Human body attribute identification method in coal mine transportation scene

Abstract

The invention relates to the field of image processing, in particular to a human body attribute identification method in a coal mine transportation scene, which comprises the following steps: constructing a danger monitoring model, and collecting data information of coal mine field environment and personnel; preprocessing the data information to obtain corresponding feature information; classifying the danger monitoring models according to the danger types and endowing the danger monitoring models with corresponding risk values; the feature information is processed to obtain action behavior information, road condition information and mineral information, the action behavior information, the road condition information and the mineral information are substituted into a danger monitoring model to obtain corresponding risk values, whether danger occurs or not is judged, and an alarm signal is sent; the danger monitoring model is trained through the detected feature information, the original danger monitoring model is covered, and therefore the problem that in the prior art, the accuracy of judging the danger type and the danger occurrence probability through experience is low, subjective factors are many, and consequently the accuracy of risk judgment during coal mine transportation is reduced is solved.

Description

Technical Field

[0001] The present invention relates to the field of image processing, and in particular to a method for recognizing human attributes in a coal mine transportation scenario. Background Art

[0002] The prevalence and severity of accidents in the mining industry vary with the nature and type of mineral. The incidence of fatal accidents in the coal mining industry is much higher than in other mining industries, and the injury rate in the coal mining industry is also significantly higher than in other mining industries. Within the same mining industry, the accident rate in underground mining is higher than that in surface open-pit mining. The fatal and non-fatal accident rates in underground mining are generally 1 to 2 times higher than those in surface or open-pit mining.

[0003] China Patent Authorization Announcement No.: CN118097492B discloses a method for intelligent identification of illegal activities in mines based on artificial intelligence deep learning, including: aggregating multiple video segments newly acquired by all cameras installed in the mine as a simultaneous video segment group, and identifying all identification object area sequences in the simultaneous video segment group; matching all identification object area sequences in the simultaneous video segment group based on the actual spatial displacement of the entity objects corresponding to all identification object area sequences in the mine to obtain an identification object area sequence group; obtaining an illegal activity identification result based on each preset artificial intelligence deep learning illegal activity discrimination model of the camera device that obtains the video segment to which the identification object area sequence in each identification object area sequence group belongs; and using the surveillance video containing multiple perspectives of the identification object and the artificial intelligence deep learning illegal activity discrimination model to accurately identify illegal activities existing in the mine operation scene.

[0004] Chinese patent authorization announcement number: CN118230397B discloses a blockchain-based face recognition method and system. The blockchain-based face recognition method of the application first fuses the face images before and after face posture adjustment to obtain two groups of target face images, and then judges the success of face recognition by the matching results of the target face images and the face library images; in this way, the user's face is actively guided to adjust its posture during face recognition to perform fusion recognition of facial features, which greatly improves the accuracy of live recognition and can avoid the safety hazards caused by recognition of non-live photos; and when the face posture is adjusted, the face posture adjustment guidance is performed according to the abnormal features of the face image before the face posture adjustment to obtain two groups of target face images with differentiated features, and the two groups of differentiated target face images are appropriately weighted to perform feature vector matching, further improving the accuracy of face recognition.

[0005] However, the above method has the following problems: the accuracy of judging the type of danger and the probability of danger occurrence through experience in the prior art is low and the error is large, so there is an urgent need for a technical method to reduce human interference and respond promptly when judging the danger. Summary of the invention

[0006] To this end, the present invention provides a method for identifying human attributes in a coal mine transportation scenario, which is used to overcome the problem in the prior art of low accuracy in judging the type of danger and the probability of danger occurrence through experience, and many subjective factors, which leads to reduced accuracy of risk judgment during coal mine transportation.

[0007] To achieve the above object, the present invention provides a method for human attribute recognition in a coal mine transportation scenario, comprising:

[0008] Build a hazard monitoring model and collect data information on the coal mine environment and personnel;

[0009] Preprocessing the data information to obtain corresponding feature information;

[0010] Classifying the risk monitoring models according to the risk type and assigning a corresponding risk value to each risk monitoring model;

[0011] Processing the characteristic information to obtain action behavior information, and bringing the action behavior information into a danger monitoring model to obtain an action risk value, determining whether the action behavior is dangerous and sending an alarm signal;

[0012] Processing the characteristic information to obtain road condition information, and bringing the road condition information into a danger monitoring model to obtain a road condition risk value, determining whether there is a danger source in the road condition and sending an alarm signal;

[0013] Processing the characteristic information to obtain mine information, and importing the mine information into a risk monitoring model to obtain a mine risk value, determining whether there is a risk of collapse in the coal mine and sending an alarm signal;

[0014] The hazard monitoring model is trained by detecting the action, road condition, characteristic information corresponding to the mine and the hazard type, and the hazard monitoring model is adjusted to cover the original hazard monitoring model.

