Method and system for identifying underground vibration of coal mine

Abstract

The invention relates to a coal mine underground vibration identification method and system. The method comprises the following steps: acquiring various vibration data of each device and an appearance image of the device in real time; the working duration, the load state and the historical operation parameters of each device are recorded and stored, so that the weight of each device is calculated; extracting vibration features according to the vibration data, and multiplying all the vibration features of each device by the weight of the corresponding device; identifying the feature point coordinate change of the equipment as an image feature by using the collected appearance image through a scale invariant feature transformation algorithm; acquiring historical vibration position data, and dividing an underground coal mine area into a plurality of areas with different vibration risk levels through a geographic information system (GIS); and fusing and splicing the weighted vibration features and image features of all devices to form a regional feature vector, and inputting the regional feature vector into a convolutional neural network for prediction. According to the invention, real-time monitoring and accurate positioning of underground vibration of the coal mine are realized, and the system adapts to complex and changeable vibration environments.

Description

Technical Field

[0001] The present invention belongs to the technical field of vibration identification, and more specifically, relates to a method and system for identifying vibration in underground coal mines. Background Art

[0002] Coal mine production safety is a key concern of the industry, among which underground vibration monitoring is crucial to prevent equipment damage and safety accidents. Traditional vibration monitoring methods that rely on a single sensor have many limitations, such as incomplete information, inaccurate positioning, and limited vibration type recognition capabilities. With the advancement of science and technology, the application of image recognition technology and GIS technology has provided new perspectives and possibilities for coal mine vibration monitoring, but there is currently a lack of an effective method to combine these two technologies to achieve comprehensive and accurate identification of underground coal mine vibration. Summary of the invention

[0003] In order to solve the deficiencies in the prior art, the present invention provides a method and system for identifying vibration in underground coal mines.

[0004] The present invention adopts the following technical solution.

[0005] The first aspect of the present invention provides a method for identifying vibration in a coal mine, which is characterized by comprising: By configuring a status sensor on each device, various vibration data of each device can be obtained in real time, and the appearance image of the device can be collected in real time; the vibration data and the appearance image of the device can be pre-processed; Establish an operating parameter recording unit, which is connected to the status sensor to record and store the working hours, load status and historical operating parameters of each device; Calculate the weight of each device based on working hours, load status and historical operating parameters; Extract vibration features based on vibration data, multiply all vibration features of each device by the weight of the corresponding device; extract feature points of the device from the collected appearance image through a scale-invariant feature transformation algorithm, identify changes in the coordinates of the feature points of the device, and output all changes in the coordinates of the feature points in the form of vectors as image features; Obtain historical vibration location data, and use the geographic information system (GIS) to combine the vibration data and vibration location data to divide the underground area of ​​the coal mine into multiple areas with different vibration risk levels; According to the divided areas, the weighted vibration features of all equipment in each vibration risk level area are fused and concatenated with the image features to form a regional feature vector. The regional feature vector is input into the trained convolutional neural network to predict the probability, type and intensity of vibration in the area.

[0006] Preferably, the state sensor is configured on each device to obtain the vibration data of each device in real time, specifically: Equipment includes coal mining machines, tunneling machines and conveyors according to equipment type; vibration data includes vibration frequency, amplitude, vibration mode, and operating acceleration changes.

[0007] Preferably, the preprocessing of the vibration data and the appearance image of the device is specifically as follows: The image is converted into a grayscale image, and then filtered to eliminate noise. The filtering formula is:

[0008]

[0009] in, k is the radius of the Gaussian kernel, is the pixel coordinate in the image, is the pixel coordinate The result after pixel filtering is is the weight of the Gaussian kernel, are the element coordinates of the Gaussian kernel, The pixel coordinates in the original image are Pixels of is a two-dimensional Gaussian function.

