Risk identification system based on deep learning

Abstract

The invention relates to the technical field of deep learning, in particular to a risk identification system based on deep learning, which comprises an image acquisition module; the model construction module is used for constructing a mountain three-dimensional model; the data collection module is used for constructing a case library; the deep learning module is used for constructing a mountaineering risk identification model and continuously updating the mountaineering risk identification model according to the information acquired by the monitoring acquisition module; the monitoring acquisition module is used for acquiring mountaineering data; the risk identification module is used for determining the preset segmented path duration of the corresponding climber; and the user communication module is used for transmitting information to the climbers and the workers. The accuracy and sensitivity of risk assessment are improved, further injury caused by untimely rescue is avoided, the rescue efficiency and accuracy are improved, and therefore the problem that accuracy and sensitivity of mountain climbing risk assessment cannot be guaranteed at the same time in the prior art is solved.

Description

Technical Field

[0001] The present invention relates to the field of deep learning, and in particular to a risk identification system based on deep learning. Background Art

[0002] Deep learning is a new research direction in the field of machine learning. It is introduced into machine learning to make it closer to its original goal - artificial intelligence.

[0003] Deep learning is the process of learning the inherent laws and representation levels of sample data. The information obtained in the learning process is of great help in interpreting data such as text, images, and sounds. Its ultimate goal is to enable machines to have analytical learning capabilities like humans and to recognize data such as text, images, and sounds. Deep learning is a complex machine learning algorithm that has achieved results in speech and image recognition that far exceed previous related technologies.

[0004] Deep learning has achieved many results in search technology, data mining, machine learning, machine translation, natural language processing, multimedia learning, speech, recommendation and personalization technology, and other related fields. Deep learning enables machines to imitate human activities such as seeing, hearing, and thinking, solves many complex pattern recognition problems, and makes great progress in artificial intelligence related technologies.

[0005] China Patent Authorization Announcement No.: CN117131444A discloses a method and system for automatic identification of solid waste risks based on deep learning, wherein the method includes: S1: establishing a case library and knowledge base for solid waste risk identification; S2: extracting the characteristics of the case cargo samples in the case library, and combining the characteristics of the cargo samples with the knowledge base to obtain the solid waste risk value of the cargo samples; S3: using the cases in the case library and the solid waste risk values ​​obtained for each case as a data set for training a solid waste risk identification model; S4: establishing a solid waste risk identification model, and the solid waste risk identification model includes models for solid waste risk identification for different types of cargo. The present invention uses machine learning to automatically identify solid waste risks, can quickly and accurately predict the solid waste risk identification results, and improve the efficiency and accuracy of identification.

[0006] China Patent Authorization Announcement No.: CN117811842A discloses a power grid security risk assessment method based on privacy computing, which belongs to the field of power system information security. The method includes obtaining historical operating status data of the power system under each power operator and performing data processing; constructing a two-layer risk assessment indicator system; calculating risk indicators based on the two-layer risk assessment indicator system; dividing risk levels according to the calculated risk indicators, and using historical operating status data and corresponding risk levels to construct local data sets corresponding to each power operator; obtaining the final local model, namely the risk assessment model, based on federated learning and homomorphic encryption methods; obtaining the operating status data of the power system to be evaluated, inputting it into the risk assessment model, and outputting the operating risk level of the power operator; this method can allow multiple power operators to jointly participate in the establishment and update of the risk assessment model without leaking privacy, thereby reducing potential risks and minimizing losses caused by risk events.

[0007] However, the above method has the following problems: the mountaineering environment is complex and diverse, and the risks of mountaineering are often sudden and unpredictable, which poses a huge challenge to the accuracy, real-time and sensitivity of the risk assessment of climbers. The assessment of mountaineering risks in the existing technology often cannot ensure accuracy and sensitivity at the same time. Summary of the invention

[0008] To this end, the present invention provides a risk identification system based on deep learning to overcome the problem that the assessment of mountaineering risks in the prior art often cannot ensure both accuracy and sensitivity.