[0015] Among them, the danger types include road danger, behavioral danger, and landslide danger.

[0016] Furthermore, the preprocessing is to perform a first convolution operation on the data information to obtain low-level features, perform a pooling operation on the low-level features to reduce the dimension of the low-level features, and perform several convolution and pooling operations on the low-level features to obtain corresponding feature information.

[0017] Furthermore, the action feature information in the feature information is screened out, the action feature information is summarized and substituted into the danger monitoring model to determine the action risk value and the risk type, and an alarm signal is sent.

[0018] Furthermore, the risk monitoring model is preset with an action alarm value, and the action risk value is compared with the action alarm value.

[0019] If the action risk value is zero, no signal is sent;

[0020] If the action risk value is not zero and has not reached the action alarm value, a low danger signal is sent;

[0021] If the action risk value reaches the action alarm value, a high-risk signal is sent.

[0022] Furthermore, the road condition characteristic information in the characteristic information is screened out, the road condition characteristic information is summarized and substituted into the danger monitoring model to determine the road condition risk value and the risk type, and an alarm signal is sent.

[0023] Furthermore, the risk monitoring model is preset with a road condition alarm value, and the road condition risk value is compared with the road condition alarm value.

[0024] If the road condition risk value is zero, no signal is sent;

[0025] If the road condition risk value is not zero and has not reached the road condition alarm value, a low-risk signal is sent;

[0026] If the road condition risk value reaches the road condition alarm value, a high-risk signal is sent.

[0027] Furthermore, the mine characteristic information in the characteristic information is screened out, the mine characteristic information is summarized and substituted into the danger monitoring model to determine the mine risk value and the risk type, and send an alarm signal.

[0028] Furthermore, the risk monitoring model is preset with a mine alarm value, and the mine risk value is compared with the mine alarm value.

[0029] If the mine risk value is zero, no signal is sent;

[0030] If the mine risk value is not zero and has not reached the mine alarm value, a low danger signal is sent;

[0031] If the mine risk value reaches the mine alarm value, a landslide danger signal will be sent.

[0032] Furthermore, the road condition risk value, the action risk value and the mine risk value are weighted to obtain a total risk value. An alarm value is preset in the danger monitoring model. The total risk value is compared with the alarm value.

[0033] If the total risk value is zero, the mine is not dangerous;

[0034] If the total risk value is not zero and has not reached the alarm value, a risk warning signal is sent;

[0035] If the total risk value reaches the alarm value, an alarm signal is sent.

[0036] Further, a cycle period is determined, and any cycle includes an adjustment period and an application period.

[0037] When in the adjustment period, the detected characteristic information and risk types are summarized and input into the risk monitoring model, the risk monitoring model is trained and the original risk monitoring model is overwritten;

[0038] When in the application period, the characteristic information is brought into the adjusted danger monitoring model and it is determined whether to send an alarm signal;

[0039] When a cycle is finished, the hazard monitoring model is reset to zero and a new hazard monitoring model is rebuilt for the next cycle.

[0040] Compared with the prior art, the beneficial effect of the present invention lies in that, by constructing a hazard monitoring model and determining the risk category and risk level according to the risk value, the accuracy of hazard identification is improved, the efficiency of hazard discovery is improved, and the probability of hazard occurrence is reduced, thereby overcoming the problem in the prior art of low accuracy in judging the hazard type and the probability of hazard occurrence through experience and many subjective factors, which leads to reduced accuracy of risk judgment during coal mine transportation.

[0041] Furthermore, by preprocessing image data through a multi-layer convolution pooling method, the precision and accuracy of image recognition are improved, which is conducive to the execution of complex image recognition, helps to prevent the overfitting of the model on the training set, reduces data dimensions, and improves computational efficiency and robustness, thereby further overcoming the problem of low accuracy in judging the type of danger and the probability of danger occurrence through experience in the prior art, and the large number of subjective factors, which leads to reduced accuracy of risk judgment during coal mine transportation.

