Mine safety monitoring method based on AI identification monitoring system

Abstract

The invention provides a mine safety monitoring method based on an AI identification monitoring system, and belongs to the technical field of safety management, and the method specifically comprises the steps: determining a matching operation type of a monitoring region according to the matching conditions of monitoring images at different moments in a preset time period and different preset operation types, when the existing AI security identification model of the monitoring device does not meet the requirement according to the historical security data of the matched operation type, the identification accuracy of different types of security risks is determined based on different AI security identification models. On the basis of the matching operation type and historical occurrence data of different types of safety risks under the historical matching operation type of the monitoring area, determining a matching recognition model corresponding to the monitoring device, and performing mine safety risk recognition processing on the monitoring device by using the matching recognition model, and the reliability of monitoring and processing the safety risk is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of safety management, and in particular relates to a mine safety monitoring method based on an AI recognition monitoring system. Background Art

[0002] Mines involve the mining, processing and transportation of minerals, which inevitably involve a variety of safety risk factors. Therefore, in order to achieve the supervision and processing of safety risk factors, in the invention patent application CN102999873A "Digital Mine Safety Monitoring and Control System", different functional subsystems are used to browse and process these interactive information, and the control information is sent to the operating system and database management system platform through the safety monitoring and control system basic platform. It has the advantages of realizing mine safety monitoring, safety warning and safety control, but there are the following technical problems: When conducting safety monitoring and processing in mines, there are often multiple dimensions of safety risks for the monitoring areas corresponding to different monitoring devices, such as falling object risk, impact risk, falling risk, environmental pollutant risk, etc. At the same time, there are certain differences in the safety risks of different monitoring areas under operation, maintenance, shutdown, etc. If a fixed AI recognition model is used, it will inevitably lead to the reliability of safety risk monitoring and processing being difficult to meet the requirements.

[0003] In response to the above technical problems, the present application specifically provides a mine safety monitoring method based on an AI recognition monitoring system. Summary of the invention

[0004] To achieve the purpose of the present invention, the present invention adopts the following technical solutions: According to one aspect of the present invention, a mine safety monitoring method based on an AI recognition monitoring system is provided.

[0005] A mine safety monitoring method based on AI recognition monitoring system, specifically comprising: S1 obtains a monitoring image of a monitoring area corresponding to a monitoring device, and when it is determined that a matching operation type in the monitoring area corresponding to the monitoring device has changed according to matching data between the monitoring image and the initial operation type within a preset period, proceeds to the next step; S2 determines the matching conditions between the surveillance images at different times and different preset operation types based on the surveillance images at different times within the preset time period, and determines the matching operation type of the surveillance area based on the matching conditions; S3 determines the historical occurrence data of different types of safety risks under the matching operation type according to the historical safety data of the matching operation type, and proceeds to the next step when it is determined by using the historical occurrence data that the existing AI safety recognition model of the monitoring device does not meet the requirements; S4 determines the matching recognition model corresponding to the monitoring device based on the recognition accuracy of different AI safety recognition models in different types of safety risks and the historical occurrence data of different types of safety risks under the matching operation type and the historical matching operation type of the monitoring area, and uses the matching recognition model to identify and process the mine safety risks of the monitoring device.

[0006] The beneficial effects of the present invention are: By utilizing the historical occurrence data of different types of safety risks under matching operation types, it is determined whether the existing AI safety identification model of the monitoring device meets the requirements. This not only takes into account the differences in the requirements for identification accuracy due to the differences in the historical number of occurrences of different types of safety risks under matching operation types, but also takes into account the identification accuracy of different types of safety risks under matching operation types. It is possible to determine from multiple angles whether the existing AI safety identification model meets the requirements, which not only avoids the technical problem that the reliability of safety risk identification cannot meet the requirements due to frequent switching of AI safety identification models, but also ensures the accuracy of mine safety risk identification.