[0010] Preferably, the weight of each device is calculated by working time, load status and historical operating parameters, specifically:

[0011] in, is the credibility weight of the device's working time, is the credibility weight of the load status of the equipment, is the credibility weight of the i-th historical operating parameter of the equipment; n is the number of historical operating parameters;

[0012]

[0013]

[0014] in, , , They are respectively the set working hours, load status and historical operating parameter weight coefficients; , , They are the thresholds of the set working hours, load status and historical operating parameters; The theoretical maximum working time; is the maximum load value; It is the maximum value of historical operating parameters; is the average value of historical operating parameters; , , are the working time, load status and the i-th historical operating parameter of the equipment respectively; It is a step function, which is 1 when its independent variable is greater than or equal to 0, and 0 otherwise.

[0015] Preferably, the various vibration characteristics include vibration waveform characteristics, vibration spectrum characteristics, and vibration energy distribution characteristics; The vibration waveform features are extracted as follows: All vibration amplitudes are arranged into a vibration amplitude sequence in time order, and the curve function of the sequence is simulated is the vibration waveform, t For time, the symmetry deviation of the vibration waveform and the derivatives of each set time point are calculated as vibration waveform features; Symmetry Deviation The calculation formula is:

[0016] in, is the total number of time points set, The time of the I-th set time point; Preferably, the extraction of vibration waveform characteristics and vibration energy distribution characteristics is specifically as follows: All vibration frequencies are arranged into a vibration frequency sequence in chronological order, and the vibration frequency sequence is converted into a spectrum diagram using the fast Fourier transform FFT algorithm. , f is the frequency, and the peak of the spectrum is the vibration spectrum feature; The vibration energy distribution characteristic is the vibration energy distribution within the set frequency range, and the calculation formula is:

[0017] in, is the frequency corresponding to the peak of the spectrum graph, f 1- f 2 is the set frequency range; is the set frequency value.

[0018] Preferably, the historical vibration position data is obtained, and the vibration position data and the geographic information system GIS are used in combination with the vibration data to divide the underground area of ​​the coal mine into multiple areas with different vibration risk levels, specifically: According to the actual terrain near each device, select a space with the device as the center and a set distance as the radius, or a rectangular space with the device as the center point and set length and width, divide all the selected spaces into regions, and merge other areas into a risk-free area; Based on the historical vibration position data, the vibration density of all unit areas within the object is calculated using the formula:

[0019] in, is the vibration density of the vth unit area; m is the total number of historical vibration position data; h is the set smoothing parameter; K() is the Gaussian kernel function; is the position of the vth unit area; is the u-th historical vibration position data; and All are in the form of two-dimensional coordinates; is the Euclidean norm; Set different vibration modes as digital indicators ranging from 0 to 1; calculate the comprehensive vibration index based on the vibration density per unit area, the vibration mode of the equipment contained in the unit area, and the change in running acceleration. , the formula is:

[0020] in, , , They are the weights of vibration density, vibration mode, and running acceleration changes; is the average value of the vibration mode of the device in the vth unit area, is the average value of the running acceleration change of the equipment in the vth unit area; The comprehensive vibration indicators of all unit area regions are input into the GIS tool, and the natural breakpoint algorithm of GIS is used to divide all the comprehensive vibration indicators into a set number of vibration risk levels; and the unit area regions with the same vibration risk level are integrated into one area through the GIS tool and visualized.

[0021] Preferably, the weights of vibration density, vibration mode, and running acceleration change are calculated as follows: The vibration density, vibration mode and running acceleration changes are all normalized; Calculate the entropy value and weight of each indicator:

[0022] in, a ={1,2,3}, where 1, 2, and 3 represent the vibration density, vibration mode, and running acceleration changes respectively; for a The corresponding vibration density, vibration mode setting or entropy value of running acceleration change, is the total number of unit area regions in the partitioned object; exist a When it is 1, 2, or 3, it means after normalization. , , ; exist a When it is 1, 2, or 3, it means , , .