[0042] Furthermore, by identifying and judging the characteristics of action behavior through the hazard monitoring model, the sensitivity to abnormal actions of workers is improved, the damage caused by abnormal worker behavior is reduced, the probability of danger is reduced, and personnel safety is improved, thereby further overcoming the problem of low accuracy in judging the type of danger and the probability of danger occurrence through experience in the existing technology, and the large number of subjective factors, which leads to reduced accuracy of risk judgment during coal mine transportation.

[0043] Furthermore, by identifying and judging the characteristics of road conditions through the hazard monitoring model, the possibility of construction safety problems on site due to road conditions is reduced, the safety of road operation is improved, and the probability of accidents is reduced, thereby further overcoming the problem of low accuracy in judging the type of hazard and the probability of hazard occurrence through experience in the prior art, and the large number of subjective factors, which leads to reduced accuracy of risk judgment during coal mine transportation.

[0044] Furthermore, through the identification and judgment of mine characteristics by the hazard monitoring model, the response time of personnel in the event of a landslide accident is improved, the probability of casualties is reduced, and the safety prevention and control awareness at the construction site is improved, thereby further overcoming the problem of low accuracy in judging the type of hazard and the probability of hazard occurrence through experience in the existing technology, and the large number of subjective factors, which leads to reduced accuracy of risk judgment during coal mine transportation.

[0045] Furthermore, by weighting the three types of risk values ​​and dividing the risk levels, the degree of danger can be accurately judged, cost waste can be reduced, safety awareness can be improved, and the probability of safety accidents can be reduced, thereby further overcoming the problem of low accuracy in judging the type of danger and the probability of danger through experience in the prior art, and the large number of subjective factors, which leads to reduced accuracy in risk judgment during coal mine transportation.

[0046] Furthermore, by determining the adjustment period and the application period, the hazard monitoring model is updated in real time, which improves the accuracy of the model and reduces the probability of error of the model, thereby further overcoming the problem in the prior art of low accuracy in judging the type of hazard and the probability of occurrence of hazard through experience and many subjective factors, which leads to reduced accuracy of risk judgment during coal mine transportation. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 This is a flow chart of the steps of the human attribute recognition method in the coal mine transportation scenario of the present invention;

[0048] Figure 2 This is a flow chart of action alarm level determination of a human attribute recognition method in a coal mine transportation scenario according to an embodiment of the present invention;

[0049] Figure 3 This is a flow chart of road condition alarm level determination in a method for human attribute recognition in a coal mine transportation scenario according to an embodiment of the present invention;

[0050] Figure 4 This is a flow chart of mine alarm level determination in a method for human attribute recognition in a coal mine transportation scenario according to an embodiment of the present invention;

[0051] Figure 5 This is a general flow chart of alarm level determination of a human attribute recognition method in a coal mine transportation scenario according to an embodiment of the present invention. DETAILED DESCRIPTION

[0052] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0053] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the protection scope of the present invention.

[0054] It should be noted that, in the description of the present invention, terms such as "up", "down", "left", "right", "inside" and "outside" indicating directions or positional relationships are based on the directions or positional relationships shown in the drawings. This is merely for the convenience of description and does not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it should not be understood as a limitation on the present invention.

[0055] In addition, it should be noted that in the description of the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected" and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, or it can be the internal communication of two components. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood according to the specific circumstances.

[0056] See also Figure 1 As shown, it is a flowchart of the steps of the human attribute recognition method in the coal mine transportation scenario, including:

[0057] Step S1, constructing a hazard monitoring model and collecting data information on the coal mine environment and personnel;

[0058] Step S2, preprocessing the data information to obtain corresponding feature information;

[0059] Step S3, classifying the risk monitoring models according to the risk type and assigning a corresponding risk value to each risk monitoring model;

[0060] Step S4, processing the characteristic information to obtain action behavior information, and bringing the action behavior information into the danger monitoring model to obtain the action risk value, judging whether the action behavior is dangerous and sending an alarm signal;

[0061] Step S5, processing the characteristic information to obtain road condition information, and importing the road condition information into a danger monitoring model to obtain a road condition risk value, determining whether there is a danger source in the road condition and sending an alarm signal;

[0062] Step S6, processing the characteristic information to obtain mine information, and importing the mine information into a risk monitoring model to obtain a mine risk value, determining whether there is a risk of collapse in the coal mine and sending an alarm signal;

[0063] Step S7, training the hazard monitoring model by detecting the action, road condition, characteristic information corresponding to the mine and the hazard type, adjusting the hazard monitoring model to cover the original hazard monitoring model, and repeating steps S1-S7.

