribution discreteness of the gray-scale values of all pixel points in each image region includes: calculating the standard deviation of the gray-scale values of all pixel points in the same image region as the gray-scale dispersion index.

[0033] In the monitoring images of coal mines underground, due to the different diameters of dust particles, the sedimentation degrees of dust with different diameters are also different, which also means that the aggregation information of dust at different positions in the same image is also different. By the details and features of the local area of the monitoring images of coal mines underground, the aggregation situation of local dust can be revealed. In the embodiments of the present invention, the standard deviation is used for discrete analysis. The larger the value of the standard deviation, the poorer the aggregation within the same image area, that is, the larger the value of the gray-scale discrete index, the poorer the dust aggregation effect in the image area.

[0034] S103: Analyze the dust quality, and determine the quality correction index of each image area according to any height and the gray-scale discrete index of the image area below it; correct the gray-scale discrete index in combination with the quality correction index to obtain the corrected discrete value.

[0035] It should be noted that the gray-scale discrete index of each area calculated using the above logic is obtained without considering the dust quality. Because in reality, the distributions of dust with different qualities in space are different. The larger the quality of the dust, the closer it is to the lower part in the same scene, and the smaller the quality of the dust, the easier it is to float in the air. Therefore, it is necessary to correct the aggregation information (that is, the gray-scale discrete index) obtained above.

[0036] Further, in some embodiments of the present invention, determining the quality correction index of each image area according to any height and the gray-scale discrete index of the image area below it includes: taking the height of any point as the target height, and taking the heights of all points below the target height as the candidate heights; calculating the sum value of the gray-scale discrete indexes at the target height and all candidate heights, and normalizing it as the quality correction index.

[0037] It should be noted that in the actual scene, the sedimentation speeds corresponding to dust particles with different quality sizes are also different. The larger the quality, the faster the sinking speed. So in the image, it is shown that the aggregation of dust in the area closer to the lower part of the image is better. As time changes, the possibility of the dust in the lower area adhering to the ground is higher. Then the proportion of disasters caused by the dust in this part is much less than that of the dust in the upper part of the image, that is, the probabilities of dust disasters caused by the lower part and the upper part will change.

[0038] In an actual scenario, due to various conditions such as wind force, human activities, and machine operation, dust will be raised below. Considering this aspect, in the embodiments of the present invention, the gray-scale discrete indexes of the target height and all candidate heights below it are directly accumulated. When approaching the lower part, since the number of accumulation parameters is small, the proportion of disasters caused by dust below cannot be represented as well as the dust above. When the dust below is more discretely distributed and more likely to be raised, the accumulated gray-scale discrete index value is larger, which also represents the raising feature. Therefore, the quality correction index in the embodiments of the present invention can accurately represent the quality feature and the raising feature, and has higher reliability.

[0039] Combining the quality correction index to correct the gray-scale discrete index to obtain a corrected discrete value, including: calculating the product value of the quality correction index and the gray-scale discrete index of the image area as the corrected discrete value.

[0040] In the embodiments of the present invention, the gray-scale discrete index is further corrected by the quality correction index, so that the corrected discrete value can fuse the quality feature and the dust raising feature, and improve the reliability of dust aggregation analysis.

[0041] S104: Combining the number of dust points belonging to each image area and the corrected discrete value, determining the possibility index of dust hazard occurring in the corresponding acquisition time frame, and making a warning judgment according to the numerical values of the possibility indexes of adjacent acquisition time frames.

[0042] Among them, due to the diffusion effect of dust, in the related art, only the dust situation at a single moment is analyzed without considering its changes. Therefore, in the embodiments of the present invention, specific analysis can be carried out in combination with the changes of different acquisition time frames. First, it is necessary to analyze and obtain the possibility index.

[0043] Furthermore, in some embodiments of the present invention, combining the number of dust points belonging to each image area and the corrected discrete value, determining the possibility index of dust hazard occurring in the corresponding acquisition time frame, including: calculating the product value of the number of dust points belonging to the image area and the corrected discrete value, and normalizing it as the possibility index.