[0023] Preferably, according to the divided areas, the weighted vibration features of all devices in each vibration risk level area are fused and spliced ​​with the image features to form a regional feature vector, and the regional feature vector is input into a trained convolutional neural network to predict the probability, type and intensity of vibration in the area, specifically: For each weighted vibration feature and image feature of all equipment in the area of ​​each vibration risk level, the average value of the feature is calculated and then concatenated to form a regional feature vector; It is input into the trained convolutional neural network and outputs a three-dimensional vector. The three-dimensional vector represents the probability of vibration in the area, the probability of the vibration in the area being of each vibration type, and the probability of the vibration intensity in the area being of various intensities. The probability of vibration is 0-1. The various vibration types include mechanical vibration caused by loose equipment parts, geological vibration caused by changes in geological structure, and vibration caused by normal equipment operation. The various intensities include weak, medium, and strong.

[0024] The second aspect of the present invention provides a coal mine underground vibration recognition system using the method described in the first aspect of the present invention, comprising a configuration state sensor, a camera device, an operation parameter recording unit, a weight calculation module, a feature extraction module, a region division module and a prediction module, characterized in that: Configuration status sensor: used to obtain various vibration data of each device in real time; Camera device: used to collect the appearance image of the equipment in real time; pre-process the vibration data and the appearance image of the equipment; Operation parameter recording unit: This unit is connected to the status sensor and is used to record and store the working hours, load status and historical operation parameters of each device; Weight calculation module: used to calculate the weight of each device based on working hours, load status and historical operating parameters; Feature extraction module: extract vibration features based on vibration data, multiply all vibration features of each device by the weight of the corresponding device; extract feature points of the device from the collected appearance image through the scale-invariant feature transformation algorithm, identify the coordinate changes of the feature points of the device, and output all feature point coordinate changes in the form of vectors as image features; Area division module: obtain historical vibration location data, and use the geographic information system GIS to divide the underground area of ​​the coal mine into multiple areas with different vibration risk levels by combining the vibration data and vibration location data; Prediction module: According to the divided areas, the weighted vibration features of all devices in each vibration risk level area are fused and concatenated with the image features to form a regional feature vector. The regional feature vector is input into the trained convolutional neural network to predict the probability, type and intensity of vibration in the area.

[0025] The beneficial effects of the present invention are that, compared with the prior art, by extracting the multi-dimensional features of vibration data and combining the features extracted from pictures, the accuracy and reliability of vibration recognition are improved, and false alarms and missed alarms are reduced; weights are calculated by using working hours, load status and historical parameters, and the prediction of equipment vibration under different states can be dynamically adapted; GIS regional division is performed by using vibration data and vibration location data, and the physical location and vibration data can be combined, and the regional division is more scientific, and regions with different risk levels can be predicted and managed in a targeted manner. Finally, real-time monitoring and precise positioning of underground vibration in coal mines are achieved through convolutional neural networks, adapting to complex and changeable vibration environments, improving emergency response efficiency, and providing data support for safe production in coal mines. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 The figure is a flow chart of the method of the present invention. DETAILED DESCRIPTION

[0027] In order to make the purpose, technical scheme and advantages of the present invention clearer, the technical scheme of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. The embodiments described in this application are only embodiments of a part of the present invention, rather than all embodiments. Based on the spirit of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work belong to the protection scope of the present invention.

[0028] like Figure 1 As shown, embodiment 1 of the present invention proposes a method for identifying vibration in a coal mine, which is characterized by comprising: By configuring a status sensor on each device, various vibration data of each device can be obtained in real time, and the appearance image of the device can be collected in real time; the vibration data and the appearance image of the device can be pre-processed; Establish an operating parameter recording unit, which is connected to the status sensor to record and store the working hours, load status and historical operating parameters of each device; Calculate the weight of each device based on working hours, load status and historical operating parameters; Extract vibration features based on vibration data, multiply all vibration features of each device by the weight of the corresponding device; extract feature points of the device from the collected appearance image through a scale-invariant feature transformation algorithm, identify changes in the coordinates of the feature points of the device, and output all changes in the coordinates of the feature points in the form of vectors as image features; Obtain historical vibration location data, and use the geographic information system (GIS) to combine the vibration data and vibration location data to divide the underground area of ​​the coal mine into multiple areas with different vibration risk levels; According to the divided areas, the weighted vibration features of all equipment in each vibration risk level area are fused and concatenated with the image features to form a regional feature vector. The regional feature vector is input into the trained convolutional neural network to predict the probability, type and intensity of vibration in the area.