[0064] Among them, the types of danger include road danger, behavioral danger, and landslide danger.

[0065] The data information includes images of workers, roads, and the interior of the mountain;

[0066] The characteristic information includes worker behavior information, safety equipment worn, road lighting, types of road debris, surface deformation, and deformation speed.

[0067] During implementation, a hazard monitoring model is constructed through a deep learning module, image information of the coal mine environment and personnel is collected through an image acquisition module, the image information is convolutionally pooled through a preprocessing module to obtain feature information, the worker behavior characteristics are summarized and processed through a behavior recognition module and brought into the hazard monitoring model to obtain a behavior risk value, the road condition characteristics are summarized and processed through a road condition recognition module and brought into the hazard monitoring model to obtain a road condition risk value, the mine characteristics are summarized and processed through a landslide recognition module and brought into the hazard monitoring model to obtain a mine risk value, it is determined whether each risk value exceeds a predetermined alarm value, the alarm level is determined according to the risk value and the risk value weighted value, and finally, the data after the monitoring response is summarized and the hazard monitoring model is trained to make the hazard monitoring model closer to the actual situation.

[0068] The experimenter can set the corresponding relationship between the alarm level and the risk value according to the actual situation of the coal mine transportation site. In this experiment, when any risk value is not zero and each risk value does not exceed the corresponding warning value, the risk level is judged to be low. When any risk value exceeds the corresponding alarm value, the risk level is judged to be medium. When the risk value exceeds the warning value after weighting, the risk level is judged to be high. In this process, each judgment is independent of each other, and the high risk level has the first priority, the medium risk level has the second priority, and the low risk level has the third priority.

[0069] By constructing a hazard monitoring model and determining the risk category and risk level based on the risk value, the accuracy of hazard identification is improved, the efficiency of hazard detection is improved, and the probability of hazard occurrence is reduced, thus overcoming the problem of low accuracy in judging hazard types and hazard occurrence probabilities through experience in the prior art and the large number of subjective factors, which leads to reduced accuracy in risk judgment during coal mine transportation.

[0070] When the action risk sends a low danger signal, the road condition risk sends a warning signal, and the mine risk does not send a danger signal, and the weighted value does not reach the alarm value, only a warning signal will be sent.

[0071] When the action risk sends a low danger signal, the road condition risk sends a warning signal, and the mine risk does not send a danger signal, when the weighted value reaches the alarm value, only a high danger signal is sent.

[0072] When the action risk does not send a danger signal, the road condition risk does not send a danger signal, and the mine risk sends a low danger signal, and the weighted value does not reach the alarm value, only a low danger signal is sent.

[0073] Specifically, the preprocessing is to perform the first convolution operation on the data information to obtain low-level features, perform pooling operations on the low-level features to reduce the dimensions of the low-level features, and perform several convolution and pooling operations on the low-level features to obtain the corresponding feature information.

[0074] The image data is preprocessed through a multi-layer convolution pooling method to improve the precision and accuracy of image recognition, which is conducive to the execution of complex image recognition, helps to prevent the overfitting of the model on the training set, reduces the data dimension, and improves computational efficiency and robustness, thereby further overcoming the problem of low accuracy in judging the type of danger and the probability of danger occurrence through experience in the existing technology, and the large number of subjective factors, which leads to reduced accuracy of risk judgment during coal mine transportation.

[0075] Specifically, the action feature information in the feature information is screened out, the action feature information is summarized and substituted into the danger monitoring model to determine the action risk value and the risk type, and an alarm signal is sent.

[0076] Please combine Figure 1 See also Figure 2 As shown, it is a flow chart of action alarm level determination of the human attribute recognition method in the coal mine transportation scenario according to an embodiment of the present invention. In the figure, the danger monitoring model is preset with an action alarm value, and the action risk value is compared with the action alarm value.

[0077] If the action risk value is zero, no signal is sent;

[0078] If the action risk value is not zero and has not reached the action alarm value, a low danger signal is sent;

[0079] If the action risk value reaches the action alarm value, a high-risk signal is sent.

