A Distributed Optical Fiber Vibration Monitoring Method and System Incorporating UAV Image Recognition

Abstract

The present invention provides a distributed optical fiber vibration monitoring method and system integrating UAV image recognition. Among them, the present invention obtains time series vibration data by using a distributed optical fiber sensor network deployed in a preset terrain, and uses a real-time analysis module to combine an adaptive threshold algorithm and a machine learning-driven abnormal pattern detection mechanism to generate threat event trigger information. According to the threat event trigger information, an optimal flight path plan for the threat event trigger information is generated, target image data obtained by the UAV and three-dimensional terrain data collected by the airborne lidar are collected to generate a panoramic view of the real environment, and a 5G communication network is used in combination with differential privacy protection technology to generate an operation guide containing visualization suggestions. The technical solution provided by the present invention improves the coverage of data collection, reduces the false alarm rate, can generate threat event trigger information in a timely manner, supports rapid response, maximizes the resource utilization rate, and shortens the emergency response time.

Description

Technical Field

[0001] The embodiments of the present invention relate to the technical field of unmanned aerial vehicles, and in particular, to a distributed optical fiber vibration monitoring method integrating unmanned aerial vehicle image recognition. Background Art

[0002] In the fields of security monitoring and emergency response, especially for real-time monitoring of large areas and complex terrains, a method that can quickly respond, accurately locate, and comprehensively analyze vibration signals is needed. Natural disasters (such as earthquakes, landslides), human activities (such as illegal intrusion, construction blasting), or other emergencies may trigger vibrations, and these vibration signals need to be captured and analyzed in a timely manner;

[0003] Currently, traditional vibration monitoring systems mainly rely on ground sensor networks or fixed cameras for local area monitoring. Although these systems can capture vibration signals to a certain extent, they are overwhelmed when faced with large areas and dynamically changing environments. Unmanned aerial vehicle monitoring systems focus on image acquisition and lack professional analytical capabilities for vibration signals, resulting in limitations in identifying and predicting potential threats. Existing solutions usually operate independently and fail to organically combine vibration monitoring with image recognition, thus limiting the overall system efficiency and response speed. Summary of the Invention

[0004] The embodiments of the present invention provide a distributed optical fiber vibration monitoring method and system integrating unmanned aerial vehicle image recognition to solve the problems of being overwhelmed and lacking professional analytical capabilities for vibration signals in the prior art when faced with large areas and dynamically changing environments.

[0005] In a first aspect, the embodiments of the present invention provide a distributed optical fiber vibration monitoring method integrating unmanned aerial vehicle image recognition, including:

[0006] Using a distributed optical fiber sensor network deployed in a preset terrain to synchronously collect vibration signals caused by external events to obtain time-series vibration data;

[0007] Using a real-time analysis module combined with an adaptive threshold algorithm and a machine learning-driven abnormal pattern detection mechanism to analyze the time-series vibration data, distinguish natural background noise and potential threat events, and generate threat event trigger information;

[0008] According to the threat event trigger information, the cloud intelligent unmanned aerial vehicle scheduling system integrates dynamic meteorological data, geographical obstacle information, and unmanned aerial vehicle performance parameters, and generates an optimal flight path plan for the threat event trigger information through a reinforcement learning algorithm;

[0009] Based on the optimal flight path planning for the threat event trigger information, collect the target image data obtained by the drone and the three-dimensional terrain data collected by the airborne lidar, fuse the target image data and the three-dimensional terrain data, and generate a panoramic view of the real environment;

[0010] Use the 5G communication network combined with differential privacy protection technology to transmit the panoramic view of the real environment to the central server, analyze the content transmitted by the panoramic view of the real environment based on the spatio-temporal context understanding model, identify the specific entity that causes the vibration signal and predict the behavior trend of the specific entity, establish a deep association rule between the specific entity and the historical vibration signal pattern, so as to evaluate the nature of the event and the scope of influence of the event nature, calculate the event priority, and generate an operation guide containing visualization suggestions.

[0011] Optionally, according to the threat event trigger information, the cloud intelligent drone scheduling system integrates dynamic meteorological data, geographical obstacle information and drone performance parameters, and generates an optimal flight path planning for the threat event trigger information through a reinforcement learning algorithm, including:

[0012] Determine the threat event location and the threat event influence range in the threat event trigger information, and predict potential event chains based on geographical information systems and historical data analysis to obtain an initial flight path planning environment;

[0013] According to the initial flight path planning environment, integrate dynamic meteorological data, geographical obstacle information, drone performance parameters, real-time traffic flow data and temporary control area information to construct a comprehensive environment model;

[0014] Based on the comprehensive environment model, through a reinforcement learning algorithm combined with a risk assessment model, evaluate different flight path selections, calculate the flight time, energy consumption and flight path risk index, and generate multiple flight path candidate plans;

[0015] For the multiple flight path candidate plans, conduct simulation tests, calculate the expected effects according to preset evaluation indicators, and select the university path with a score greater than the preset value as the optimal flight path planning through a multi-objective optimization algorithm.

[0016] Optionally, according to the initial flight path planning environment, integrate dynamic meteorological data, geographical obstacle information, drone performance parameters, real-time traffic flow data and temporary control area information to construct a comprehensive environment model, including:

[0017] Use the threat event location and the threat event influence range determined in the initial flight path planning environment to integrate the dynamic meteorological data, and analyze the weather change trend of the time based on historical meteorological data to obtain the environmental basic data;

[0018] Based on the environmental basic data, combined with geographical obstacle information, analyze the physical obstacles encountered outside the preset target of the UAV flight through 3D modeling technology to generate an obstacle distribution map;

[0019] Based on the obstacle distribution map and UAV performance parameters, combined with the changes in UAV performance under different meteorological conditions, comprehensively evaluate the UAV's ability to perform tasks under various conditions to generate a UAV performance evaluation report;

[0020] Integrate real-time traffic flow data and temporary control area information, combined with the data of the air traffic management system, comprehensively analyze the airspace usage to generate the analysis results of airspace usage, and establish a dynamic update mechanism to cope with the changes in meteorological conditions, ground traffic flow, air traffic conditions, the setting of temporary control areas, and the UAV's own state during the flight;

[0021] Combine the environmental basic data, obstacle distribution map, UAV performance evaluation report, and airspace usage analysis results, and construct a comprehensive environment model through a multi-source data fusion algorithm.

[0022] Optionally, use the 5G communication network combined with differential privacy protection technology to transmit the panoramic view of the real environment to the central server, parse the content of the panoramic view of the real environment based on the spatio-temporal context understanding model, identify the specific entities that cause vibration signals and predict the behavior trends of the specific entities, establish deep association rules between the specific entities and historical vibration signal patterns to evaluate the nature of the event and the scope of influence of the event nature, and calculate the event priority to generate an operation guide including visualization suggestions, including:

[0023] Use the 5G communication network combined with differential privacy protection technology to securely encrypt and transmit the panoramic view of the real environment, and use distributed edge computing nodes to assist the central server in receiving data to obtain an enhanced reality panoramic view of the environment;

[0024] Use the spatio-temporal context understanding model and multi-modal data analysis framework, combined with deep learning algorithms, to parse the content of the enhanced reality panoramic view of the environment, identify the specific entities that cause vibration signals, analyze the spatial distribution, time series behavior patterns of the specific entities, and the dynamic interaction relationships between the specific entities to obtain specific entity information, and refine the specific entity information through semantic segmentation and target tracking technologies to generate specific entity information and analysis results of specific entity behavior patterns;

[0025] Based on the analysis results of specific entity information and specific entity behavior patterns, combined with machine learning algorithms and Bayesian networks, predict the behavior trends of specific entities. By comparing and matching with the historical vibration signal pattern library, and introducing a causal inference model to explore the potential causal relationships between threat events, establish deep association rules between specific entities and historical vibration signal patterns, and evaluate the geographical features and socioeconomic backgrounds of the locations where threat events occur to enhance the accuracy and relevance of predictions, obtaining the predicted behavior trends and deep association rules;

[0026] Evaluate the nature of threat events and the scope of influence of threat events. By comprehensively evaluating the predicted behavior trends, locations, as well as historical data and socioeconomic factors, use a risk assessment model to calculate the event priority and dynamically adjust the weight parameters to meet the requirements in different scenarios. Integrate the scenario simulation function to generate a threat event impact report and a scenario plan outside the preset goals. Based on the threat event impact report and the scenario plan outside the preset goals, generate an operation guide containing visual suggestions.