[0009] To achieve the above objectives, the present invention provides a risk identification system based on deep learning, comprising:

[0010] An image acquisition module, used for acquiring mountain images;

[0011] A model building module is used to build a three-dimensional model based on the mountain image collected by the image acquisition module;

[0012] The data collection module is used to collect mountaineering cases and build a case library in combination with the three-dimensional model of the mountain image;

[0013] The deep learning module is used to build a mountaineering risk identification model and continuously update the mountaineering identification model based on the information collected by the monitoring and acquisition module;

[0014] The monitoring and collection module is used to collect the travel time and rest time of climbers to each rest platform in real time;

[0015] A risk identification module, used to bring the travel time and rest time collected by the monitoring and collection module into a mountaineering risk identification model, so as to determine the estimated time for the climber to walk through each section of the path;

[0016] User communication module, used to transmit information to climbers and staff;

[0017] If the monitoring and acquisition module fails to detect the climber arriving at the rest platform within the estimated time, the risk identification module sends an alarm signal to the staff until the climber is detected arriving at the rest platform again.

[0018] Furthermore, the deep learning module includes a case analysis module, a data preprocessing module, and a model training module;

[0019] The case analysis module is used to analyze the mountaineering cases in the case library to obtain accident occurrence data;

[0020] The data preprocessing module is used to extract relevant features from the accident data, standardize the features and form a training set;

[0021] The model training module is used to train and adjust the mountain climbing risk identification model to determine the mountain climbing duration and classify it into different mountain climbing risk assessment models;

[0022] The accident occurrence data include weather changes, climbing routes, altitude changes, slopes, climbing time for each route, and rest time at each rest platform.

[0023] Furthermore, the model training module includes a long short-term memory network, a convolutional neural network and a model training network;

[0024] The LSTM network is used to process time series data and capture the dependency between mountain climbing duration and risk;

[0025] The convolutional neural network is used to process terrain images and extract spatial features to determine the dependency between paths between different rest platforms and climbing time;

[0026] The model training network uses the training set to perform model training, and uses cross-validation and hyperparameter adjustment to optimize model performance.

[0027] Furthermore, the monitoring acquisition module is connected to the monitoring camera and the cruise drone signal;

[0028] The monitoring cameras are located at various nodes on the mountain trails and are used to collect climbing time data and user data of climbers;

[0029] The cruise drone is used to collect real-time weather data and terrain data.

[0030] Furthermore, each rest platform of the mountain climbing path is determined as a node, and the sections between the rest platforms are divided into data collection sections, data verification sections and risk identification sections;

[0031] The data collection section is used to collect the action data of the climbers and transmit it to the deep learning module;

[0032] The data verification section is used to verify and correct the climber risk assessment model updated by the deep learning module;

[0033] The risk identification section is used to identify the risk of the climber's climbing situation and determine whether it is necessary to send an alarm signal.

[0034] Furthermore, the risk identification module determines the climbing risk assessment model closest to the climber through the time when the climber arrives at a plurality of rest platforms and the rest duration in the data collection section;

[0035] The risk identification module adjusts and optimizes the mountaineering risk assessment model according to the time when climbers arrive at several rest platforms and the rest duration in the data verification section to obtain an actual risk assessment model;

[0036] The risk identification module determines whether it is necessary to send an alarm signal based on the duration of each rest platform determined by the actual risk assessment model in the risk identification section.

[0037] Furthermore, the risk identification system is provided with an alarm value, and the path between any two rest platforms is provided with a corresponding standard passing time, and the monitoring and acquisition module measures the actual passing time of the climber at the rest platform;

[0038] If the climber does not reach the rest platform at the designated climbing time, the risk value will continue to increase linearly until the climber appears on the rest platform;

[0039] The standard passing time is the passing time of the climber on this section of the path determined according to the actual risk assessment model.

[0040] Furthermore, if the climber reaches the rest platform, the monitoring and acquisition module detects the passing time and the rest time, and transmits the passing time and the rest time to the deep learning module to update the actual risk assessment model.

[0041] Furthermore, an alarm value is provided in the risk identification module, and the risk value is compared with the alarm value;

[0042] If the risk value is lower than the alarm value, the user communication module transmits information to the climber and waits for the climber to respond;

[0043] If the risk value is not lower than the alarm value, the user communication module transmits an alarm signal to the staff;

[0044] Among them, the alarm value is related to temperature, precipitation, wind speed, humidity, path length and width, altitude, slope, vegetation coverage, climber's experience level, physical fitness, and equipment condition.

[0045] Further, after the user communication module transmits a confirmation signal to the climber;

[0046] If the climber responds to the confirmation signal, the risk value increases normally;

[0047] If the climber does not respond to the confirmation signal, the risk value increases at twice the original rate.