[0029] Preferably, the state sensor is configured on each device to obtain the vibration data of each device in real time, specifically: Equipment includes coal mining machines, tunneling machines and conveyors according to equipment type; vibration data includes vibration frequency, amplitude, vibration mode, and operating acceleration changes.

[0030] Preferably, the preprocessing of the vibration data and the appearance image of the device is specifically as follows: The vibration data is standardized using the formula:

[0031] in, is the original vibration data, is the mean value of the vibration data, is the standard deviation of the vibration data, is the standardized vibration data; The image is converted into a grayscale image, and then filtered to eliminate noise. The filtering formula is:

[0032]

[0033] in, k is the radius of the Gaussian kernel, is the pixel coordinate in the image, is the pixel coordinate The result after pixel filtering is is the weight of the Gaussian kernel, are the element coordinates of the Gaussian kernel, The pixel coordinates in the original image are Pixels of is a two-dimensional Gaussian function.

[0034] Preferably, the weight of each device is calculated by working time, load status and historical operating parameters, specifically:

[0035] in, is the credibility weight of the device's working time, is the credibility weight of the load status of the equipment, is the credibility weight of the i-th historical operating parameter of the equipment; n is the number of historical operating parameters;

[0036]

[0037]

[0038] in, , , They are respectively the set working hours, load status and historical operating parameter weight coefficients; , , They are the thresholds of the set working hours, load status and historical operating parameters; The theoretical maximum working time; is the maximum load value; It is the maximum value of historical operating parameters; is the average value of historical operating parameters; , , are the working time, load status and the i-th historical operating parameter of the equipment respectively; It is a step function, which is 1 when its independent variable is greater than or equal to 0, and 0 otherwise.

[0039] Preferably, the various vibration characteristics include vibration waveform characteristics, vibration spectrum characteristics, and vibration energy distribution characteristics; The vibration waveform features are extracted as follows: All vibration amplitudes are arranged into a vibration amplitude sequence in time order, and the curve function of the sequence is simulated is the vibration waveform, t For time, the symmetry deviation of the vibration waveform and the derivatives of each set time point are calculated as vibration waveform features; Symmetry Deviation The calculation formula is:

[0040] in, is the total number of time points set, The time of the I-th set time point; Preferably, the extraction of vibration waveform characteristics and vibration energy distribution characteristics is specifically as follows: All vibration frequencies are arranged into a vibration frequency sequence in chronological order, and the vibration frequency sequence is converted into a spectrum diagram using the fast Fourier transform FFT algorithm. , f is the frequency, and the peak of the spectrum is the vibration spectrum feature; The vibration energy distribution characteristic is the vibration energy distribution within the set frequency range, and the calculation formula is:

[0041] in, is the frequency corresponding to the peak of the spectrum graph, f 1- f 2 is the set frequency range; is the set frequency value.

[0042] Preferably, the historical vibration position data is obtained, and the vibration position data and the geographic information system GIS are used in combination with the vibration data to divide the underground area of ​​the coal mine into multiple areas with different vibration risk levels, specifically: According to the actual terrain near each device, select a space with the device as the center and a set distance as the radius, or a rectangular space with the device as the center point and set length and width, divide all the selected spaces into regions, and merge other areas into a risk-free area; Specifically, the space selected in this embodiment is a space with the device as the center and the set distance as the radius, and the set distance is 5m.