[0007] The matching recognition model corresponding to the monitoring device is determined based on the historical occurrence data of different types of safety risks under the matching operation type and the historical matching operation type of the monitoring area, as well as the recognition accuracy of different AI safety recognition models for different types of safety risks. This not only takes into account the recognition accuracy of different AI recognition models for mine safety risks under the current matching operation type, but also takes into account different types of safety risks under historical matching operation types. This ensures both the recognition accuracy under the current matching operation type and the recognition accuracy under the historical matching operation type, thereby avoiding frequent switching of matching recognition models.

[0008] A further technical solution is that the monitoring image is determined according to an analysis result of an image storage device corresponding to the monitoring device.

[0009] A further technical solution is that the initial operation type is the operation type determined in the monitoring area before the current moment.

[0010] A further technical solution is that the operation type includes an operation status type, a maintenance status type of different types of equipment, and a shutdown status type.

[0011] A further technical solution is to determine that a matching operation type in a monitoring area corresponding to the monitoring device has changed, specifically including: Based on the initial job type, determine the image feature corresponding to the initial job type and use it as the matching image feature; Determine, based on surveillance images at different times within a preset period, surveillance images with matching image features, and use them as matching surveillance images; The number of the matching monitoring images within a preset time period is used to determine whether a matching operation type in a monitoring area corresponding to the monitoring device changes.

[0012] A further technical solution is that, when the number of the matching monitoring images is less than a preset number of monitoring images within a preset time period, it is determined that a matching operation type in a monitoring area corresponding to the monitoring device has changed.

[0013] A further technical solution is that the method for determining the matching recognition model corresponding to the monitoring device is: Based on the historical occurrence data of different types of safety risks under the matching operation type, the historical occurrence times of different types of safety risks are determined; based on the historical occurrence data of different types of safety risks under different historical matching operation types, the historical occurrence times of different types of safety risks under different historical matching operation types are determined; According to the historical occurrence times of different types of security risks, the weight coefficients of different types of security risks under matching operation types and the weight coefficients under different historical matching operation types are determined according to a preset mapping function; Determine the comprehensive recognition accuracy of different AI security recognition models under matching operation types and under different historical matching operation types according to the recognition accuracy of different AI security recognition models under different types of security risks and the weight coefficients of different types of security risks; Determine whether there is a job type whose comprehensive recognition accuracy does not meet the requirements. If so, determine that the AI ​​safety recognition model does not belong to a matching recognition model. If not, use the average value of the comprehensive recognition accuracy to determine the matching recognition model corresponding to the monitoring device.

[0014] A further technical solution is that the comprehensive recognition accuracy of the AI ​​security recognition model is determined based on the sum of the products of the recognition accuracy of different types of security risks and the weight coefficients.

[0015] A further technical solution is that the matching recognition model corresponding to the monitoring device is an AI security recognition model with the largest average value of comprehensive recognition accuracy.

[0016] Other features and advantages will be described in the following description. The objects and other advantages of the present invention are realized and obtained by the structures particularly pointed out in the description and drawings.

[0017] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The above and other features and advantages of the present invention will become more apparent by describing in detail exemplary embodiments thereof with reference to the accompanying drawings; Figure 1 It is a flow chart of a mine safety monitoring method based on AI recognition monitoring system; Figure 2 is a flow chart for determining a change in a matching operation type in a monitoring area corresponding to a monitoring device; Figure 3 is a flow chart of a method for determining a matching job type for a monitoring area; Figure 4 It is a flow chart for determining that the existing AI security recognition model of the monitoring device does not meet the requirements. DETAILED DESCRIPTION

[0019] In order to enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the drawings in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on the embodiments of this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this specification.

[0020] There are certain differences in the safety risks of different monitoring areas in operation, maintenance, shutdown, etc. If a fixed AI recognition model is used, it will inevitably lead to the reliability of safety risk monitoring and processing being difficult to meet the requirements.

[0021] Determine whether the matching job type in the monitoring area corresponding to the monitoring device has changed: based on the initial job type, determine the image features corresponding to the initial job type, and use them as matching image features. According to the monitoring images at different times within a preset time period, determine the monitoring images with matching image features, and use them as matching monitoring images. Use the number of matching monitoring images within the preset time period to determine whether the matching job type in the monitoring area corresponding to the monitoring device has changed.