[0027] Optionally, use a spatio-temporal context understanding model and a multimodal data analysis framework, combined with deep learning algorithms, to analyze the content of the panoramic view of the augmented reality environment, identify the specific entities that cause vibration signals, analyze the spatial distribution, time series behavior patterns of the specific entities, and the dynamic interaction relationships between the specific entities to obtain specific entity information, and refine the specific entity information through semantic segmentation and object tracking technologies to generate the analysis results of specific entity information and specific entity behavior patterns, including:

[0028] Use a spatio-temporal context understanding model and a multimodal data analysis framework, combined with deep learning algorithms and an adaptive perception network, to comprehensively analyze and process the images and three-dimensional terrain data in the panoramic view of the augmented reality environment. By introducing an adaptive perception network, adjust the parsing accuracy to meet the requirements of scenarios with different complexities, identify the specific entities that cause vibration signals, and extract the visual features, spatial location information, and dynamic behavior features of the specific entities to obtain the initial specific entity information and specific entity dynamic attributes;

[0029] Based on the initial specific entity information and specific entity dynamic attributes, apply geographic information systems and geocoding technologies to analyze the spatial distribution of specific entities, determine the real-time locations of each specific entity in the geographic coordinate system, generate a specific entity distribution map with geographic tags, and accelerate the spatial query efficiency by introducing a geographic spatio-temporal index to obtain an optimized specific entity distribution map;

[0030] According to the optimized specific entity distribution map, a time series prediction model combining a long short-term memory network and a variational autoencoder is used to track and record the behavioral trajectories of each specific entity over time, establish a dynamic interaction relationship model, use the dynamic interaction relationship model to describe the interactions and influences between different entities, and predict the behavioral trends within a preset time to provide forward-looking analysis results, so as to generate the analysis results of the specific entity behavior patterns;

[0031] Based on the analysis results of the specific entity behavior patterns, semantic segmentation and object tracking technologies are applied, combined with the Transformer architecture, to refine the specific entity information and generate the target specific entity information and the analysis results of the target specific entity behavior patterns.

[0032] Optionally, a distributed fiber optic sensor network deployed in a preset terrain is used to synchronously collect vibration signals caused by external events to obtain time series vibration data, including:

[0033] A distributed fiber optic sensor network deployed in a preset terrain is used to synchronously collect vibration signals caused by external events. The optical time domain reflectometer technology is combined with frequency division multiplexing and wavelength division multiplexing technologies to ensure the time and space resolution of the vibration signals, and an edge computing node is introduced for real-time data preprocessing to obtain the initial time series vibration data;

[0034] Through the adaptive data preprocessing module built in the distributed fiber optic sensor network, a deep learning-driven noise filtering algorithm is applied to intelligently clean and format the initial time series vibration data, identify and remove background noise, and retain the vibration signals under preset conditions to generate the target time series vibration data;

[0035] According to the target time series vibration data, a multi-scale adaptive filtering algorithm is applied to separate the natural background noise and the potential threat event signals, capture the characteristics of the vibration signals at different time scales, enhance the recognition of the potential threat event signals, obtain the enhanced time series vibration data, and generate feature vectors for subsequent analysis;

[0036] Using the enhanced time series vibration data and feature vectors, combined with the geographic information system and the global positioning system, the vibration signals are mapped to specific geographical locations. Through spatio-temporal correlation analysis, the real-time location where the vibration occurs and the distribution pattern of the vibration occurrence are determined, and a spatio-temporal database of the vibration signals is established to generate the optimal time series vibration data.

[0037] Optionally, based on the optimal flight path planning for the threat event trigger information, collect the target image data obtained by the drone and the three-dimensional terrain data collected by the airborne lidar, fuse the target image data and the three-dimensional terrain data, and generate a panoramic view of the real environment, including:

[0038] Based on the optimal flight path planning for the threat event trigger information, capture images of the target area and collect three-dimensional terrain data to obtain target image data and three-dimensional terrain data;

[0039] Use the target image data obtained by the drone, combine it with the three-dimensional terrain data collected by the airborne lidar, and perform preliminary processing through the edge computing node to obtain the optimized target image data and the optimized three-dimensional terrain data;

[0040] According to the optimized target image data and the optimized three-dimensional terrain data, apply a multimodal data fusion algorithm to align and fuse the optimized target image data and the optimized three-dimensional terrain data to obtain the fused data, and introduce a time synchronization mechanism to ensure the docking of the fused data in the spatio-temporal dimension to generate an intermediate view of the real environment;

[0041] Integrate the geographic information system and semantic segmentation technology, analyze the intermediate view of the real environment, identify and label the key entities in the intermediate view of the real environment, and generate a panoramic view of the real environment.

[0042] In a second aspect, an embodiment of the present invention provides a distributed optical fiber vibration monitoring system integrating drone image recognition, including:

[0043] An acquisition module, configured to synchronously acquire vibration signals caused by external events by using a distributed optical fiber sensor network deployed in a preset terrain to obtain time series vibration data;

[0044] An analysis module, configured to analyze the time series vibration data by using a real-time analysis module in combination with an adaptive threshold algorithm and a machine learning-driven anomaly pattern detection mechanism to distinguish natural background noise and potential threat events, and generate threat event trigger information;

[0045] An integration module, configured to integrate dynamic meteorological data, geographical obstacle information, and drone performance parameters by a cloud intelligent drone scheduling system according to the threat event trigger information, and generate an optimal flight path planning for the threat event trigger information through a reinforcement learning algorithm;

[0046] A fusion module, configured to collect target image data obtained by a drone and three-dimensional terrain data collected by an airborne lidar based on the optimal flight path planning for threat event trigger information, and fuse the target image data and the three-dimensional terrain data to generate a panoramic view of the real environment;

[0047] An identification module, configured to use a 5G communication network in combination with differential privacy protection technology to transmit the panoramic view of the real environment to a central server, analyze the content of the panoramic view of the real environment based on a spatio-temporal context understanding model, identify the specific entity causing the vibration signal and predict the behavior trend of the specific entity, establish a deep association rule between the specific entity and the historical vibration signal pattern, so as to evaluate the nature of the event and the influence range of the nature of the event, calculate the event priority, and generate an operation guide including visualization suggestions.

[0048] In a third aspect, an embodiment of the present invention provides a computing device, including a processor and a memory, where a computer program is stored in the memory, and the processor is configured to run the computer program to execute a distributed optical fiber vibration monitoring method for fusing drone image recognition according to any one of the first aspect.

[0049] In a fourth aspect, an embodiment of the present invention provides a computer storage medium, on which computer program instructions are stored, and when the computer program instructions are executed by a processor, a distributed optical fiber vibration monitoring method for fusing drone image recognition according to any one of the first aspect is implemented.

[0050] In an embodiment of the present invention, a distributed fiber optic sensor network deployed in a preset terrain is used to synchronously collect vibration signals triggered by external events to obtain time series vibration data; a real-time analysis module combines an adaptive threshold algorithm and a machine learning-driven anomaly pattern detection mechanism to analyze the time series vibration data, distinguish natural background noise and potential threat events, and generate threat event trigger information; according to the threat event trigger information, a cloud intelligent UAV scheduling system integrates dynamic meteorological data, geographical obstacle information, and UAV performance parameters, and generates an optimal flight path plan for the threat event trigger information through a reinforcement learning algorithm; based on the optimal flight path plan for the threat event trigger information, target image data obtained by the UAV and three-dimensional terrain data collected by an airborne lidar are collected, and the target image data and the three-dimensional terrain data are fused to generate a panoramic view of the real environment; a 5G communication network is used in combination with differential privacy protection technology to transmit the panoramic view of the real environment to a central server, and the content of the panoramic view of the real environment is analyzed based on a spatio-temporal context understanding model to identify the specific entity that causes the vibration signal and predict the behavior trend of the specific entity, establish a deep association rule between the specific entity and the historical vibration signal pattern, evaluate the nature of the event and the scope of influence of the event nature, calculate the event priority, and generate an operation guide including visual suggestions; the technical solution provided by the present invention improves the coverage of data collection, enhances the accuracy of the system, reduces the false alarm rate, can generate threat event trigger information in a timely manner, supports rapid response, maximizes the resource utilization rate, shortens the emergency response time, ensures the efficiency and security of data transmission, protects sensitive information at the same time, and meets the requirements of modern monitoring systems for data privacy;

[0051] Furthermore, by determining the threat event location and impact scope in the threat event trigger information and predicting potential event chains based on Geographic Information System (GIS) and historical data analysis, an initial flight path planning environment was obtained. This ensured the accuracy of the basic data for path planning, providing a reliable basis for subsequent optimization; integrating dynamic meteorological data, geographical obstacle information, UAV performance parameters, real-time traffic flow data, and temporary control area information, a comprehensive environment model was constructed. This model considered the influence of various factors, making the path planning closer to the actual situation and improving the safety and feasibility of UAV flight; through a reinforcement learning algorithm combined with a risk assessment model, different flight path options were evaluated, calculating flight time, energy consumption, and flight path risk index, generating multiple flight path candidate plans. This method not only considered flight efficiency but also took into account risk control, ensuring the safe completion of UAV tasks; using simulation tests and calculating the expected effect according to preset evaluation indicators, the path with the highest score was selected as the optimal flight path plan through a multi-objective optimization algorithm. This process ensured the scientificity and rationality of path selection and improved the UAV's ability to cope with complex environments

[0052] Even further, by using the 5G communication network combined with differential privacy protection technology, the panoramic view of the real environment was securely encrypted and transmitted, and distributed edge computing nodes were used to assist the central server in receiving data. This method not only ensured the security and low latency of data transmission but also reduced the burden on the central server and improved the overall system response speed; based on the spatio-temporal context understanding model and multi-modal data analysis framework, combined with deep learning algorithms, the content of the panoramic view of the augmented reality environment was analyzed. This analysis process could not only identify the specific entities causing vibration signals but also analyze the spatial distribution, time series behavior patterns of these entities, and their dynamic interaction relationships with each other, providing a detailed description of the on-site situation; according to the specific entity information and behavior pattern analysis results, combined with machine learning algorithms and Bayesian networks, the behavior trends of specific entities were predicted, and a causal reasoning model was introduced to explore the potential causal relationships between threat events. This method enhanced the accuracy and relevance of predictions, helping to take preventive measures in advance and reduce potential risks; by comprehensively evaluating the predicted behavior trends, locations, historical data, and socio-economic factors, the event priority was calculated using a risk assessment model, and the weight parameters were dynamically adjusted to adapt to different scenario requirements. Integrating a scenario simulation function, a threat event impact report and scenario plans outside the preset goals were generated, providing strong support for decision-making; based on all the above analysis results, an operation guide containing visual suggestions was generated. This guide not only provided decision-making support information required for emergency response but also included customized communication strategies and action plans for different stakeholders, ensuring the effective transmission of information and rapid response.