[0043] Based on the historical vibration position data, the vibration density of all unit areas within the object is calculated using the formula:

[0044] in, is the vibration density of the vth unit area; m is the total number of historical vibration position data; h is the set smoothing parameter; K() is the Gaussian kernel function; is the position of the vth unit area; is the u-th historical vibration position data; and All are in the form of two-dimensional coordinates; is the Euclidean norm; Set different vibration modes as digital indicators ranging from 0 to 1; calculate the comprehensive vibration index based on the vibration density per unit area, the vibration mode of the equipment contained in the unit area, and the change in running acceleration. , the formula is:

[0045] in, , , They are the weights of vibration density, vibration mode, and running acceleration changes; is the average value of the vibration mode of the device in the vth unit area, is the average value of the running acceleration change of the equipment in the vth unit area; The comprehensive vibration indicators of all unit area regions are input into the GIS tool, and the natural breakpoint algorithm of GIS is used to divide all the comprehensive vibration indicators into a set number of vibration risk levels; and the unit area regions with the same vibration risk level are integrated into one area through the GIS tool and visualized.

[0046] Preferably, the weights of vibration density, vibration mode, and running acceleration change are calculated as follows: The vibration density, vibration mode and running acceleration changes are all normalized; Calculate the entropy value and weight of each indicator:

[0047] in, a ={1,2,3}, where 1, 2, and 3 represent the vibration density, vibration mode, and running acceleration changes respectively; for a The corresponding vibration density, vibration mode setting or entropy value of running acceleration change, is the total number of unit area regions in the partitioned object; exist a When it is 1, 2, or 3, it means after normalization. , , ; exist a When it is 1, 2, or 3, it means , , .

[0048] Preferably, according to the divided areas, the weighted vibration features of all devices in each vibration risk level area are fused and spliced ​​with the image features to form a regional feature vector, and the regional feature vector is input into a trained convolutional neural network to predict the probability, type and intensity of vibration in the area, specifically: For each weighted vibration feature and image feature of all equipment in the area of ​​each vibration risk level, the average value of the feature is calculated and then concatenated to form a regional feature vector; It is input into the trained convolutional neural network and outputs a three-dimensional vector. The three-dimensional vector represents the probability of vibration in the area, the probability of the vibration in the area being of each vibration type, and the probability of the vibration intensity in the area being of various intensities. The probability of vibration is 0-1. The various vibration types include mechanical vibration caused by loose equipment parts, geological vibration caused by changes in geological structure, and vibration caused by normal equipment operation. The various intensities include weak, medium, and strong.

[0049] Embodiment 2 of the present invention proposes a system for identifying vibration in underground coal mines using the method described in Embodiment 1 of the present invention, comprising a configuration state sensor, a camera device, an operation parameter recording unit, a weight calculation module, a feature extraction module, a region division module and a prediction module, characterized in that: Configuration status sensor: used to obtain various vibration data of each device in real time; Camera device: used to collect the appearance image of the equipment in real time; pre-process the vibration data and the appearance image of the equipment; Operation parameter recording unit: This unit is connected to the status sensor and is used to record and store the working hours, load status and historical operation parameters of each device; Weight calculation module: used to calculate the weight of each device based on working hours, load status and historical operating parameters; Feature extraction module: extract vibration features based on vibration data, multiply all vibration features of each device by the weight of the corresponding device; extract feature points of the device from the collected appearance image through the scale-invariant feature transformation algorithm, identify the coordinate changes of the feature points of the device, and output all feature point coordinate changes in the form of vectors as image features; Area division module: obtain historical vibration location data, and use the geographic information system GIS to divide the underground area of ​​the coal mine into multiple areas with different vibration risk levels by combining the vibration data and vibration location data; Prediction module: According to the divided areas, the weighted vibration features of all devices in each vibration risk level area are fused and concatenated with the image features to form a regional feature vector. The regional feature vector is input into the trained convolutional neural network to predict the probability, type and intensity of vibration in the area.