[0048] Compared with the prior art, the beneficial effect of the present invention lies in that a mountain climbing risk identification model is constructed through a deep learning module, and the climbing time of the climber is collected through a monitoring and collection module to determine whether the climber is in danger, so as to more comprehensively evaluate the risks of the climbing process, improve the accuracy and sensitivity of risk assessment, avoid further injuries caused by untimely rescue, and improve rescue efficiency and accuracy, thereby overcoming the problem that the assessment of mountain climbing risks in the prior art often cannot ensure accuracy and sensitivity at the same time.

[0049] Furthermore, through the deep learning module, the case analysis, training and learning are carried out, and multiple data sources are integrated and effectively fused to form a more comprehensive risk assessment model, which improves the accuracy of the data and reduces the risk judgment errors, thus overcoming the problem that the existing technology for the assessment of mountaineering risks often cannot ensure both accuracy and sensitivity at the same time.

[0050] Furthermore, the monitoring and acquisition module is used to control the monitoring cameras and cruise drones to collect real-time mountain climbing data, and the mountain climbing sections are divided into data collection sections, data verification sections, and risk identification sections. The actual mountain climbing risk identification model is continuously updated and further verified, providing real-time risk identification information, reducing the probability of accidents, improving data accuracy, and reducing risk misjudgment, thereby overcoming the problem that the assessment of mountain climbing risks in the existing technology often cannot ensure both accuracy and sensitivity at the same time.

[0051] Furthermore, by setting risk values ​​and alarm values, judging the degree of crisis of the situation according to the size of the risk value warning value, and providing different solutions, a scientific decision-making basis is provided for climbers and managers, thereby improving the safety of climbers and reducing losses caused by erroneous transmission of alarms, thereby overcoming the problem that the assessment of climbing risks in the existing technology often cannot ensure accuracy and sensitivity at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 It is a schematic diagram of the structure of the risk identification system based on deep learning of the present invention;

[0053] Figure 2 This is a flow chart of a risk identification system based on deep learning according to an embodiment of the present invention. DETAILED DESCRIPTION

[0054] 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.

[0055] 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.

[0056] 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 component 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.

[0057] 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 it can be indirectly connected 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 specific circumstances.

[0058] See also Figure 1 , which is a schematic diagram of the structure of the risk identification system based on deep learning of the present invention, including:

[0059] An image acquisition module, used for acquiring mountain images;

[0060] A model building module is used to build a three-dimensional model based on the mountain image collected by the image acquisition module;

[0061] The data collection module is used to collect mountaineering cases and build a case library in combination with the three-dimensional model of the mountain image;

[0062] The deep learning module is used to build a mountaineering risk identification model and continuously update the mountaineering identification model based on the information collected by the monitoring and acquisition module;

[0063] The monitoring and collection module is used to collect the travel time and rest time of climbers to each rest platform in real time;

[0064] The risk identification module is used to bring the travel time and rest time collected by the monitoring and collection module into the mountaineering risk identification model to determine the estimated time for the climber to walk each section of the path;

[0065] User communication module, used to transmit information to climbers and staff;

[0066] Among them, if the monitoring and collection module does not detect that the climber arrives at the rest platform within the expected time, the risk identification module sends an alarm signal to the staff until the climber is detected to arrive at the rest platform again.

[0067] A mountaineering risk identification model is constructed through a deep learning module, and the climbing time of the climbers is collected through the monitoring and collection module to determine whether the climbers are in danger. The risks of the climbing process are evaluated more comprehensively, which improves the accuracy and sensitivity of risk assessment, avoids further injuries caused by untimely rescue, and improves rescue efficiency and accuracy, thus overcoming the problem that the assessment of mountaineering risks in existing technologies often cannot ensure both accuracy and sensitivity at the same time.

[0068] Please combine Figure 1 See also Figure 2 , which is a flow chart of a risk identification system based on deep learning in an embodiment, including:

[0069] Step S1, the image acquisition module acquires mountain images; the model construction module constructs a three-dimensional model of the mountain; the data collection module constructs a corresponding case library of the three-dimensional model of the mountain;

[0070] Step S2, the deep learning module constructs a mountain climbing risk identification model and continuously updates the mountain climbing identification model based on the information collected by the monitoring and collection module;

[0071] Step S3, the monitoring and collecting module collects the travel time and rest duration of the climber to any rest platform to form climbing data;

[0072] Step S4, the risk identification module determines the actual mountain climbing risk identification model according to the mountain climbing data to determine the corresponding preset segment path duration, and determines the climber's risk value according to the actual duration;

[0073] Step S5, determining the relationship between the climber's risk value and the preset alarm value to determine whether the user communication module transmits an alarm signal to the staff;

[0074] Step S6: When a climber appears on the rest platform, the risk value is reset to zero and the corresponding alarm signal is released.