[0053] These aspects or other aspects of the present invention will become more clearly understandable in the following description of the embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0055] Figure 1 Flowchart of a distributed optical fiber vibration monitoring method integrating UAV image recognition provided for an embodiment of the present invention;

[0056] Figure 2 Structural schematic diagram of a distributed optical fiber vibration monitoring system integrating UAV image recognition provided for an embodiment of the present invention;

[0057] Figure 3 Structural schematic diagram of a computing device provided for an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0058] In order to enable those skilled in the art to better understand the solutions of the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention.

[0059] In some processes described in the specification, claims and above-mentioned drawings of the present invention, a plurality of operations appear in a specific order. However, it should be clearly understood that these operations can be executed not in the order in which they appear in this article or in parallel. The serial numbers of the operations, such as 101, 102, etc., are only used to distinguish different operations, and the serial numbers themselves do not represent any execution order. In addition, these processes may include more or fewer operations, and these operations can be executed in sequence or in parallel. It should be noted that the descriptions such as "first", "second", etc. in this article are used to distinguish different messages, devices, modules, etc., and do not represent a sequence, nor do they limit that "first" and "second" are different types.

[0060] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative efforts belong to the scope of protection of the present invention.

[0061] Figure 1The following is a flowchart of a distributed optical fiber vibration monitoring method integrating UAV image recognition provided by an embodiment of the present invention. As Figure 1 shown, the method includes:

[0062] Step 101: Use a distributed optical fiber sensor network deployed in a preset terrain to synchronously collect vibration signals caused by external events, and obtain time-series vibration data;

[0063] In this step, use a distributed optical fiber sensor network deployed in a preset terrain to synchronously collect vibration signals caused by external events, and obtain time-series vibration data. The distributed optical fiber sensor network is a highly sensitive sensing system. Through optical time domain reflectometer (OTDR) technology, it can monitor minute changes along the optical fiber in real time and capture vibration signals caused by external events such as earthquakes, landslides, and explosions. Time-series vibration data refers to the change in vibration intensity recorded in chronological order for subsequent analysis and processing;

[0064] The key to this step is to achieve wide-area and high-precision vibration signal acquisition. The optical fiber sensors are pre-buried in the terrain to be monitored to form a dense monitoring network. When an external event occurs, the sensors will capture the vibrations caused by these events and convert these vibration signals into electrical signals, which are transmitted to the data center through a communication link. The data center performs preliminary processing on the received data to ensure the time synchronization and integrity of the data, thereby generating time-series vibration data that can be used for further analysis;

[0065] In a large-scale mine monitoring project, the optical fiber sensor network covers the entire mining area and surrounding areas. Whenever there is blasting operation or geological activity, the sensors immediately capture the vibration signals and quickly transmit them to the data center through the optical fiber. The data center processes these signals in real time, identifies abnormal vibration patterns, and triggers an alarm mechanism to notify relevant personnel to take necessary safety measures..

[0066] Step 102: Use a real-time analysis module to combine an adaptive threshold algorithm and a machine learning-driven abnormal pattern detection mechanism to analyze the time-series vibration data, distinguish natural background noise and potential threat events, and generate threat event trigger information;

[0067] In this step, the real-time analysis module refers to a system component that can instantaneously process and parse data. It can analyze data immediately when it arrives, without waiting for the entire dataset to be collected. The adaptive threshold algorithm is a technique for dynamically adjusting the judgment criterion, automatically adjusting the detection threshold according to environmental or data characteristics, so as to more accurately distinguish normal and abnormal situations. The machine learning-driven anomaly pattern detection mechanism uses machine learning models to identify atypical features or behaviors in time series vibration data. These models can learn to distinguish natural background noise (including vibrations caused by natural factors such as wind and rain) and potential threat events (such as illegal intrusion, geological activities, etc., which may pose risks to the safety of facilities) through training, and generate threat event trigger information based on this. This information is used to initiate subsequent response measures or further data collection;

[0068] In the embodiment of the present application, the process of the real-time analysis module combining the adaptive threshold algorithm and the machine learning-driven anomaly pattern detection mechanism to parse time series vibration data is as follows: First, the real-time analysis module receives time series vibration data from the sensor network, and these data reflect the physical activity status in the target area. Then, the data stream is preliminarily screened by the adaptive threshold algorithm to remove most of the background noise caused by natural phenomena. Next, the pre-trained machine learning model is used to deeply analyze the preprocessed data to identify abnormal vibration features that conform to known threat patterns. Once a possible threat event is detected, the system will generate threat event trigger information, which usually includes a preliminary assessment of the nature, location, and severity of the suspected event. Finally, the trigger information is sent to the drone scheduling system or other emergency response units to provide decision support for the next step;

[0069] For example, in the scenario of a mine monitoring project, while the drone is executing tasks according to the optimal flight path planning, the vibration sensors deployed on the ground continuously monitor the vibration conditions around the mining area and transmit the time-series vibration data to the central control system in real time. After receiving these data, the real-time analysis module immediately starts working. It uses an adaptive threshold algorithm to filter out the background noise caused by daily mine operations and natural factors, ensuring that resources are not wasted due to false alarms. Then, an anomaly pattern detection mechanism based on machine learning intervenes. By comparing historical data and current data, it successfully captures an abnormal vibration pattern, which is similar to the previously recorded earthquake precursors or illegal excavation activities. The system quickly generates threat event trigger information, which not only details the location and possible causes of the abnormal vibration but also suggests dispatching a drone to the incident location for further investigation. After receiving the instruction, the drone scheduling system immediately arranges the nearby standby drones to fly along the optimal path to the specified coordinates, ready to capture images and collect three-dimensional terrain data, and at the same time prepare to fuse these data to generate a panoramic view of the real environment for technicians to make more informed decisions. In this process, step 102, as a key link, ensures the rapid linkage from data collection to drone response, improving the efficiency and accuracy in dealing with potential threats.

[0070] Step 103: According to the threat event trigger information, the cloud-based intelligent drone scheduling system integrates dynamic meteorological data, geographical obstacle information, and drone performance parameters, and generates an optimal flight path planning for the threat event trigger information through a reinforcement learning algorithm;

[0071] In this step, the cloud-based intelligent drone scheduling system integrates dynamic meteorological data, geographical obstacle information, and drone performance parameters, and generates an optimal flight path planning for the threat event trigger information through a reinforcement learning algorithm. The dynamic meteorological data includes information such as wind speed, temperature, and humidity that affect flight conditions; the geographical obstacle information covers factors such as terrain and buildings that may impede drone flight; the drone performance parameters involve indicators such as maximum flight speed and endurance. The optimal flight path planning is the best flight route formulated after comprehensively considering the above factors, ensuring that the drone can reach the target location efficiently and safely;

[0072] Based on the threat event trigger information, the cloud-based intelligent drone scheduling system first determines the specific location and its influence range of the event, and constructs an initial flight path planning environment. Then, the system collects the latest dynamic meteorological data, geographical obstacle information, and drone performance parameters, and constructs a comprehensive environment model. On this basis, the system uses a reinforcement learning algorithm combined with a risk assessment model to evaluate the selection of different flight paths, calculates the flight time, energy consumption, and flight path risk index, finally generates multiple candidate solutions, and selects the optimal flight path planning through a multi-objective optimization algorithm;

[0073] In the mine monitoring project, once the threat event trigger information is generated, the cloud-based intelligent drone dispatching system starts working immediately. The system first confirms that the incident occurred near an abandoned mine in the southeast corner of the mining area, and then collects the latest meteorological data in the area (such as strong wind warnings), geographical obstacle information (such as nearby high-voltage towers), and specific performance parameters of the drones used. Through the reinforcement learning algorithm, the system simulates multiple flight paths, evaluates the risks and efficiency of each path, and finally selects the shortest path that avoids high-voltage towers as the optimal flight path planning, ensuring that the drone can arrive at the scene quickly and safely.