[0022] When the matching operation type in the monitoring area corresponding to the monitoring device has not changed, there is no need to switch the existing AI safety recognition model of the monitoring device.

[0023] The AI ​​security identification model is built based on one or more of CNN, autoencoder, deep belief network, and transfer learning.

[0024] Matching job types: Based on the matching number of image features between surveillance images at different times and those corresponding to preset job types, determine surveillance images whose matching number is greater than the preset matching number, take surveillance images whose matching number is greater than the preset matching number as matching images, and take the preset job type with the largest number of matching images as the matching job type.

[0025] Determine whether the existing AI security recognition model of the monitoring device does not meet the requirements: determine the attention risk type among different types of security risks based on the historical number of occurrences, and determine the recognition accuracy of different attention risk types based on the existing AI security recognition model of the monitoring device and the historical recognition data under different attention risk types; when the average value of the recognition accuracy of different attention risk types does not meet the requirements, determine that the existing AI security recognition model of the monitoring device does not meet the requirements.

[0026] Matching recognition model: Determine the matching recognition model based on the average recognition accuracy of different types of security risks under different matching job types and historical matching job types, and take the AI ​​security recognition model with the largest average recognition accuracy as the matching recognition model.

[0027] To solve the above problems, according to one aspect of the present invention, Figure 1 As shown, a first aspect is provided. The present invention provides a mine safety monitoring method based on an AI recognition monitoring system, specifically comprising: S1 obtains a monitoring image of a monitoring area corresponding to a monitoring device, and when it is determined that a matching operation type in the monitoring area corresponding to the monitoring device has changed according to matching data between the monitoring image and the initial operation type within a preset period, proceeds to the next step; Furthermore, the monitoring image is determined according to an analysis result of an image storage device corresponding to the monitoring device.

[0028] Specifically, the initial operation type is the operation type determined in the monitoring area before the current moment.

[0029] It should be noted that the operation types include operating status types, maintenance status types of different types of equipment, and shutdown status types.

[0030] Optionally, if Figure 2 As shown, determining that a matching operation type in a monitoring area corresponding to the monitoring device has changed specifically includes: Based on the initial job type, determine the image feature corresponding to the initial job type and use it as the matching image feature; Determine, based on surveillance images at different times within a preset period, surveillance images with matching image features, and use them as matching surveillance images; The number of the matching monitoring images within a preset time period is used to determine whether a matching operation type in a monitoring area corresponding to the monitoring device changes.

[0031] Furthermore, when the number of the matching monitoring images is less than a preset number of monitoring images within a preset time period, it is determined that a matching operation type in the monitoring area corresponding to the monitoring device has changed.

[0032] It should also be noted that the image features corresponding to the initial operation type are constructed based on the equipment used corresponding to the initial operation type, the way the equipment is used by people, and the connection relationship between the equipment.

[0033] In another embodiment, determining that a matching job type in a monitoring area corresponding to the monitoring device has changed specifically includes: Based on the initial job type, determine the image feature corresponding to the initial job type and use it as the matching image feature; According to the surveillance images at different times within the preset time period, determining the matching data between the surveillance images and the image features within the preset time period, and using the matching data to determine the number of surveillance images matching different image features; The number of monitoring images matched by different image features is used to determine whether the matching operation type in the monitoring area corresponding to the monitoring device changes.

[0034] Furthermore, when there is no image feature whose number of matched surveillance images is greater than a preset number of surveillance images, it is determined that a matching operation type in the surveillance area corresponding to the surveillance device has changed.

[0035] In another embodiment, determining that a matching job type in a monitoring area corresponding to the monitoring device has changed specifically includes: S11 determines the image features corresponding to the initial operation type based on the initial operation type, and uses the image features as matching image features, and determines the monitoring images with matching image features according to the monitoring images at different times in a preset period of time, and uses the monitoring images as matching monitoring images; S12 determines feature matching coefficients of different matching monitoring images according to the number of matching image features in different matching monitoring images and the types of different matching image features; S13 determines a comprehensive matching coefficient using the feature matching coefficients of different matching monitoring images within a preset time period, and determines whether a matching operation type in a monitoring area corresponding to the monitoring device changes according to the comprehensive matching coefficient.