[0075] Specifically, the deep learning module includes a case analysis module, a data preprocessing module, and a model training module;

[0076] The case analysis module is used to analyze mountaineering cases in the case library to obtain accident occurrence data;

[0077] The data preprocessing module is used to extract relevant features from the accident data, standardize the features and form a training set;

[0078] Model training module, used to train and adjust the mountaineering risk identification model to determine the mountaineering duration and classify it into different mountaineering risk assessment models;

[0079] Among them, the accident data include weather changes, climbing routes, altitude changes, slopes, climbing time for each section of the route, and rest time at each rest platform.

[0080] In implementation, it is foreseeable that the accident data packet may also include temperature, precipitation, humidity, wind speed, vegetation coverage, climber's experience level, physical fitness, equipment condition, etc.

[0081] Specifically, the model training module includes long short-term memory network, convolutional neural network and model training network;

[0082] Long short-term memory networks are used to process time series data and capture the dependency between climbing duration and risk;

[0083] Convolutional neural networks are used to process terrain images and extract spatial features to determine the dependency between paths between different rest platforms and climbing time.

[0084] The model training network uses the training set for model training, and uses cross-validation and hyperparameter adjustment to optimize model performance.

[0085] Through the deep learning module, case analysis, training and learning are carried out, and multiple data sources are integrated and effectively fused to form a more comprehensive risk assessment model, which improves the accuracy of the data and reduces the risk judgment errors, thus overcoming the problem that the existing technology for the assessment of mountaineering risks often cannot ensure both accuracy and sensitivity at the same time.

[0086] Specifically, the monitoring acquisition module is connected with the monitoring camera and the cruise drone signal;

[0087] Surveillance cameras are located at various nodes on the mountain trail to collect climbers’ climbing time data and user data;

[0088] Cruise drones are used to collect real-time weather and terrain data.

[0089] Specifically, each rest platform on the mountain climbing path is determined as a node, and the sections between the rest platforms are divided into data collection sections, data verification sections, and risk identification sections;

[0090] The data collection section is used to collect climbers’ action data and transmit it to the deep learning module;

[0091] The data verification section is used to verify and correct the climber risk assessment model updated by the deep learning module;

[0092] The risk identification section is used to identify the risk of climbers' climbing conditions and determine whether an alarm signal needs to be transmitted.

[0093] Specifically, the risk identification module determines the climbing risk assessment model that is closest to the climber through the time when the climber arrives at several rest platforms and the rest duration in the data collection section;

[0094] The risk identification module adjusts and optimizes the mountaineering risk assessment model in the data verification section based on the time it takes climbers to reach several rest platforms and the length of rest to obtain the actual risk assessment model;

[0095] The risk identification module determines whether it is necessary to send an alarm signal based on the duration of each rest platform determined by the actual risk assessment model on the risk identification section.

[0096] Through the monitoring and acquisition module, monitoring cameras and cruise drones are controlled to collect real-time mountain climbing data, and the mountain climbing sections are divided into data collection sections, data verification sections, and risk identification sections. The actual mountain climbing risk identification model is continuously updated and further verified, providing real-time risk identification information, reducing the probability of accidents, improving data accuracy, and reducing risk judgment errors, thereby overcoming the problem that the assessment of mountain climbing risks in the existing technology often cannot ensure accuracy and sensitivity at the same time.

[0097] Specifically, the risk identification system has an alarm value, and the path between any two rest platforms has a corresponding standard passing time. The monitoring and acquisition module measures the actual passing time of the climbers at the rest platform;

[0098] If the climber does not reach the rest platform at the designated climbing time, the risk value will continue to increase linearly until the climber appears on the rest platform;

[0099] The standard passing time is the time it takes for climbers to pass this section of the path determined based on the actual risk assessment model.