[0074] Step 104: Based on the optimal flight path planning for the threat event trigger information, the target image data acquired by the drone and the three-dimensional terrain data collected by the airborne laser radar are collected, and the target image data and the three-dimensional terrain data are integrated to generate a panoramic view of the real environment;

[0075] In this step, based on the optimal flight path planning, the UAV flies along the predetermined path, obtains the target image data and the three-dimensional terrain data collected by the airborne LiDAR, and fuses these data to generate a panoramic view of the real environment. The target image data refers to the photos or videos taken by the UAV camera, which provide visual information; the three-dimensional terrain data is obtained through LiDAR, which contains height information and is used to construct accurate landform structures. The panoramic view of the real environment is a comprehensive description of the target area, integrating visual and terrain information to support subsequent analysis and decision-making;

[0076] The drone flies to the designated location according to the optimal flight path planning, uses a high-resolution camera to capture images of the target area, and uses the onboard laser radar to scan the terrain and collect detailed 3D point cloud data. After the drone returns to the base, it uploads the acquired data to the central server. The multimodal data fusion algorithm on the server processes the image and terrain data to ensure the consistency of the two in time and space dimensions, and finally generates a panoramic view of the real environment, intuitively showing the full picture of the target area;

[0077] In the mine monitoring project, the drone flew along the optimal path to the abandoned mine, used a high-definition camera to capture images around the mine, and created a three-dimensional terrain model of the area through lidar. After the drone returned to the base, all data was uploaded to the central server, and after multi-modal data fusion processing, a detailed panoramic view of the real environment was generated. This view not only shows the appearance of the mine, but also provides detailed information about the surrounding terrain, helping technicians assess potential risks and plan the next step.

[0078] Step 105: Utilize the 5G communication network in combination with differential privacy protection technology to transmit the panoramic view of the real environment to the central server. Based on the spatio-temporal context understanding model, analyze the content of the transmitted panoramic view of the real environment, identify the specific entities that cause the vibration signals, predict the behavior trends of the specific entities, establish deep association rules between the specific entities and historical vibration signal patterns, evaluate the nature of the event and the scope of influence of the event nature, calculate the event priority, and generate an operation guide containing visualization suggestions;

[0079] In this step, utilize the 5G communication network in combination with differential privacy protection technology to securely transmit the panoramic view of the real environment to the central server. Based on the spatio-temporal context understanding model, analyze the content, identify the specific entities that cause the vibration signals, predict the behavior trends of the specific entities, establish deep association rules between the specific entities and historical vibration signal patterns, evaluate the nature of the event and its scope of influence, calculate the event priority, and generate an operation guide containing visualization suggestions. The 5G communication network provides a high-speed and low-latency data transmission channel; differential privacy protection technology ensures the privacy of data during the transmission process; the spatio-temporal context understanding model is an advanced data analysis tool used to analyze complex relationships in image and terrain data; the operation guide provides specific countermeasures based on the analysis results;

[0080] The core task of this step is to analyze the content of the panoramic view of the real environment and extract valuable information from it. The 5G communication network ensures the efficiency and security of data transmission, while differential privacy protection technology safeguards the non-disclosure of sensitive information. After receiving the panoramic view, the central server conducts in-depth analysis through the spatio-temporal context understanding model, identifies the specific vibration source entities, and predicts their future behaviors. The system also compares the current event with historical vibration signal patterns to explore causal relationships, evaluate the scope and severity of the event. Finally, the system calculates the event priority based on the evaluation results and generates an operation guide containing visualization suggestions to guide the emergency response team to take corresponding actions;

[0081] In a mine monitoring project, the data collected by drones is securely transmitted back to the central server through the 5G network. The server uses the spatio-temporal context understanding model to analyze the panoramic view of the real environment, identifies the specific location of the mine collapse, and predicts the possible risk of secondary collapse. The system compares this event with historical vibration data, discovers similar patterns, and indicates the possible existence of a larger area of geological instability. Based on this information, the system calculates the event priority as the highest level and generates a detailed operation guide, suggesting immediate evacuation of nearby personnel and setting up a warning area. Relevant departments act quickly according to the guide, effectively avoiding potential safety accidents.

[0082] Through the above steps, the present invention realizes a complete chain from the high-precision synchronous acquisition of vibration signals to the generation of a panoramic view of the real environment, and then to the final analysis and decision support. Each step is closely connected, forming an efficient and intelligent monitoring system, significantly improving the response speed and accuracy to external events. Especially in the field of unmanned aerial vehicles, this method not only enhances the flexibility and safety of unmanned aerial vehicle missions, but also provides a scientific basis for decision-making, greatly improving the effectiveness and efficiency of emergency response.

[0083] To solve the problems of insufficient intelligence and flexibility in the existing technology path planning, in some embodiments, as described in step 103, according to the threat event trigger information, the cloud intelligent unmanned aerial vehicle scheduling system integrates dynamic meteorological data, geographical obstacle information, and unmanned aerial vehicle performance parameters, and generates an optimal flight path planning for the threat event trigger information through a reinforcement learning algorithm, specifically including:

[0084] Determine the threat event location and threat event influence range in the threat event trigger information, and predict potential event chains based on geographical information systems and historical data analysis to obtain an initial flight path planning environment; according to the initial flight path planning environment, integrate dynamic meteorological data, geographical obstacle information, unmanned aerial vehicle performance parameters, as well as real-time traffic flow data and temporary control area information to construct a comprehensive environment model; based on the comprehensive environment model, through a reinforcement learning algorithm combined with a risk assessment model, evaluate different flight path selections, calculate flight time, energy consumption, and flight path risk indices, and generate multiple flight path candidate plans; for the multiple flight path candidate plans, conduct simulation tests, calculate the expected effects according to preset evaluation indicators, and select the university path with a score greater than the preset value as the optimal flight path planning through a multi-objective optimization algorithm;

[0085] In this embodiment, the threat event trigger information includes the specific location and its influence range, which are obtained based on the analysis of the previous vibration data and are used to guide the drone to the correct location for further investigation. Initializing the flight path planning environment means constructing a preliminary flight environment model according to the location and influence range in the threat event trigger information, combining with the Geographic Information System (GIS) and historical data analysis to predict potential event chains, ensuring the accuracy of the basic data for subsequent path planning. The comprehensive environment model integrates dynamic meteorological data (such as wind speed, temperature, humidity, etc.), geographical obstacle information (such as terrain, buildings, etc.), drone performance parameters (such as maximum flight speed, endurance time, etc.), as well as real-time traffic flow data and temporary control area information, comprehensively considering the influence of various factors to make the path planning closer to the actual situation. Multiple flight path candidate schemes are a series of possible flight paths generated by evaluating different flight path selections through a reinforcement learning algorithm combined with a risk assessment model. The flight time, energy consumption, and flight path risk index are calculated for each path for subsequent optimization and selection. The multi-objective optimization algorithm is a method that can simultaneously consider multiple evaluation indicators (such as flight time, safety, and energy consumption) and select the optimal solution. By conducting simulation tests on multiple flight path candidate schemes, calculating the expected effect according to the preset evaluation indicators, and finally selecting the best path with a score greater than the preset value as the optimal flight path planning;

[0086] In the embodiment of the present application, after receiving the threat event trigger information, the cloud intelligent drone scheduling system first determines the specific location and its influence range of the event, and uses GIS and historical data analysis to predict potential event chains to construct an initial flight path planning environment. Next, the system collects the latest dynamic meteorological data, geographical obstacle information, drone performance parameters, as well as real-time traffic flow data and temporary control area information to construct a comprehensive environment model. On this basis, the system uses a reinforcement learning algorithm combined with a risk assessment model to evaluate different flight path selections, calculate the flight time, energy consumption, and flight path risk index, and generate multiple flight path candidate schemes. Finally, the system conducts simulation tests on these candidate schemes through a multi-objective optimization algorithm, calculates the expected effect according to the preset evaluation indicators, and selects the best path as the optimal flight path planning;

[0087] For example, in a mine monitoring project, when the monitoring system detects an abnormal vibration in the southeast corner of the mining area, the cloud-based intelligent drone scheduling system immediately initiates the path planning process. First, the system confirms that the vibration occurred near an abandoned mine and predicts the possible risk of geological instability. Subsequently, the system collects the latest meteorological data in the area (such as strong wind warnings), geographical obstacle information (such as nearby high-voltage pylons), and the specific performance parameters of the drones used. In addition, considering the busy air traffic flow and temporarily established control areas in the region, the system also obtains relevant real-time data.

[0088] Based on the above information, the system constructs a comprehensive environmental model and simulates various possible flight paths. Through a reinforcement learning algorithm, the system evaluates the safety, efficiency, and energy consumption of each path and generates multiple flight path candidate plans. After simulation tests, the system selects the shortest path that avoids the high-voltage pylons as the optimal flight path plan. This path not only ensures that the drone can quickly reach the scene but also minimizes the flight risk and ensures the successful completion of the task. The drone quickly arrives at the scene according to the planned path, takes high-definition images, and conducts lidar scans, providing valuable first-hand information to help technicians take effective measures in a timely manner and avoid potential safety accidents.