[0036] It should be noted that the comprehensive matching coefficient is determined according to the weighted sum of the feature matching coefficients of different matching monitoring images.

[0037] Optionally, the above step S11 includes the following contents: S111 determines the image features corresponding to the initial operation type based on the initial operation type, and uses the image features as matching image features. When it is determined that there is no monitoring image with matching image features according to the monitoring images at different times in the preset time period, it is determined that the matching operation type in the monitoring area corresponding to the monitoring device has changed. When there is a monitoring image with matching image features, the process proceeds to step S112. S112: taking the surveillance images with matching image features as matching surveillance images, and when the number of the matching surveillance images is greater than the preset surveillance image number threshold, proceeding to step S113; when the number of the matching surveillance images is not greater than the preset surveillance image number threshold, proceeding to step S12; S113 determines that when the number of monitoring images matching different image features is greater than the preset image number threshold based on the number of monitoring images matching different image features, it is determined that the matching operation type in the monitoring area corresponding to the monitoring device has not changed. When there is an image feature whose number of matching monitoring images is not greater than the preset image number threshold, proceed to step S12.

[0038] Optionally, the above step S12 includes the following contents: S121 determines a feature matching coefficient based on the number of monitoring images matched with different image features. When the feature matching coefficient is greater than a preset feature matching coefficient threshold, it is determined that the matching operation type in the monitoring area corresponding to the monitoring device has not changed. When the feature matching coefficient is not greater than the preset feature matching coefficient threshold, the process proceeds to step S122. S122: when the feature matching coefficient is within the preset matching coefficient interval, the process proceeds to step S123; when the feature matching coefficient is not within the preset matching coefficient interval, it is determined that the matching operation type in the monitoring area corresponding to the monitoring device has changed; S123: When determining a matching monitoring image whose number of matching image features is greater than a preset number of image features through the number of matching image features in different matching monitoring images, proceed to step S124; when there is no matching monitoring image whose number of matching image features is greater than the preset number of image features, proceed to step S125; S124: When the number of matching monitoring images whose number of matching image features is greater than the preset number of image features is greater than the preset number threshold, it is determined that the matching operation type in the monitoring area corresponding to the monitoring device has changed; when the number of matching monitoring images whose number of matching image features is greater than the preset number of image features is not greater than the preset number threshold, the process proceeds to step S125; S125 determines the feature matching coefficients of different matching monitoring images by the number of matching image features in different matching monitoring images and the types of different matching image features. When the number of matching monitoring images whose feature matching coefficients are greater than a preset feature matching coefficient threshold is greater than the preset number threshold, it is determined that the matching operation type in the monitoring area corresponding to the monitoring device has changed. When the number of matching monitoring images whose feature matching coefficients are greater than the preset feature matching coefficient threshold is not greater than the preset number threshold, proceed to step S13.

[0039] S2 determines the matching conditions between the surveillance images at different times and different preset operation types based on the surveillance images at different times within the preset time period, and determines the matching operation type of the surveillance area based on the matching conditions; Furthermore, the matching status of the monitoring image with different preset job types is determined according to the matching number of the monitoring image with the image features corresponding to the different preset job types.

[0040] Specifically, Figure 3 As shown, the method for determining the matching operation type of the monitoring area is: Determine the surveillance images whose matching number is greater than a preset matching number based on the matching number of the surveillance images at different times and the image features corresponding to the preset job type; The surveillance images whose matching number is greater than the preset matching number are regarded as matching images; A matching job type among the preset job types is determined according to the number of matching images.

[0041] Furthermore, the matching job type is a preset job type having the largest number of matching images.