[0100] During implementation, the mountain road is divided into data collection sections, data verification sections and risk identification sections. When the climber reaches the starting point of the mountain road, the monitoring and collection module collects the climber's information and screens out the climbing risk assessment model that suits the climber through the risk identification module. When the climber climbs the data collection section, the monitoring and collection module collects the time when the climber arrives at each rest platform and the rest duration in real time. The risk assessment model screens this information regardless of the climber. When the climber reaches the end of the data collection section, the risk identification module determines the actual climbing risk assessment model that best suits the climber.

[0101] When a climber sets foot on the data verification section, the actual climbing risk identification module determines the time it takes for the climber to pass between each rest platform, and the monitoring and acquisition module collects the actual passing time to verify whether the actual climbing risk identification module is suitable for this climber. If there is a discrepancy in the data, the deep learning module will update and modify the actual climbing risk identification module.

[0102] When a climber steps onto the risk identification module, the actual climbing risk assessment model determines the time it takes for the climber to pass through each section of the path. When the climber's arrival is not detected at the specified time, the user communication module sends an alarm signal.

[0103] Specifically, if a climber reaches the rest platform, the monitoring and acquisition module detects the passing time and rest time, and transmits the passing time and rest time to the deep learning module to update the actual risk assessment model.

[0104] Specifically, the risk identification module has an alarm value, and the risk value is compared with the alarm value;

[0105] If the risk value is lower than the alarm value, the user communication module transmits information to the climber and waits for the climber's response;

[0106] If the risk value is not lower than the alarm value, the user communication module transmits an alarm signal to the staff;

[0107] Among them, the alarm value is the climber identified by the risk identification model.

[0108] Specifically, after the user communication module transmits a confirmation signal to the climber;

[0109] If the climber does not respond to the confirmation signal, the risk value increases at twice the original rate.

[0110] In the implementation, after the risk identification module screens out the actual climbing risk assessment model for the climber, it is determined that the estimated time for the climber to pass from the first rest platform to the second rest platform is 30 minutes, and a warning value of 100 is set. At the same time, it is determined that after the climber's estimated passing time exceeds 30 minutes, the risk value increases by 5 points per minute;

[0111] When the climber reaches the second rest platform in 20 minutes and rests for 15 minutes, the monitoring and acquisition module transmits the climber's real-time data to the deep learning module, and the deep learning module learns and updates the actual risk assessment model information;

[0112] When the climber does not reach the second rest platform within 30 minutes, the user communication module sends a confirmation message to the climber. If the climber responds to the confirmation message, the risk value increases by 5 points per minute. If the climber does not respond to the message, the risk value increases by 10 points per minute.

[0113] When the climber does not reach the second rest platform within 40 minutes and does not respond to the confirmation message, the risk value reaches 100 points, and the user communication module sends an alarm signal to the staff until the climber appears on any rest platform.

[0114] When the climber does not reach the second rest platform within 40 minutes but responds to the confirmation signal, the risk value is 50 points, which does not reach the warning value. The user communication module does not send an alarm signal to the staff. When the climber reaches the rest platform 50 minutes ago, the risk value is reset and the actual climbing risk identification model is updated.

[0115] It is understandable that the warning line is related to temperature, precipitation, wind speed, humidity, path length and width, altitude, slope, vegetation coverage, climber's experience level, physical fitness, and equipment conditions. Those skilled in the art can determine it themselves during implementation and will not be elaborated here.

[0116] By setting risk values ​​and alarm values, judging the degree of crisis according to the size of the risk value warning value, and providing different solutions, a scientific decision-making basis is provided for climbers and managers, thereby improving the safety of climbers and reducing losses caused by erroneous alarm transmission, thereby overcoming the problem that the assessment of climbing risks in existing technologies often cannot ensure both accuracy and sensitivity at the same time.

[0117] 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.

[0118] 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 risk identification system based on deep learning, characterized in that: include: An image acquisition module, used for acquiring mountain images; A model building module, connected to the image acquisition module, for building a three-dimensional mountain model based on the mountain image acquired by the image acquisition module; A data collection module, connected to the model building module, for collecting mountaineering cases and building a corresponding case library in combination with the three-dimensional model of the mountain; A deep learning module, connected to the data collection module, is used to build a mountaineering risk identification model and continuously update the mountaineering identification model based on the information collected by the monitoring and collection module; The monitoring and acquisition module is connected to the deep learning module and is used to collect the travel time and rest time of each climber to any rest platform in real time, and form the climbing data of the corresponding climber; A risk identification module, connected to the monitoring collection module and the deep learning module, for determining the preset segmented path duration of the corresponding climber according to the climbing data using the climbing risk identification model; A user communication module, connected to the risk identification module, for transmitting information to climbers and staff; Among them, if the monitoring and collection module does not detect that the climber arrives at the corresponding rest platform within the preset segmented path time, the risk identification module sends an alarm signal until the corresponding climber is detected to arrive at any rest platform again.