[0089] To solve the problems of insufficient intelligence and flexibility in existing technology path planning, according to the previous embodiment, based on the initialized flight path planning environment, integrating dynamic meteorological data, geographical obstacle information, drone performance parameters, and real-time traffic flow data and temporary control area information, a comprehensive environmental model is constructed, specifically including:

[0090] Using the location and influence range of threat events determined in the initialized flight path planning environment, integrating the dynamic meteorological data, and based on the analysis of historical meteorological data to obtain the weather change trend over time, environmental basic data is obtained; according to the environmental basic data, combined with geographical obstacle information, through three-dimensional modeling technology, the physical obstacles encountered outside the preset target of the drone flight are analyzed to generate an obstacle distribution map; based on the obstacle distribution map and drone performance parameters, combined with the changes in drone performance under different meteorological conditions, a comprehensive evaluation of the drone's ability to perform tasks under various conditions is carried out to generate a drone performance evaluation report; integrating real-time traffic flow data and temporary control area information, combined with the data of the air traffic management system, a comprehensive analysis of the airspace usage is carried out to generate an airspace usage analysis result, and a dynamic update mechanism is established to cope with changes in meteorological conditions, ground traffic flow, air traffic conditions, and the setting of temporary control areas and the state of the drone itself during flight; combining the environmental basic data, obstacle distribution map, drone performance evaluation report, and airspace usage analysis result, through a multi-source data fusion algorithm, a comprehensive environmental model is constructed;

[0091] In this embodiment, the environmental basic data includes threat event locations and their influence ranges, dynamic meteorological data and its historical change trends. These data are used to construct the basic conditions for flight path planning to ensure that the path planning takes into account the impact of weather changes on flight. The obstacle distribution map is a chart generated through three-dimensional modeling technology based on geographical obstacle information, which details the locations and forms of all physical obstacles encountered outside the preset target of the UAV flight, helping to avoid collision risks. The UAV performance evaluation report comprehensively evaluates the performance of the UAV under different meteorological conditions, such as the maximum flight speed, endurance time, etc., and analyzes the UAV's mission execution ability in combination with the obstacle distribution map to ensure the selection of the most suitable flight plan for the current conditions. The analysis result of airspace usage integrates real-time traffic flow data, temporary control area information, and data from the air traffic management system, comprehensively analyzes the usage of the airspace, and establishes a dynamic update mechanism to handle various changes that may occur during flight. The multi-source data fusion algorithm is a technology that integrates data from different sources to ensure that all relevant information can be fully utilized, thereby constructing an accurate and comprehensive integrated environment model;

[0092] In the embodiment of the present application, after receiving the initialized flight path planning environment, the system first integrates the dynamic meteorological data using the determined threat event locations and their influence ranges, and analyzes the weather change trend based on historical meteorological data to obtain the environmental basic data. Next, according to these environmental basic data, combined with geographical obstacle information, an obstacle distribution map is generated through three-dimensional modeling technology to clarify all physical obstacles that the UAV may encounter. Subsequently, based on the obstacle distribution map and the UAV performance parameters, combined with the changes in the UAV performance under different meteorological conditions, a comprehensive evaluation of the UAV's mission execution ability under various conditions is carried out to generate a UAV performance evaluation report. In addition, the system integrates real-time traffic flow data and temporary control area information, combined with data from the air traffic management system, comprehensively analyzes the usage of the airspace, generates an analysis result of airspace usage, and establishes a dynamic update mechanism to handle changes in meteorological conditions, ground traffic flow, air traffic conditions, and the setting of temporary control areas and the UAV's own state during flight. Finally, combining the environmental basic data, the obstacle distribution map, the UAV performance evaluation report, and the analysis result of airspace usage, an integrated environment model is constructed through the multi-source data fusion algorithm;

[0093] For example, in a large-scale mine monitoring project, when the monitoring system detects an abnormal vibration in the southeast corner of the mining area, the cloud-based intelligent drone scheduling system starts the path planning process. The system confirms that the vibration occurs near an abandoned mine and predicts the risk of potential geological instability. To develop the safest and most effective flight path, the system first integrates the latest dynamic meteorological data in the area and predicts the future weather change trend based on historical meteorological data to form the environmental basic data. Next, the system combines geographical obstacle information (such as nearby high-voltage line towers) and generates a detailed obstacle distribution map through 3D modeling technology to ensure that the drone can avoid all potential physical obstacles. Then, based on the obstacle distribution map and the drone performance parameters, the system evaluates the flight ability and endurance of the drone under different meteorological conditions and generates a drone performance evaluation report, providing a scientific basis for subsequent path selection. At the same time, the system also integrates real-time traffic flow data and temporary control area information, and comprehensively analyzes the airspace usage situation in combination with the data provided by the air traffic management system. Considering possible weather changes, increased ground traffic flow, air traffic congestion, and adjustments to temporary control areas, the system establishes a dynamic update mechanism to ensure that the flight path planning can flexibly respond to various emergencies. Finally, through a multi-source data fusion algorithm, the system combines the environmental basic data, the obstacle distribution map, the drone performance evaluation report, and the airspace usage analysis results to construct a comprehensive environment model. This model not only covers all necessary flight conditions but also has high adaptability and flexibility, ensuring that the drone can efficiently and safely complete tasks in a complex and changing environment. The drone quickly arrives at the scene according to the optimized path, takes high-definition images and conducts lidar scans, providing valuable first-hand information to help technicians take effective measures in a timely manner and avoid potential safety accidents.

[0094] To solve the problems of insufficient data transmission security, insufficient parsing depth, and limited decision support in the existing technology, and to increase real-time performance and privacy protection, as another embodiment, according to step 105, the panoramic view of the real environment is transmitted to the central server by using the 5G communication network in combination with differential privacy protection technology. Based on the spatio-temporal context understanding model, the content of the transmitted panoramic view of the real environment is parsed, the specific entity causing the vibration signal is identified, and the behavior trend of the specific entity is predicted. Deep association rules between the specific entity and the historical vibration signal pattern are established to evaluate the nature of the event and the scope of influence of the event nature, and the event priority is calculated to generate an operation guide containing visual suggestions, specifically including:

[0095] Using the 5G communication network in combination with differential privacy protection technology, securely encrypt and transmit the panoramic view of the real environment, and use distributed edge computing nodes to assist the central server in receiving data to obtain a panoramic view of the augmented reality environment; use a spatio-temporal context understanding model and a multi-modal data analysis framework, combined with deep learning algorithms, to analyze the content of the panoramic view of the augmented reality environment, identify the specific entities that cause vibration signals, analyze the spatial distribution, time series behavior patterns of the specific entities, and the dynamic interaction relationships between the specific entities to obtain specific entity information, and refine the specific entity information through semantic segmentation and object tracking technologies to generate specific entity information and analysis results of specific entity behavior patterns; according to the specific entity information and analysis results of specific entity behavior patterns, combine machine learning algorithms and Bayesian networks to predict the behavior trends of specific entities, compare and match with the historical vibration signal pattern library, and introduce a causal reasoning model to explore the potential causal relationships between threat events, establish deep association rules between specific entities and historical vibration signal patterns, and evaluate the geographical features and socio-economic backgrounds of the locations where threat events occur to enhance the accuracy and relevance of predictions, obtaining predicted behavior trends and deep association rules; evaluate the nature of threat events and the scope of influence of threat events, calculate the event priority through a risk assessment model by comprehensively evaluating the predicted behavior trends, locations, historical data, and socio-economic factors, and dynamically adjust the weight parameters to meet the requirements in different scenarios, integrate scenario simulation functions, generate threat event impact reports and scenario plans outside the preset goals, and generate an operation guide containing visual suggestions based on the threat event impact reports and scenario plans outside the preset goals;

[0096] In this embodiment, the panoramic view of the augmented reality environment includes a dataset formed by the target image data obtained by the drone and the three-dimensional terrain data collected by the airborne lidar, after preliminary processing and fusion, for providing a comprehensive and detailed description of the on-site situation; the spatio-temporal context understanding model is an advanced data analysis tool that can analyze the complex relationships in the image and terrain data and extract spatio-temporal features; the multi-modal data analysis framework combines various types of data (such as visual, geographical, time series, etc.) to provide a richer analysis perspective; the specific entity information refers to the objects identified from the panoramic view and their attributes, such as location, shape, behavior pattern, etc.; the analysis results of specific entity behavior patterns are the refined specific entity information through semantic segmentation and object tracking technologies, providing the behavior patterns of these entities over time; the predicted behavior trends and deep association rules are the results obtained through machine learning algorithms and Bayesian networks, aiming to predict possible future behaviors and developments and explore the causal relationships between different events; the threat event impact reports and scenario plans outside the preset goals are documents generated based on the above analysis, providing countermeasures and suggestions;