[0080] The action alert value can be determined by the experimenter, and its main function is to distinguish the severity of the danger.

[0081] In the implementation, the action alarm value is preset to 100. Through the risk value preset in the hazard monitoring model, the risk value of each behavior of not wearing a safety helmet is 10, and the risk value of each behavior of not wearing safety shoes is 5.

[0082] If the data collected on site shows dangerous behavior of seven people not wearing helmets, the action risk value will be 70, which is less than the action alarm value, and a low-risk signal will be sent.

[0083] If no dangerous behavior occurs in the data collected on-site, the action risk value is 0 and no danger signal is sent.

[0084] If the data collected on site shows dangerous behavior of eight people wearing safety helmets and five people not wearing safety shoes, the action risk value is 105, which exceeds the action alarm value and sends a warning signal.

[0085] By identifying and judging the characteristics of action behavior through the hazard monitoring model, the sensitivity to abnormal actions of workers is improved, the damage caused by abnormal worker behavior is reduced, the probability of danger is reduced, and personnel safety is improved, thereby further overcoming the problem of low accuracy in judging the type of danger and the probability of danger through experience in the existing technology, and the large number of subjective factors, which leads to reduced accuracy of risk judgment during coal mine transportation.

[0086] Specifically, the road condition characteristic information in the characteristic information is screened out, the road condition characteristic information is summarized and substituted into the danger monitoring model to determine the road condition risk value and the risk type, and an alarm signal is sent.

[0087] Please combine Figure 1 See also Figure 3 As shown, it is a flow chart of road condition alarm level determination of the human attribute recognition method in the coal mine transportation scenario according to an embodiment of the present invention. In the figure, the danger monitoring model is preset with a road condition alarm value, and the road condition risk value is compared with the road condition alarm value.

[0088] If the road condition risk value is zero, no signal is sent;

[0089] If the road condition risk value is not zero and has not reached the road condition alarm value, a low-risk signal is sent;

[0090] If the road condition risk value reaches the road condition alarm value, a high-risk signal is sent.

[0091] The road condition warning value can be determined by the experimenter, and its main function is to distinguish the severity of the danger.

[0092] In the implementation, the preset road condition alarm value is 100. Through the risk value preset in the hazard monitoring model, the risk value of each meter of road lighting that does not meet the mine construction standard is 10, and the risk value of each meter of road cracks is 20.

[0093] If the data collected on-site shows that the three-meter road lighting does not meet the mine construction standards and cracks appear on the two-meter road, the road condition risk value is 70, which is less than the road condition alarm value and a low-risk signal is sent.

[0094] If no dangerous behavior occurs in the data collected on-site, the road condition risk value is 0 and no danger signal is sent.

[0095] If the data collected on-site shows that the three-meter road lighting does not meet the mine construction standards and cracks appear on the four-meter road, the road condition risk value is 110, exceeding the road condition alarm value, and a warning signal is sent.

[0096] By identifying and judging the characteristics of road conditions through the hazard monitoring model, the possibility of construction safety problems on site due to road conditions is reduced, the safety of road operation is improved, and the probability of accidents is reduced, thereby further overcoming the problem of low accuracy in judging hazard types and hazard probability through experience in the existing technology, and the large number of subjective factors, which leads to reduced accuracy of risk judgment during coal mine transportation.

[0097] Specifically, the mine characteristic information in the characteristic information is screened out, the mine characteristic information is summarized and substituted into the danger monitoring model to determine the mine risk value and the risk type, and an alarm signal is sent.

[0098] Please combine Figure 1 See also Figure 4 As shown, it is a flow chart of mine alarm level determination of a human attribute recognition method in a coal mine transportation scenario according to an embodiment of the present invention. In the figure, a mine alarm value is preset in a danger monitoring model, and a mine risk value is compared with the mine alarm value.

[0099] If the mine risk value is zero, no signal is sent;

[0100] If the mine risk value is not zero and has not reached the mine alarm value, a low danger signal is sent;

[0101] If the mine risk value reaches the mine alarm value, a landslide danger signal will be sent.

[0102] The mine warning value can be determined by the experimenters, and its main function is to distinguish the severity of the danger.

[0103] In the implementation, the preset mine alarm value is 100. The risk value is determined by the preset risk value in the hazard monitoring model. The risk value is 10 for every 1 degree of surface deformation. The surface deformation rate increases by 1 degree per day and the risk value increases by 20.