2. The risk identification system based on deep learning according to claim 1 is characterized in that: The deep learning module includes a case analysis module, a data preprocessing module, and a model training module; The case analysis module is used to analyze the mountaineering cases in the case library to obtain traveling sample data; The data preprocessing module is connected to the case analysis module and is used to extract the travel sample features from the travel sample data, standardize the features and form a training set; The model training module is connected to the data preprocessing module and is used to train and adjust the mountain climbing risk identification model to determine the mountain climbing duration and classify it into different mountain climbing risk assessment models; The travel sample characteristics include weather changes, mountain climbing paths, altitude changes, slopes, duration of each preset segment, and duration of stay on each rest platform.

3. The risk identification system based on deep learning according to claim 2 is characterized in that: The model training module includes a long short-term memory network, a convolutional neural network and a model training network; The long short-term memory network is used to process time series data and capture the mapping relationship between mountain climbing time and risk; The convolutional neural network is used to process terrain images and extract spatial features to determine the mapping relationship between the paths between different rest platforms and the climbing time; The model training network uses the training set to perform model training, and uses cross-validation and hyperparameter adjustment to optimize model performance.

4. The risk identification system based on deep learning according to claim 3 is characterized in that: The monitoring acquisition module is connected with the monitoring camera and the cruise drone signal; The monitoring camera is located at each node preset in the three-dimensional model of the mountain, and is used to collect the climbing time data of each climber and the corresponding user data; The cruise drone is used to collect real-time weather data and real-time terrain data.

5. The risk identification system based on deep learning according to claim 4 is characterized in that: The model building module sets each rest platform of the mountain climbing path as a node, and divides the section between the rest platforms into a data collection section, a data verification section and a risk identification section; The data collection section is used to collect the action data of the climbers and transmit it to the deep learning module; The data verification section is used to verify and correct the climber risk assessment model updated by the deep learning module; The risk identification section is used to identify the risk of the climber's climbing situation and determine whether it is necessary to send an alarm signal.

6. The deep learning-based risk identification system according to claim 5, characterized in that: The risk identification module determines the climbing risk assessment model that is closest to the climber in the data collection section based on the time when the climber arrives at a plurality of rest platforms and the rest duration; The risk identification module adjusts and optimizes the mountaineering risk assessment model according to the time when climbers arrive at several rest platforms and the rest duration in the data verification section to obtain an actual risk assessment model; The risk identification module determines whether it is necessary to send an alarm signal based on the duration of each rest platform determined by the actual risk assessment model in the risk identification section.

7. The deep learning-based risk identification system according to claim 6, characterized in that: The risk identification system is provided with an alarm value, and the path between any two rest platforms is provided with a corresponding standard passing time. The monitoring and acquisition module measures the actual passing time of the climber at the rest platform; If the climber does not reach the rest platform at the designated climbing time, the risk value will continue to increase linearly until the climber appears on the rest platform; The standard passing time is the passing time of the climber on this section of the path determined according to the actual risk assessment model.

8. The deep learning-based risk identification system according to claim 7, characterized in that: If the climber reaches the rest platform, the monitoring and acquisition module detects the passing time and the duration of staying on the platform, and transmits the passing time and the duration of staying on the platform to the deep learning module to update the actual risk assessment model.

9. The deep learning-based risk identification system according to claim 8, characterized in that: The risk identification module is provided with an alarm value, and the risk value is compared with the alarm value; If the risk value is lower than the alarm value, the user communication module transmits information to the climber and waits for the climber to respond; If the risk value is not lower than the alarm value, the user communication module transmits an alarm signal to the staff in the corresponding preset segmented path; Among them, the alarm value is related to temperature, precipitation, wind speed, humidity, path length and width, altitude, slope, vegetation coverage, climber's experience level, physical fitness, and equipment condition.

10. The risk identification system based on deep learning according to claim 9, characterized in that: After the user communication module transmits a confirmation signal to the climber; If the climber responds to the confirmation signal, the risk value increases normally; If the climber does not respond to the confirmation signal, the risk value increases at twice the original rate.

Images