[0097] In the embodiments of the present application, after receiving the panoramic view of the real environment collected by the drone, the system first uses the 5G communication network combined with differential privacy protection technology to securely encrypt and transmit the panoramic view, and adopts distributed edge computing nodes to assist the central server in receiving data to ensure the security and low latency of data transmission. Next, the system uses a spatio-temporal context understanding model and a multimodal data analysis framework, combined with deep learning algorithms, to analyze the content of the panoramic view of the augmented reality environment, identify the specific entities that cause vibration signals, and analyze the spatial distribution, time series behavior patterns of these entities, and their dynamic interaction relationships with each other to obtain specific entity information. Subsequently, the system further refines the specific entity information through semantic segmentation and object tracking technologies to generate specific entity information and analysis results of specific entity behavior patterns. Then, based on these analysis results, the system combines machine learning algorithms and Bayesian networks to predict the behavior trends of specific entities, and through comparison and matching with the historical vibration signal pattern library, introduces a causal inference model to explore the potential causal relationships between threat events and establish deep association rules between specific entities and historical vibration signal patterns. Finally, the system evaluates the nature and scope of influence of threat events, calculates the event priority using a risk assessment model by comprehensively evaluating the predicted behavior trends, locations, historical data, and socioeconomic factors, and dynamically adjusts the weight parameters to meet the requirements in different scenarios, integrates a scenario simulation function, generates a threat event impact report and a scenario plan outside the preset goals, and finally generates an operation guide containing visual suggestions;

[0098] For example, in a large-scale mine monitoring project, when the monitoring system detected an abnormal vibration in the southeast corner of the mining area, the drone quickly flew to the scene to collect high-resolution images and three-dimensional terrain data, forming a panoramic view of the real environment. To ensure the security and efficiency of data transmission, the system used a 5G communication network combined with differential privacy protection technology to securely encrypt and transmit the panoramic view, and adopted distributed edge computing nodes to assist the central server in receiving data, ensuring that the data was transmitted to the data center completely and accurately; after receiving the data, the data center immediately initiated the parsing process. Through the spatio-temporal context understanding model and multi-modal data analysis framework, combined with deep learning algorithms, the system identified the specific entity that caused the vibration - an abandoned mine, and analyzed in detail the terrain and structural characteristics around the mine. The system further refined this information through semantic segmentation and object tracking technologies, generating specific entity information and the analysis results of the specific entity behavior pattern, revealing the potential risk of geological instability around the mine; next, based on these analysis results, the system combined machine learning algorithms and Bayesian networks to predict the possible future behavior trends of the mine, such as whether a secondary collapse might occur, and through comparison and matching with the historical vibration signal pattern library, found similar geological activity patterns, suggesting the possible existence of a larger range of geological instability. The system introduced a causal reasoning model to explore the potential causal relationships between this event and other historical events, establishing deep association rules between specific entities and historical vibration signal patterns; finally, the system evaluated the nature of the threat event and its impact range, calculated the event priority as the highest level through a comprehensive evaluation of the predicted behavior trends, locations, historical data, and socio-economic factors, and dynamically adjusted the weight parameters to adapt to the current emergency situation. The system generated a threat event impact report and a scenario plan outside the preset goals, suggesting an immediate evacuation of nearby personnel and the setting up of a warning area, and relevant departments acted quickly according to the guidelines, effectively avoiding potential safety accidents.

[0099] To solve the problems of insufficient parsing accuracy, insufficiently fine spatial distribution analysis, and limited behavior pattern prediction ability in the existing technology, according to the previous embodiment, using the spatio-temporal context understanding model and multi-modal data analysis framework, combined with deep learning algorithms, to parse the content of the panoramic view of the augmented reality environment, identify the specific entity that causes the vibration signal, analyze the spatial distribution, time series behavior pattern of the specific entity, and the dynamic interaction relationship between the specific entities, obtain specific entity information, and refine the specific entity information through semantic segmentation and object tracking technologies, generating specific entity information and the analysis results of the specific entity behavior pattern, specifically including:

[0100] Using a spatio-temporal context understanding model and a multi-modal data analysis framework, combined with deep learning algorithms and an adaptive perception network, comprehensively analyze and process the images and 3D terrain data in the panoramic view of the augmented reality environment. By introducing an adaptive perception network, adjust the parsing accuracy to adapt to the scene requirements of different complexities, identify the specific entities that cause vibration signals, and extract the visual features, spatial location information, and dynamic behavior features of the specific entities to obtain initial specific entity information and specific entity dynamic attributes; Based on the initial specific entity information and specific entity dynamic attributes, apply geographic information systems and geocoding technologies to analyze the spatial distribution of specific entities, determine the real-time position of each specific entity in the geographic coordinate system, generate a specific entity distribution map with geographic tags, and accelerate the spatial query efficiency by introducing a geographic spatio-temporal index to obtain an optimized specific entity distribution map; According to the optimized specific entity distribution map, use a time series prediction model combined with a long short-term memory network and a variational autoencoder to track and record the behavior trajectories of each specific entity over time, establish a dynamic interaction relationship model, use the dynamic interaction relationship model to describe the interactions and influences between different entities, and predict the behavior trends within a preset time to provide forward-looking analysis results to generate specific entity behavior pattern analysis results; Based on the specific entity behavior pattern analysis results, apply semantic segmentation and object tracking technologies, combined with the Transformer architecture, to refine the specific entity information to generate target specific entity information and target specific entity behavior pattern analysis results;

[0101] In this embodiment, the spatio-temporal context understanding model is an advanced data processing tool that can analyze the complex relationships in image and terrain data, extract spatio-temporal features, and be used to understand the objects and their behaviors in the scene; the multi-modal data analysis framework combines various types of data (such as vision, geography, time series, etc.), provides a richer analysis perspective, and enhances the understanding of complex scenes; the adaptive perception network is a neural network structure that can adjust the parsing accuracy according to the scene complexity, ensuring the best parsing effect under different conditions; the initial specific entity information and the specific entity dynamic attributes include the objects identified from the panoramic view and their visual features, spatial location information, and dynamic behavior features, which are the basis for further analysis; the specific entity distribution map with geographical tags refers to the distribution map formed by determining the real-time positions of each specific entity in the geographical coordinate system through geographic information system (GIS) and geocoding technologies, and the geographical spatio-temporal index is a technology that accelerates the spatial query efficiency, making it possible to quickly locate and retrieve information in a specific area in a large-scale or high-density dataset; the time series prediction model combined with long short-term memory network (LSTM) and variational autoencoder (VAE) is an advanced method for tracking the behavioral trajectories of each specific entity over time, which can not only describe the interactions and influences between different entities, but also predict future behavioral trends; the dynamic interaction relationship model describes the interactions and influences between different entities, helping to understand the development process of events; the target specific entity information and the analysis results of the target specific entity behavior pattern are the results generated after refining the specific entity information, providing more accurate target objects and their behavior patterns;

[0102] In the embodiments of the present application, the system first uses a spatio-temporal context understanding model and a multi-modal data analysis framework, combines deep learning algorithms and an adaptive perception network, and comprehensively analyzes and processes the images and three-dimensional terrain data in the panoramic view of the augmented reality environment. By introducing the adaptive perception network, the system can adjust the parsing accuracy to meet the requirements of different complexity scenarios, identify the specific entities that cause vibration signals, and extract the visual features, spatial position information, and dynamic behavior features of these entities to obtain initial specific entity information and specific entity dynamic attributes. Based on this information, applying geographic information system (GIS) and geocoding technologies, the system analyzes the spatial distribution of specific entities, determines the real-time position of each specific entity in the geographic coordinate system, generates a specific entity distribution map with geographic tags, and accelerates the spatial query efficiency by introducing a geographic spatio-temporal index to obtain an optimized specific entity distribution map. Next, the system uses a time series prediction model combined with a long short-term memory network (LSTM) and a variational autoencoder (VAE) to track and record the behavior trajectories of each specific entity over time, establish a dynamic interaction relationship model, describe the interactions and influences between different entities, and predict the behavior trends within a preset time to provide forward-looking analysis results to generate specific entity behavior pattern analysis results. Finally, based on the specific entity behavior pattern analysis results, the system applies semantic segmentation and object tracking technologies, combines the Transformer architecture, and refines the specific entity information to generate more accurate target specific entity information and target specific entity behavior pattern analysis results;

[0103] For example, in a large-scale mine monitoring project, when the monitoring system detected an abnormal vibration in the southeast corner of the mining area, the drone quickly flew to the scene to collect high-resolution images and three-dimensional terrain data, forming a panoramic view of the augmented reality environment. To deeply analyze the content of this view, the system first used a spatio-temporal context understanding model and a multi-modal data analysis framework, combined with deep learning algorithms and an adaptive perception network, to comprehensively analyze and process the images and three-dimensional terrain data. Through the adaptive perception network, the system adjusted the analysis accuracy, successfully identified the specific entity causing the vibration - an abandoned mine, and extracted the visual features, spatial location information, and dynamic behavior characteristics of the mine, obtaining the initial specific entity information and specific entity dynamic attributes; based on this information, the system applied geographic information system (GIS) and geocoding technologies to analyze the spatial distribution of specific entities, determine the real-time position of each specific entity in the geographic coordinate system, and generate a distribution map of specific entities with geographic tags. By introducing a geographic spatio-temporal index, the system significantly improved the spatial query efficiency, obtained an optimized distribution map of specific entities, which not only clarified the exact location of the mine but also revealed the detailed conditions of the surrounding geological structures; next, the system used a time series prediction model combining long short-term memory network (LSTM) and variational autoencoder (VAE) to track and record the behavior trajectory of the abandoned mine over time, establishing a dynamic interaction relationship model. This model describes the interactions and influences between the mine and other geological features, and predicts the behavior trends within a preset time, providing forward-looking analysis results. For example, the system predicts the risk of a secondary collapse of the mine in the next few days, indicating greater geological instability; finally, based on the analysis results of the specific entity behavior patterns, the system applied semantic segmentation and object tracking technologies, combined with the Transformer architecture, to refine the specific entity information, generating more accurate target specific entity information and target specific entity behavior pattern analysis results. These results not only helped technicians accurately evaluate the status of the abandoned mine but also provided detailed decision-making support for the emergency response team, suggesting immediate evacuation of nearby personnel and setting up a warning area, and relevant departments quickly took action according to the guidelines, effectively avoiding potential safety accidents.