[0104] If the data collected on-site shows a surface deformation of 2 degrees and a surface deformation rate increase of 3 degrees per day, the mine risk value is 80, which is less than the mine alarm value and a low-risk signal is sent.

[0105] If no dangerous behavior is found in the data collected on-site, the mine risk value is 0 and no danger signal is sent.

[0106] If the data collected on site shows a surface deformation of 5 degrees and an increase in the surface deformation rate of 3 degrees per day, the mine risk value is 110, exceeding the mine alarm value, and a warning signal is sent.

[0107] By identifying and judging the characteristics of the mine through the hazard monitoring model, the response time of personnel in the event of a landslide accident is improved, the probability of casualties is reduced, and the safety prevention and control awareness at the construction site is improved, thereby further overcoming the problem of low accuracy in judging the type of hazard and the probability of hazard occurrence through experience in the existing technology, and the large number of subjective factors, which leads to reduced accuracy of risk judgment during coal mine transportation.

[0108] Please combine Figure 1 See also Figure 5 As shown, it is a general flow chart of alarm level determination of the human attribute recognition method in the coal mine transportation scenario according to an embodiment of the present invention. In the figure, the road condition risk value, the action risk value and the mine risk value are weighted to obtain a total risk value. An alarm value is preset in the danger monitoring model. The total risk value is compared with the alarm value.

[0109] If the total risk value is zero, the mine is not dangerous;

[0110] If the total risk value is not zero and has not reached the alarm value, a risk warning signal is sent;

[0111] If the total risk value reaches the alarm value, an alarm signal is sent.

[0112] The warning value can be determined by the experimenter, and its main function is to distinguish the severity of the danger.

[0113] In the implementation, the preset alarm value is 100, the weighted value of action risk value is 0.2, the weighted value of road condition risk value is 0.2, and the weighted value of mine risk value is 0.6.

[0114] When the action risk value is 110, the road condition risk value is 80, and the mine risk value is 120, the weighted total risk value is 110, and a high-risk signal is sent.

[0115] When the action risk value is 40, the road condition risk value is 60, and the mine risk value is 20, the weighted total risk value is 32, sending a low danger signal.

[0116] When the action risk value is 80, the road condition risk value is 110, and the mine risk value is 40, the weighted total risk value is 62, and a warning signal is sent.

[0117] By weighting the three types of risk values ​​and dividing the risk levels, the degree of danger can be accurately judged, cost waste can be reduced, safety awareness can be improved, and the probability of safety accidents can be reduced, thereby further overcoming the problem of low accuracy in judging the type of danger and the probability of danger through experience in the existing technology, and the large number of subjective factors, which leads to reduced accuracy in risk judgment during coal mine transportation.

[0118] Specifically, determine the cycle period, any cycle includes an adjustment period and an application period,

[0119] When in the adjustment period, the detected characteristic information and risk types are summarized and input into the hazard monitoring model, the hazard monitoring model is trained and the original hazard monitoring model is overwritten;

[0120] When in the application period, the characteristic information is brought into the adjusted danger monitoring model and it is determined whether to send an alarm signal;

[0121] When a cycle is completed, the hazard monitoring model is reset to zero and rebuilt for the next cycle.

[0122] In practice, one day is a cycle, and the adjustment period in one cycle is from 14:00 to 16:00 (including 14:00 and 16:00), and the application period is from 16:00 to 24:00 and from 0:00 to 14:00 the next day (excluding 14:00 and 16:00)

[0123] When the time is 15:00, after the risk information is detected, the data is transmitted to the hazard monitoring model and the hazard monitoring model is trained to cover the original hazard monitoring model. When it is 18:00, the adjusted hazard monitoring model is used and it is determined whether to send an alarm signal. When it reaches 14:00 the next day, the hazard monitoring model clears the adjustment data of the previous cycle and readjusts the hazard monitoring model.

[0124] By determining the adjustment period and the application period, the hazard monitoring model can be updated in real time, which improves the accuracy of the model and reduces the probability of error of the model, thereby further overcoming the problem in the existing technology of low accuracy in judging the type of hazard and the probability of hazard occurrence through experience and many subjective factors, which leads to reduced accuracy of risk judgment during coal mine transportation.