[0042] It can be understood that the method for determining the matching operation type of the monitoring area is: Determine the number of matches between the monitoring images at different times and the image features corresponding to the preset job types based on the matching conditions between the monitoring images at different times and the image features corresponding to the preset job types; Determining the total number of matches of the image features within the preset time period according to the number of matches between the monitoring images at different times within the preset time period and the image features corresponding to the preset job type; A matching job type among the preset job types is determined according to the total matching quantity.

[0043] Furthermore, the matching job type is a preset job type with the largest total matching quantity.

[0044] Optionally, the method for determining the matching operation type of the monitoring area is: S21 determines the number of matches between the surveillance images at different times and the image features corresponding to the preset job types based on the matching conditions between the surveillance images at different times and the image features corresponding to the preset job types, and uses the surveillance images with matching image features as job type matching images; Optionally, the above step S21 includes the following contents: S211 determines the number of matches between the monitoring images at different times and the image features corresponding to the preset job type based on the matching conditions between the monitoring images at different times and the image features corresponding to the preset job type. When the number of matches between the image features corresponding to the preset job type does not meet the requirement, it is determined that the preset job type does not belong to the matching job type. When the number of matches between the image features corresponding to the preset job type meets the requirement, the process proceeds to step S212. S212: When the number of matches of the image features corresponding to the preset job type is within the preset matching number interval, the process proceeds to step S22; when the number of matches of the image features corresponding to the preset job type is not within the preset matching number interval, the process proceeds to step S213; S213 uses the matching monitoring image with image features as the job type matching image. When the number of the job type matching images does not meet the requirement, it is determined that the preset job type does not belong to the matching job type. When the number of the job type matching images meets the requirement, proceed to step S22.

[0045] S22 determines the job type matching coefficients between the job type matching images at different times and the preset job type according to the number of matches between the job type matching images at different times within the preset time period and the image features corresponding to the preset job type; Optionally, the above step S22 includes the following contents: S221 determines whether there is a job type matching image whose matching number of image features is greater than a preset matching number limit value according to the matching number of job type matching images at different moments in the preset time period and the image features corresponding to the preset job type. If so, proceed to step S223. When there is a job type matching image whose matching number of image features is greater than the preset matching number limit value, proceed to step S222. S222: When the number of images matching the job type is within the preset image number interval, it is determined that the preset job type does not belong to the matching job type; when the number of images matching the job type is not within the preset image number interval, the process proceeds to step S223; S223 determines the job type matching coefficients of the job type matching images at different times and the preset job type according to the number of matches between the job type matching images at different times within a preset time period and the image features corresponding to the preset job type. When the average value of the job type matching coefficients of the job type matching images at different times and the preset job type meets the requirement, proceeds to step S23. When the average value of the job type matching coefficients of the job type matching images at different times and the preset job type does not meet the requirement, it is determined that the preset job type does not belong to the matching job type.

[0046] S23 determines the image matching coefficient of the preset job type by summing the weights of the job type matching images at different times and the job type matching coefficients of the preset job type, and determines the matching job type of the monitoring area by using the image matching coefficient.

[0047] Optionally, the matching job type is a preset job type with a maximum image matching coefficient.

[0048] S3 determines the historical occurrence data of different types of safety risks under the matching operation type according to the historical safety data of the matching operation type, and proceeds to the next step when it is determined by using the historical occurrence data that the existing AI safety recognition model of the monitoring device does not meet the requirements; Furthermore, the historical occurrence data of the safety risks is determined according to the historical occurrence times of different types of safety risks under the matching operation type.

[0049] Specifically, Figure 4 As shown, determining that the existing AI security recognition model of the monitoring device does not meet the requirements specifically includes: Determine the historical occurrence times of different types of security risks based on the historical occurrence data, and determine the attention risk type among the different types of security risks based on the historical occurrence times; Determine the recognition accuracy of different attention risk types based on the existing AI safety recognition model of the monitoring device and the historical recognition data under different attention risk types; Whether the existing AI safety recognition model of the monitoring device meets the requirements is determined by the average value of the recognition accuracy of different attention risk types.

[0050] Furthermore, the attention risk type is a type of security risk whose historical occurrence times are greater than a preset occurrence times.