[0104] To solve the problems of insufficient vibration signal acquisition accuracy, severe background noise interference, and low data analysis efficiency in the existing technology, and to increase the real-time processing ability and spatio-temporal resolution, as another embodiment, according to what is described in step 101, a distributed fiber optic sensor network deployed in a preset terrain is used to synchronously collect vibration signals caused by external events, obtaining time series vibration data, specifically including:

[0105] Using a distributed fiber optic sensor network deployed in a preset terrain, synchronously collect vibration signals triggered by external events. Adopt optical time domain reflectometry technology combined with frequency division multiplexing and wavelength division multiplexing technologies to ensure the time and space resolution of vibration signals, and introduce edge computing nodes for real-time data preprocessing to obtain initial time series vibration data; through an adaptive data preprocessing module built into the distributed fiber optic sensor network, apply a noise filtering algorithm driven by deep learning to intelligently clean and format the initial time series vibration data, identify and remove background noise, and retain vibration signals under preset conditions to generate target time series vibration data; according to the target time series vibration data, apply a multi-scale adaptive filtering algorithm to separate natural background noise from potential threat event signals, capture the characteristics of vibration signals at different time scales, enhance the distinguishability of potential threat event signals, obtain enhanced time series vibration data, and generate feature vectors for subsequent analysis; use the enhanced time series vibration data and feature vectors, combined with geographic information systems and global positioning systems, map the vibration signals to specific geographical locations, determine the real-time location where the vibration occurs and the distribution pattern of the vibration occurrence through spatio-temporal correlation analysis, and establish a spatio-temporal database of vibration signals to generate optimal time series vibration data;

[0106] In this embodiment, the optical time domain reflectometry technology combined with frequency division multiplexing and wavelength division multiplexing technologies ensures the time and space resolution of vibration signals, and edge computing nodes are introduced for real-time data preprocessing to obtain initial time series vibration data; the adaptive data preprocessing module built into the distributed fiber optic sensor network applies a noise filtering algorithm driven by deep learning to intelligently clean and format the initial time series vibration data, identify and remove background noise, and retain vibration signals under preset conditions to generate target time series vibration data; the multi-scale adaptive filtering algorithm is used to separate natural background noise from potential threat event signals, capture the characteristics of vibration signals at different time scales, enhance the distinguishability of potential threat event signals, obtain enhanced time series vibration data, and generate feature vectors for subsequent analysis; geographic information systems and global positioning systems map the vibration signals to specific geographical locations, determine the real-time location where the vibration occurs and the distribution pattern of the vibration occurrence through spatio-temporal correlation analysis, and establish a spatio-temporal database of vibration signals to generate optimal time series vibration data;

[0107] In the embodiments of the present application, first, the system uses a distributed fiber optic sensor network deployed in a preset terrain to synchronously collect vibration signals caused by external events. By adopting the optical time domain reflectometer (OTDR) technology combined with frequency division multiplexing (FDM) and wavelength division multiplexing (WDM) technologies, high time and space resolutions are ensured, and edge computing nodes are introduced for real-time data preprocessing to obtain initial time series vibration data. Next, through the adaptive data preprocessing module built in the distributed fiber optic sensor network, a deep learning-driven noise filtering algorithm is applied to intelligently clean and format the initial time series vibration data, identify and remove background noise, and retain vibration signals that meet preset conditions at the same time, generating target time series vibration data. Then, the system applies a multi-scale adaptive filtering algorithm to separate natural background noise from potential threat event signals, capture the characteristics of vibration signals at different time scales, enhance the recognition of potential threat event signals, obtain enhanced time series vibration data, and generate feature vectors for subsequent analysis. Finally, the system uses the enhanced time series vibration data and feature vectors, combines with the geographic information system (GIS) and the global positioning system (GPS), accurately maps the vibration signals to specific geographical locations, determines the real-time location where the vibration occurs and the distribution pattern of the vibration occurrence through spatio-temporal correlation analysis, and establishes a spatio-temporal database of vibration signals, and finally generates the best time series vibration data;

[0108] For example, in a large-scale mine monitoring project, when the monitoring system detects an abnormal vibration in the southeast corner of the mining area, the distributed fiber optic sensor network quickly initiates the acquisition process. The system uses optical time domain reflectometry (OTDR) technology combined with frequency division multiplexing (FDM) and wavelength division multiplexing (WDM) technologies to ensure the time and spatial resolution of the vibration signals, and introduces edge computing nodes for real-time data preprocessing to obtain the initial time series vibration data. Next, through the built-in adaptive data preprocessing module, the system applies a deep learning-driven noise filtering algorithm to intelligently clean and format these initial time series vibration data, identifies and removes background noises such as wind and vehicle movement, retains the vibration signals that may be caused by geological activities or human activities, and generates the target time series vibration data. Subsequently, the system applies a multi-scale adaptive filtering algorithm to further separate the natural background noise from the potential threat event signals, captures the characteristics of the vibration signals at different time scales, enhances the recognition of the potential threat event signals, obtains the enhanced time series vibration data, and generates feature vectors for subsequent analysis. Finally, the system uses the enhanced time series vibration data and feature vectors, combines with the geographic information system (GIS) and the global positioning system (GPS), accurately maps the vibration signals to specific geographical locations, determines the real-time location where the vibration occurs and the distribution pattern of the vibration through spatio-temporal correlation analysis, and establishes a spatio-temporal database of the vibration signals. Ultimately, the generated optimal time series vibration data not only clarifies the exact location of the vibration but also reveals the detailed conditions of the surrounding geological structure.

[0109] To solve the problems of insufficient fusion and shallow analysis depth of existing technology image and terrain data, and to increase the real-time processing ability and multi-modal data analysis ability, as another embodiment, according to what is described in step 103, based on the optimal flight path planning for threat event trigger information, collect the target image data obtained by the unmanned aerial vehicle and the three-dimensional terrain data collected by the airborne lidar, fuse the target image data and the three-dimensional terrain data, and generate a panoramic view of the real environment, specifically including:

[0110] Based on the optimal flight path planning for threat event trigger information, image capture is performed on the target area, and three-dimensional terrain data is collected to obtain target image data and three-dimensional terrain data; using the target image data obtained by the UAV, combined with the three-dimensional terrain data collected by the airborne lidar, preliminary processing is carried out through the edge computing node to obtain the optimized target image data and the optimized three-dimensional terrain data; according to the optimized target image data and the optimized three-dimensional terrain data, applying a multimodal data fusion algorithm, aligning and fusing the optimized target image data and the optimized three-dimensional terrain data to obtain the fused data, and introducing a time synchronization mechanism to ensure the docking of the fused data in the spatio-temporal dimension to generate an intermediate view of the real environment; integrating a geographic information system and semantic segmentation technology, parsing the intermediate view of the real environment, identifying and labeling the key entities in the intermediate view of the real environment to generate a panoramic view of the real environment;

[0111] In this embodiment, the target image data includes photos or videos taken by the UAV camera, which are used to provide visual information; the three-dimensional terrain data is the height information collected by the airborne lidar (LiDAR), which is used to construct an accurate geomorphic structure; the edge computing node is a computing unit that performs preliminary processing near the data source, reducing data transmission latency and improving processing efficiency; the optimized target image data and the optimized three-dimensional terrain data refer to the data sets with improved quality after preliminary processing, ensuring the accuracy and efficiency of subsequent fusion; the multimodal data fusion algorithm is a technology that combines multiple types of data (such as vision, geography, time series, etc.), providing a richer analysis perspective and ensuring the consistency of data from different sources in the spatio-temporal dimension; the intermediate view of the real environment is the preliminary view formed by the fused data, which contains the optimized target image data and the three-dimensional terrain data but has not been finally parsed; the time synchronization mechanism ensures that the timestamps of all data are consistent, enabling the perfect docking of the fused data in the spatio-temporal dimension; the geographic information system (GIS) is a system used to capture, store, operate, analyze, manage, and display all forms of geographic data; the semantic segmentation technology is an image processing method that can identify different objects in an image and label their categories, thereby enhancing the understanding of the environment; the panoramic view of the real environment is a comprehensive description of the target area, integrating visual and terrain information, and supporting subsequent analysis and decision-making;