[0125] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

[0126] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for human attribute recognition in coal mine transportation scenarios, characterized in that: include: Build a hazard monitoring model and collect data information on the coal mine environment and personnel; Preprocessing the data information to obtain corresponding feature information; Classifying the risk monitoring models according to the risk type and assigning a corresponding risk value to each risk monitoring model; Processing the characteristic information to obtain action behavior information, and bringing the action behavior information into a danger monitoring model to obtain an action risk value, determining whether the action behavior is dangerous and sending an alarm signal; Processing the characteristic information to obtain road condition information, and bringing the road condition information into a danger monitoring model to obtain a road condition risk value, determining whether there is a danger source in the road condition and sending an alarm signal; Processing the characteristic information to obtain mine information, and importing the mine information into a risk monitoring model to obtain a mine risk value, determining whether there is a risk of collapse in the coal mine and sending an alarm signal; The hazard monitoring model is trained by detecting the action, road condition, characteristic information corresponding to the mine and the hazard type, and the hazard monitoring model is adjusted to cover the original hazard monitoring model; Among them, the danger types include road danger, behavioral danger, and landslide danger.

2. The human attribute recognition method in coal mine transportation scenario according to claim 1 is characterized in that: The preprocessing is to perform a first convolution operation on the data information to obtain low-level features, perform a pooling operation on the low-level features to adjust the dimensions of the low-level features, and perform several convolution and pooling operations on the low-level features to obtain corresponding feature information.

3. The human attribute recognition method in coal mine transportation scenario according to claim 2 is characterized in that: The action feature information in the feature information is screened out, the action feature information is summarized and substituted into the danger monitoring model to determine the action risk value and the risk type, and an alarm signal is sent.

4. The human attribute recognition method in coal mine transportation scenario according to claim 3 is characterized in that: The risk monitoring model is preset with an action alarm value, and the action risk value is compared with the action alarm value. If the action risk value is zero, no signal is sent; If the action risk value is not zero and has not reached the action alarm value, a low danger signal is sent; If the action risk value reaches the action alarm value, a high-risk signal is sent.

5. The human attribute recognition method in coal mine transportation scenario according to claim 4 is characterized in that: The road condition characteristic information in the characteristic information is screened out, the road condition characteristic information is summarized and substituted into the danger monitoring model to determine the road condition risk value and the risk type, and an alarm signal is sent.

6. The human attribute recognition method in coal mine transportation scenario according to claim 5 is characterized in that: The risk monitoring model is preset with a road condition alarm value, and the road condition risk value is compared with the road condition alarm value. If the road condition risk value is zero, no signal is sent; If the road condition risk value is not zero and has not reached the road condition alarm value, a low-risk signal is sent; If the road condition risk value reaches the road condition alarm value, a high-risk signal is sent.

7. The human attribute recognition method in coal mine transportation scenario according to claim 6 is characterized in that: The mine characteristic information in the characteristic information is screened out, the mine characteristic information is summarized and substituted into the danger monitoring model to determine the mine risk value and the risk type, and an alarm signal is sent.

8. The human attribute recognition method in coal mine transportation scenario according to claim 7 is characterized in that: The risk monitoring model is preset with a mine alarm value, and the mine risk value is compared with the mine alarm value. If the mine risk value is zero, no signal is sent; If the mine risk value is not zero and has not reached the mine alarm value, a low danger signal is sent; If the mine risk value reaches the mine alarm value, a landslide danger signal will be sent.

9. The human attribute recognition method in coal mine transportation scenario according to claim 8 is characterized in that: The road condition risk value, the action risk value and the mine risk value are weighted to obtain a total risk value. An alarm value is preset in the danger monitoring model. The total risk value is compared with the alarm value. If the total risk value is zero, the mine is not dangerous; If the total risk value is not zero and has not reached the alarm value, a risk warning signal is sent; If the total risk value reaches the alarm value, an alarm signal is sent.

10. The human attribute recognition method in coal mine transportation scene according to claim 9 is characterized in that: Determine the cycle period. Any cycle includes an adjustment period and an application period. When in the adjustment period, the detected characteristic information and risk types are summarized and input into the risk monitoring model, the risk monitoring model is trained and the original risk monitoring model is overwritten; When in the application period, the characteristic information is brought into the adjusted danger monitoring model and it is determined whether to send an alarm signal; When a cycle is finished, the hazard monitoring model is reset to zero and a new hazard monitoring model is rebuilt for the next cycle.

Images