[0050] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents, and any modifications or equivalent replacements that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for identifying vibration in coal mines, characterized in that: include: By configuring status sensors on each device, various vibration data of each device can be obtained in real time, and the appearance image of the device can be collected in real time; Preprocessing vibration data and appearance images of equipment; Establish an operating parameter recording unit, which is connected to the status sensor to record and store the working hours, load status and historical operating parameters of each device; Calculate the weight of each device based on working hours, load status and historical operating parameters; Extract vibration features based on vibration data, multiply all vibration features of each device by the weight of the corresponding device; extract feature points of the device from the collected appearance image through a scale-invariant feature transformation algorithm, identify changes in the coordinates of the feature points of the device, and output all changes in the coordinates of the feature points in the form of vectors as image features; Obtain historical vibration location data, and use the geographic information system (GIS) to combine the vibration data and vibration location data to divide the underground area of ​​the coal mine into multiple areas with different vibration risk levels; According to the divided areas, the weighted vibration features of all equipment in each vibration risk level area are fused and concatenated with the image features to form a regional feature vector. The regional feature vector is input into the trained convolutional neural network to predict the probability, type and intensity of vibration in the area.

2. A method for identifying vibration in underground coal mines according to claim 1, characterized in that: include: The state sensor is configured on each device to obtain the vibration data of each device in real time, specifically: Equipment includes coal mining machines, tunneling machines and conveyors according to equipment type; vibration data includes vibration frequency, amplitude, vibration mode, and operating acceleration changes.

3. A method for identifying vibration in coal mines according to claim 2, characterized in that: include: The vibration data and the appearance image of the device are preprocessed, specifically: The image is converted into a grayscale image, and then filtered to eliminate noise. The filtering formula is: in, k is the radius of the Gaussian kernel, is the pixel coordinate in the image, is the pixel coordinate The result after pixel filtering is is the weight of the Gaussian kernel, are the element coordinates of the Gaussian kernel, The pixel coordinates in the original image are Pixels of is a two-dimensional Gaussian function.

4. A method for identifying vibration in coal mines according to claim 1, characterized in that: include: The weight of each device is calculated by working time, load status and historical operating parameters, specifically: in, is the credibility weight of the working time of the equipment, is the credibility weight of the load status of the equipment, is the credibility weight of the i-th historical operating parameter of the equipment; n is the number of historical operating parameters; in, , , They are respectively the set working hours, load status and historical operating parameter weight coefficients; , , They are the thresholds of the set working hours, load status and historical operating parameters; The theoretical maximum working time; is the maximum load value; It is the maximum value of historical operating parameters; is the average value of historical operating parameters; , , are the working time, load status and the i-th historical operating parameter of the equipment respectively; It is a step function, which is 1 when its independent variable is greater than or equal to 0, and 0 otherwise.

5. A method for identifying vibration in underground coal mines according to claim 3, characterized in that: include: Various vibration characteristics include vibration waveform characteristics, vibration spectrum characteristics, and vibration energy distribution characteristics; The vibration waveform features are extracted as follows: All vibration amplitudes are arranged into a vibration amplitude sequence in time order, and the curve function of the sequence is simulated is the vibration waveform, t For time, the symmetry deviation of the vibration waveform and the derivatives of each set time point are calculated as vibration waveform features; Symmetry Deviation The calculation formula is: in, is the total number of time points set, The time of the I-th set time point.

6. A method for identifying vibration in underground coal mines according to claim 5, characterized in that: include: Extraction of vibration waveform characteristics and vibration energy distribution characteristics, specifically: All vibration frequencies are arranged into a vibration frequency sequence in chronological order, and the vibration frequency sequence is converted into a spectrum diagram using the fast Fourier transform FFT algorithm. , f is the frequency, and the peak of the spectrum is the vibration spectrum feature; The vibration energy distribution characteristic is the vibration energy distribution within the set frequency range, and the calculation formula is: in, is the frequency corresponding to the peak of the spectrum graph, f 1- f 2 is the set frequency range; is the set frequency value.