[0044] Among them, since the number of dust points objectively represents the dust concentration, the more the number of dust points in the area of the same area, the greater the dust concentration, and the corrected discrete value represents the state of the dust. The larger the value of the corrected discrete value, the more likely the dust distribution state is to cause an explosion hazard. Therefore, directly calculate the product value of the number of dust points belonging to the image area and the corrected discrete value, and normalize it to obtain the possibility index of dust hazard occurring.

[0045] Further, in some embodiments of the present invention, early warning judgment is performed according to the numerical values of the adjacent acquisition time frame possibility indicators, including: determining the image regions in the same acquisition frame where the possibility indicators of all image regions are greater than a preset possibility threshold as the early warning regions; determining whether to issue an early warning according to the area change of the early warning regions in the current and the previous acquisition time frames, and the average value of the possibility indicators of all early warning regions in the current acquisition time frame.

[0046] Among them, the preset possibility threshold is the threshold value of the possibility indicator. The preset possibility threshold in the embodiments of the present invention can be specifically, for example, 0.75. That is, when the possibility indicator is greater than 0.75, the corresponding image region is used as the early warning region. The early warning region represents a region with a relatively dangerous dust state.

[0047] Determining whether to issue an early warning according to the area change of the early warning regions in the current and the previous acquisition time frames, and the average value of the possibility indicators of all early warning regions in the current acquisition time frame, includes: calculating the area difference of the early warning regions between the current acquisition time frame and the previous acquisition time frame, and normalizing it as the area influence indicator; normalizing the product of the average value of the possibility indicators of all early warning regions and the area influence indicator as the early warning judgment value; when the early warning judgment value is greater than the preset early warning threshold, determining to perform early warning processing, otherwise, determining not to perform early warning processing.

[0048] In the embodiments of the present invention, while directly analyzing the average value of the possibility indicators of the early warning regions, the area change feature is introduced. When the area gradually decreases, it indicates that the dust concentration is decreasing, and when the area gradually increases, it indicates that the dust concentration is rising, which needs to be paid special attention to. Therefore, the product of the average value of the possibility indicators of all early warning regions and the area influence indicator is used as the early warning judgment value.

[0049] Among them, the preset early warning threshold is the threshold value of the early warning judgment value. It can be specifically, for example, 0.8, and it is analyzed according to the actual scenario, and there is no limitation in this regard. When the early warning judgment value is greater than 0.8, it is determined to perform early warning processing, otherwise, it is determined not to perform early warning processing. The early warning processing in the embodiments of the present invention can be specifically, for example, humidifying the air to make the dust settle, and the buzzer reminding relevant inspection personnel, etc., and there is no limitation in this regard.

[0050] In an embodiment of the present invention, coal mine underground monitoring images are obtained for each acquisition time frame; and based on the high-frequency characteristics of dust, high-frequency information and low-frequency information are analyzed to obtain dust points; then, in combination with the distribution dispersion index of dust in different blocks, as well as the mass characteristics and lifting characteristics of the dust itself, the distribution of dust is analyzed to obtain a corrected dispersion value, which can effectively integrate the mass characteristics and the dust lifting characteristics and improve the reliability of dust aggregation analysis; afterwards, using the number of dust points as the density analysis feature and combining with the corrected dispersion value, a possibility index of dust hazard occurrence is determined, so that the analysis of the possibility index can effectively combine multi-dimensional data characteristics and avoid inaccurate results caused by only analyzing the gray-scale characteristics. Through the numerical change of the possibility index of adjacent acquisition time frames, early warning judgment is carried out, which combines the change characteristics of adjacent acquisition time frames, thereby making the early warning judgment combine dynamic change analysis and improving the reliability of the overall intelligent hidden danger investigation.

[0051] The present invention also provides an intelligent early warning system for coal mine safety accidents. The system includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of an intelligent early warning method for coal mine safety accidents as described above are implemented.

[0052] It should be noted that the above sequence of embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-tasking and parallel processing are also possible or may be advantageous.

[0053] Each embodiment in this specification is described in a progressive manner. The same or similar parts among the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments.