[0051] Optionally, when the average value of the recognition accuracy of different attention risk types is less than a preset accuracy threshold, it is determined that the existing AI safety recognition model of the monitoring device does not meet the requirements.

[0052] It should be noted that when the existing AI security recognition model of the monitoring device meets the requirements, there is no need to switch the recognition model corresponding to the monitoring device.

[0053] Optionally, determining that the existing AI security recognition model of the monitoring device does not meet the requirements specifically includes: Determine the recognition accuracy of different types of security risks based on the existing AI security recognition model of the monitoring device and the historical recognition data of different types of security risks, and when there is no type of security risk whose recognition accuracy meets the requirements, determine that the existing AI security recognition model of the monitoring device does not meet the requirements; When there is a type of security risk whose recognition accuracy meets the requirement: the type of security risk that meets the requirement is taken as an accurate recognition type; when the proportion of the accurate recognition type in the number of different types of security risks does not meet the requirement, it is determined that the existing AI security recognition model of the monitoring device does not meet the requirement; When the proportion of the accurate identification types in the number of different types of security risks meets the requirements: The types of security risks other than the accurately identified types are used as identification deviation types, and the historical occurrence times of different types of security risks are determined based on the historical occurrence data. When it is determined that the total number of historical occurrences of the identification deviation types does not meet the requirements, it is determined that the existing AI security identification model of the monitoring device does not meet the requirements; When it is determined that the total number of historical occurrences of the identification deviation type meets the requirements: Determine the historical occurrence times of different types of security risks based on the historical occurrence data, and determine the attention risk type among the different types of security risks based on the historical occurrence times, and when there is an identification deviation type in the attention risk type, determine that the existing AI security identification model of the monitoring device does not meet the requirements; When there is no identification deviation type in the noted risk type: Whether the existing AI safety recognition model of the monitoring device meets the requirements is determined by the average value of the recognition accuracy of different attention risk types.

[0054] S4 determines the matching recognition model corresponding to the monitoring device based on the recognition accuracy of different AI safety recognition models in different types of safety risks and the historical occurrence data of different types of safety risks under the matching operation type and the historical matching operation type of the monitoring area, and uses the matching recognition model to identify and process the mine safety risks of the monitoring device.

[0055] Specifically, the method for determining the matching recognition model corresponding to the monitoring device is: Based on the historical occurrence data of different types of safety risks under the matching operation type, the historical occurrence times of different types of safety risks are determined; based on the historical occurrence data of different types of safety risks under different historical matching operation types, the historical occurrence times of different types of safety risks under different historical matching operation types are determined; According to the historical occurrence times of different types of security risks, the weight coefficients of different types of security risks under matching operation types and the weight coefficients under different historical matching operation types are determined according to a preset mapping function; Determine the comprehensive recognition accuracy of different AI security recognition models under matching operation types and under different historical matching operation types according to the recognition accuracy of different AI security recognition models under different types of security risks and the weight coefficients of different types of security risks; Determine whether there is a job type whose comprehensive recognition accuracy does not meet the requirements. If so, determine that the AI ​​safety recognition model does not belong to a matching recognition model. If not, use the average value of the comprehensive recognition accuracy to determine the matching recognition model corresponding to the monitoring device.

[0056] Furthermore, the comprehensive recognition accuracy of the AI ​​security recognition model is determined based on the sum of the products of the recognition accuracy of different types of security risks and the weight coefficients.

[0057] It can be understood that the matching recognition model corresponding to the monitoring device is an AI security recognition model with the largest average value of comprehensive recognition accuracy.

[0058] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device, equipment, and non-volatile computer storage medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0059] The above is a description of a specific embodiment of the specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, 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, multitasking and parallel processing are also possible or may be advantageous.

[0060] The above description is only one or more embodiments of this specification and is not intended to limit this specification. For those skilled in the art, one or more embodiments of this specification may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of one or more embodiments of this specification shall be included in the scope of the claims of this specification.