7. A method for identifying vibration in underground coal mines according to claim 3, characterized in that: include: The historical vibration location data is obtained, and the vibration location data and the geographic information system GIS are combined to divide the underground area of ​​the coal mine into multiple areas with different vibration risk levels, specifically: According to the actual terrain near each device, select a space with the device as the center and a set distance as the radius, or a rectangular space with the device as the center point and set length and width, divide all the selected spaces into regions, and merge other areas into a risk-free area; Based on the historical vibration position data, the vibration density of all unit areas within the object is calculated using the formula: in, is the vibration density of the vth unit area; m is the total number of historical vibration position data; h is the set smoothing parameter; K() is the Gaussian kernel function; is the position of the vth unit area; is the u-th historical vibration position data; and All are in the form of two-dimensional coordinates; is the Euclidean norm; Set different vibration modes as digital indicators ranging from 0 to 1; calculate the comprehensive vibration index based on the vibration density per unit area, the vibration mode of the equipment contained in the unit area, and the change in running acceleration. , the formula is: in, , , They are the weights of vibration density, vibration mode, and running acceleration changes; is the average value of the vibration mode of the device in the vth unit area, is the average value of the running acceleration change of the equipment in the vth unit area; The comprehensive vibration indicators of all unit area regions are input into the GIS tool, and the natural breakpoint algorithm of GIS is used to divide all the comprehensive vibration indicators into a set number of vibration risk levels; and the unit area regions with the same vibration risk level are integrated into one area through the GIS tool and visualized.

8. A method for identifying vibration in underground coal mines according to claim 6, characterized in that: include: Calculate the weights of vibration density, vibration mode, and running acceleration changes, specifically: The vibration density, vibration mode and running acceleration changes are all normalized; Calculate the entropy value and weight of each indicator: in, a ={1,2,3}, where 1, 2, and 3 represent the vibration density, vibration mode, and running acceleration changes respectively; for a The corresponding vibration density, vibration mode setting or entropy value of running acceleration change, is the total number of unit area regions in the partitioned object; exist a When it is 1, 2, or 3, it means after normalization. , , ; exist a When it is 1, 2, or 3, it means , , .

9. A method for identifying vibration in underground coal mines according to claim 6, characterized in that: include: According to the divided areas, the weighted vibration features of all devices in each vibration risk level area are fused and spliced ​​with the image features to form a regional feature vector, and the regional feature vector is input into the trained convolutional neural network to predict the probability, type and intensity of vibration in the area, specifically: For each weighted vibration feature and image feature of all equipment in the area of ​​each vibration risk level, the average value of the feature is calculated and then concatenated to form a regional feature vector; It is input into the trained convolutional neural network and outputs a three-dimensional vector. The three-dimensional vector represents the probability of vibration in the area, the probability of the vibration in the area being of each vibration type, and the probability of the vibration intensity in the area being of various intensities. The probability of vibration is 0-1. The various vibration types include mechanical vibration caused by loose equipment parts, geological vibration caused by changes in geological structure, and vibration caused by normal equipment operation. The various intensities include weak, medium, and strong.

10. A system for identifying underground vibration in a coal mine using the method according to any one of claims 1 to 9, comprising a configuration state sensor, a camera device, an operation parameter recording unit, a weight calculation module, a feature extraction module, a region division module and a prediction module, characterized in that: Configuration status sensor: used to obtain various vibration data of each device in real time; Camera device: used to collect the appearance image of the equipment in real time; Preprocessing vibration data and appearance images of equipment; Operation parameter recording unit: This unit is connected to the status sensor and is used to record and store the working hours, load status and historical operation parameters of each device; Weight calculation module: used to calculate the weight of each device based on working hours, load status and historical operating parameters; Feature extraction module: extract vibration features based on vibration data, multiply all vibration features of each device by the weight of the corresponding device; extract feature points of the device from the collected appearance image through the scale-invariant feature transformation algorithm, identify the coordinate changes of the feature points of the device, and output all feature point coordinate changes in the form of vectors as image features; Area division module: obtain historical vibration location data, and use the geographic information system GIS to divide the underground area of ​​the coal mine into multiple areas with different vibration risk levels by combining the vibration data and vibration location data; Prediction module: According to the divided areas, the weighted vibration features of all devices in each vibration risk level area are fused and concatenated with the image features to form a regional feature vector. The regional feature vector is input into the trained convolutional neural network to predict the probability, type and intensity of vibration in the area.

Images