Claims

1. An intelligent early warning method for coal mine safety accidents, characterized in that, The method includes: Obtaining the monitoring images of the coal mine underground for each acquisition time frame; based on Gaussian filtering, obtaining the high-frequency information and low-frequency information of the overall monitoring images of the coal mine underground, and extracting the dust points belonging to the high-frequency features; Performing grayscale processing on each monitoring image of the coal mine underground and evenly dividing it to obtain different image regions, and determining the grayscale dispersion index of each image region according to the numerical distribution dispersion of the grayscale values of all pixel points within each image region; Analyzing the dust quality, and determining the quality correction index of each image region according to any height and the grayscale dispersion index of the image region below it; combining the quality correction index to correct the grayscale dispersion index to obtain a corrected dispersion value; Combining the number of dust points belonging to each image region and the corrected dispersion value, determining the possibility index of dust hazard occurring under the corresponding acquisition time frame, and making a warning judgment according to the numerical values of the possibility index of adjacent acquisition time frames.

2. The intelligent early warning method for coal mine safety accidents according to claim 1, wherein The obtaining the high-frequency information and low-frequency information of the overall monitoring images of the coal mine underground based on Gaussian filtering, and extracting the dust points belonging to the high-frequency features includes: Performing smoothing processing on the monitoring images of the coal mine underground based on Gaussian filtering as the low-frequency image; Calculating the ratio of the brightness values of the monitoring images of the coal mine underground and the low-frequency image at the same point as the high-frequency feature index at the corresponding point; Determining the dust points according to the numerical value of the high-frequency feature index.

3. The intelligent early warning method for coal mine safety accidents according to claim 2, wherein, The determining the dust points according to the numerical value of the high-frequency feature index includes: Taking the points where the high-frequency feature index is greater than the preset index threshold as the dust points under the high-frequency features.

4. An intelligent early warning method for coal mine safety accidents according to claim 1, characterized in that, The determining the grayscale dispersion index of each image region according to the numerical distribution dispersion of the grayscale values of all pixel points within each image region includes: Calculating the standard deviation of the grayscale values of all pixel points within the same image region as the grayscale dispersion index.

5. The intelligent early warning method for coal mine safety accidents according to claim 1, wherein, The determining the quality correction index of each image region according to any height and the grayscale dispersion index of the image region below it includes: Taking any point height as the target height, and taking all point heights below the target height as the candidate heights; Calculating the sum value of the grayscale dispersion indexes at the target height and all candidate heights, and performing normalization processing as the quality correction index.

6. The intelligent early warning method for coal mine safety accidents according to claim 1, characterized in that, The combining the quality correction index to correct the grayscale dispersion index to obtain a corrected dispersion value includes: Calculating the product value of the quality correction index and the grayscale dispersion index of the image region as the corrected dispersion value.

7. An intelligent early warning method for coal mine safety accidents according to claim 1, characterized in that, The combining the number of dust points belonging to each image region and the corrected dispersion value, and determining the possibility index of dust hazard occurring under the corresponding acquisition time frame includes: Calculating the product value of the number of dust points belonging to the image region and the corrected dispersion value, and performing normalization processing as the possibility index.

8. The intelligent early warning method for coal mine safety accidents according to claim 1, characterized in that, The making a warning judgment according to the numerical values of the possibility index of adjacent acquisition time frames includes: Determining the image regions where the possibility indexes of all image regions within the same acquisition frame are greater than the preset possibility threshold as the warning regions; Determining whether to give a warning according to the area change of the warning regions within the current and the previous acquisition time frames, and the mean value of the possibility indexes of all warning regions within the current acquisition time frame.

9. The intelligent early warning method for coal mine safety accidents according to claim 8, wherein Determining whether to issue an early warning according to the area change of the early warning area in the current and previous acquisition time frames and the mean value of the possibility indicators of all early warning areas in the current acquisition time frame, includes: Calculating the area difference of the early warning area between the current acquisition time frame and the previous acquisition time frame, and performing normalization processing as the area influence index; Multiplying the mean value of the possibility indicators of all early warning areas by the area influence index, and performing normalization processing as the early warning judgment value; When the early warning judgment value is greater than the preset early warning threshold, determining to perform early warning processing, otherwise, determining not to perform early warning processing.

10. An intelligent early warning system for coal mine safety accidents, the system comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, the steps of the method according to any one of claims 1 to 9 are implemented.

Images