Claims

1. A mine safety monitoring method based on AI recognition monitoring system, characterized in that: Specifically include: Acquire a monitoring image of the monitoring area corresponding to the monitoring device, and when it is determined that the matching operation type in the monitoring area corresponding to the monitoring device has changed according to the matching data of the monitoring image and the initial operation type within a preset time period, proceed to the next step; Based on the surveillance images at different times within a preset time period, determining the matching conditions between the surveillance images at different times and different preset operation types, and determining the matching operation type of the surveillance area based on the matching conditions; Determine historical occurrence data of different types of safety risks under the matching operation type according to the historical safety data of the matching operation type, and proceed to the next step when determining by using the historical occurrence data that the existing AI safety recognition model of the monitoring device does not meet the requirements; Based on the recognition accuracy of different AI safety identification models in different types of safety risks, and based on the historical occurrence data of different types of safety risks under the matching operation type and the historical matching operation type of the monitoring area, the matching identification model corresponding to the monitoring device is determined, and the matching identification model is used to identify and process the mine safety risks of the monitoring device.

2. The mine safety monitoring method based on AI recognition monitoring system according to claim 1, characterized in that: The monitoring image is determined according to the analysis result of the image storage device corresponding to the monitoring device.

3. The mine safety monitoring method based on AI recognition monitoring system according to claim 1, characterized in that: The initial operation type is the operation type determined in the monitoring area before the current moment.

4. The mine safety monitoring method based on AI recognition monitoring system according to claim 3 is characterized in that: The operation types include operating status types, maintenance status types of different types of equipment, and shutdown status types.

5. The mine safety monitoring method based on AI recognition monitoring system according to claim 1, characterized in that: Determining that a matching operation type in a monitoring area corresponding to the monitoring device changes specifically includes: Based on the initial job type, determine the image feature corresponding to the initial job type and use it as the matching image feature; Determine, based on surveillance images at different times within a preset period, surveillance images with matching image features, and use them as matching surveillance images; The number of the matching monitoring images within a preset time period is used to determine whether the matching operation type in the monitoring area corresponding to the monitoring device changes.

6. The mine safety monitoring method based on AI recognition monitoring system according to claim 5, characterized in that: When the number of the matching monitoring images is less than the preset number of monitoring images within the preset time period, it is determined that the matching operation type in the monitoring area corresponding to the monitoring device has changed.

7. The mine safety monitoring method based on AI recognition monitoring system according to claim 5, characterized in that: The image features corresponding to the initial operation type are constructed according to the equipment used corresponding to the initial operation type, the way in which the equipment is used by people, and the connection relationship between the equipment.

8. The mine safety monitoring method based on AI recognition monitoring system according to claim 1, characterized in that: The method for determining the matching recognition model corresponding to the monitoring device is: Based on the historical occurrence data of different types of safety risks under the matching operation type, the historical occurrence times of different types of safety risks are determined; based on the historical occurrence data of different types of safety risks under different historical matching operation types, the historical occurrence times of different types of safety risks under different historical matching operation types are determined; According to the historical occurrence times of different types of security risks, the weight coefficients of different types of security risks under matching operation types and the weight coefficients under different historical matching operation types are determined according to a preset mapping function; Determine the comprehensive recognition accuracy of different AI security recognition models under matching operation types and under different historical matching operation types according to the recognition accuracy of different AI security recognition models under different types of security risks and the weight coefficients of different types of security risks; Determine whether there is a job type whose comprehensive recognition accuracy does not meet the requirements. If so, determine that the AI ​​safety recognition model does not belong to a matching recognition model. If not, use the average value of the comprehensive recognition accuracy to determine the matching recognition model corresponding to the monitoring device.

9. The mine safety monitoring method based on AI recognition monitoring system according to claim 8, characterized in that: The comprehensive recognition accuracy of the AI ​​security recognition model is determined based on the sum of the products of the recognition accuracy of different types of security risks and the weight coefficients.

10. The mine safety monitoring method based on AI recognition monitoring system according to claim 1, characterized in that: The matching recognition model corresponding to the monitoring device is an AI security recognition model with the largest average value of comprehensive recognition accuracy.

Images