[0112] In the embodiments of the present application, first, based on the optimal flight path planning for threat event trigger information, the system guides the unmanned aerial vehicle (UAV) to fly to the target area for image capture and collect three-dimensional terrain data, obtaining target image data and three-dimensional terrain data. Next, using the target image data obtained by the UAV and combining with the three-dimensional terrain data collected by the airborne lidar, preliminary processing is carried out through the edge computing node to remove noise and optimize the image quality and terrain accuracy, obtaining the optimized target image data and the optimized three-dimensional terrain data. Then, based on these optimized data, applying the multi-modal data fusion algorithm, the optimized target image data and the optimized three-dimensional terrain data are aligned and fused, and at the same time, a time synchronization mechanism is introduced to ensure the consistency of the fused data in the spatio-temporal dimension to generate an intermediate view of the real environment. Finally, integrating the geographic information system (GIS) and semantic segmentation technology, the intermediate view of the real environment is analyzed to identify and label key entities, generating the final panoramic view of the real environment, providing detailed and intuitive information for subsequent analysis and decision-making;

[0113] For example, in a large-scale mine monitoring project, when the monitoring system detected an abnormal vibration in the southeast corner of the mining area, the cloud-based intelligent drone scheduling system immediately initiated the path planning process and determined the optimal flight path plan. The drone quickly flew to the vicinity of the abandoned mine according to the planned path and began to collect high-resolution images and three-dimensional terrain data. First, the drone captured detailed images of the target area and used the on-board lidar to collect the three-dimensional terrain data of the area, obtaining the target image data and three-dimensional terrain data. Next, the system utilized the target image data obtained by the drone, combined with the three-dimensional terrain data collected by the on-board lidar, and performed preliminary processing through the edge computing node to remove background noise, optimize the image quality and terrain accuracy, and obtain the optimized target image data and optimized three-dimensional terrain data. Then, based on these optimized data, the system applied the multi-modal data fusion algorithm to align and fuse the optimized target image data and optimized three-dimensional terrain data, and at the same time introduced a time synchronization mechanism to ensure the consistency of the fused data in the spatio-temporal dimension, generating an intermediate view of the real environment, which not only ensured the quality of the data but also enhanced the accuracy of subsequent analysis. Finally, the system integrated the Geographic Information System (GIS) and semantic segmentation technology to deeply analyze the intermediate view of the real environment, identify and label key entities, such as the specific location of the abandoned mine and the characteristics of the surrounding geological structure, and generate a panoramic view of the real environment. This view not only shows the appearance of the mine but also provides detailed information about the surrounding terrain, helping technicians evaluate potential risks and plan the next steps. This intelligent data processing method not only improves the accuracy of image and terrain data fusion but also enhances the analysis depth, greatly improving the effectiveness and efficiency of emergency response. Based on these accurate data, the drone can complete tasks more quickly and accurately, providing valuable on-site information to help technicians take effective measures in a timely manner to avoid potential safety accidents.

[0114] Figure 2 FIG. provides a schematic structural diagram of a distributed optical fiber vibration monitoring system integrating drone image recognition according to an embodiment of the present invention, as Figure 2 shown, the system includes:

[0115] An acquisition module 21, configured to synchronously acquire vibration signals caused by external events by using a distributed optical fiber sensor network deployed in a preset terrain, and obtain time-series vibration data;

[0116] An analysis module 22, configured to analyze the time-series vibration data by using a real-time analysis module in combination with an adaptive threshold algorithm and a machine learning-driven abnormal pattern detection mechanism, distinguish natural background noise and potential threat events, and generate threat event trigger information;

[0117] An integration module 23, configured to integrate dynamic meteorological data, geographical obstacle information, and UAV performance parameters according to the threat event trigger information by means of a cloud intelligent UAV scheduling system, and generate an optimal flight path plan for the threat event trigger information through a reinforcement learning algorithm;

[0118] A fusion module 24, configured to collect target image data obtained by a UAV and three-dimensional terrain data collected by an airborne lidar based on the optimal flight path plan for the threat event trigger information, and fuse the target image data and the three-dimensional terrain data to generate a panoramic view of the real environment;

[0119] An identification module 25, configured to transmit the panoramic view of the real environment to a central server by using a 5G communication network in combination with differential privacy protection technology, analyze the content of the panoramic view of the real environment transmitted based on a spatio-temporal context understanding model, identify the specific entity that causes the vibration signal and predict the behavior trend of the specific entity, establish a deep association rule between the specific entity and the historical vibration signal pattern, so as to evaluate the nature of the event and the influence range of the nature of the event, calculate the event priority, and generate an operation guide including visualization suggestions.

[0120] Figure 2 The described distributed optical fiber vibration monitoring system integrating UAV image recognition can execute Figure 1 The described distributed optical fiber vibration monitoring method integrating UAV image recognition in the illustrated embodiment, the implementation principle and technical effects will not be elaborated. For the distributed optical fiber vibration monitoring system integrating UAV image recognition in the above embodiment, the specific ways for each module and unit to execute operations have been described in detail in the embodiments related to the method, and will not be elaborated here.

[0121] In a possible design, Figure 2 The distributed optical fiber vibration monitoring system integrating UAV image recognition in the illustrated embodiment can be implemented as a computing device, such as Figure 3 As shown, the computing device may include a storage component 31 and a processing component 32;

[0122] The storage component 31 stores one or more computer instructions, wherein the one or more computer instructions are called and executed by the processing component 32.

[0123] The processing component 32 is configured to synchronously collect vibration signals triggered by external events by using a distributed optical fiber sensor network deployed in a preset terrain to obtain time-series vibration data;

[0124] Using a real-time analysis module in combination with an adaptive threshold algorithm and a machine learning-driven anomaly pattern detection mechanism, parse the time series vibration data, distinguish natural background noise and potential threat events, and generate threat event trigger information;

[0125] According to the threat event trigger information, the cloud intelligent UAV scheduling system integrates dynamic meteorological data, geographical obstacle information, and UAV performance parameters, and generates an optimal flight path plan for the threat event trigger information through a reinforcement learning algorithm;

[0126] Based on the optimal flight path plan for the threat event trigger information, collect the target image data obtained by the UAV and the three-dimensional terrain data collected by the airborne lidar, fuse the target image data and the three-dimensional terrain data, and generate a panoramic view of the real environment;

[0127] Using a 5G communication network in combination with differential privacy protection technology, transmit the panoramic view of the real environment to the central server, parse the content of the panoramic view of the real environment based on a spatio-temporal context understanding model, identify the specific entity that causes the vibration signal and predict the behavior trend of the specific entity, establish a deep association rule between the specific entity and the historical vibration signal pattern, evaluate the nature of the event and the scope of influence of the event nature, calculate the event priority, and generate an operation guide including visual suggestions.

[0128] Among them, the processing component 32 may include one or more processors to execute computer instructions to complete all or part of the steps in the above method. Of course, the processing component may also be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors or other electronic components for executing the above method.

[0129] The storage component 31 is configured to store various types of data to support the operation of the terminal. The storage component can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0130] Of course, the computing device may also necessarily include other components, such as input / output interfaces, display components, communication components, etc.

[0131] The input / output interface provides an interface between the processing component and the peripheral interface module, and the above-mentioned peripheral interface module can be an output device, an input device, etc.

[0132] The communication component is configured to facilitate communication between the computing device and other devices in a wired or wireless manner, etc.

[0133] Among them, the computing device can be a physical device or an elastic computing host provided by a cloud computing platform, etc. At this time, the computing device can refer to a cloud server, and the above-mentioned processing component, storage component, etc. can be basic server resources rented or purchased from a cloud computing platform.

[0134] An embodiment of the present invention also provides a computer storage medium storing a computer program, and when the computer program is executed by a computer, it can implement the above-mentioned Figure 1 distributed optical fiber vibration monitoring method integrating UAV image recognition shown in the embodiment.

[0135] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the above-described systems, devices, and units can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0136] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art can understand and implement it without creative efforts.

[0137] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the essence of the above technical solution or the part that contributes to the prior art can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0138] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A distributed optical fiber vibration monitoring method integrating UAV image recognition, characterized in that, Including: Utilize a distributed fiber optic sensor network deployed in a preset terrain to synchronously collect vibration signals triggered by external events, obtaining time series vibration data; Utilize a real-time analysis module combined with an adaptive threshold algorithm and a machine learning-driven anomaly pattern detection mechanism to analyze the time series vibration data, distinguish natural background noise and potential threat events, and generate threat event trigger information; According to the threat event trigger information, the cloud intelligent UAV scheduling system integrates dynamic meteorological data, geographical obstacle information, and UAV performance parameters, and generates an optimal flight path plan for the threat event trigger information through a reinforcement learning algorithm; Based on the optimal flight path plan for the threat event trigger information, collect the target image data obtained by the UAV and the three-dimensional terrain data collected by the airborne lidar, fuse the target image data and the three-dimensional terrain data, and generate a panoramic view of the real environment; Utilize a 5G communication network combined with differential privacy protection technology to securely encrypt and transmit the panoramic view of the real environment, and use distributed edge computing nodes to assist the central server in receiving data to obtain an enhanced reality environment panoramic view; Using a spatio-temporal context understanding model and a multi-modal data analysis framework, combined with deep learning algorithms and an adaptive perception network, comprehensively analyze and process the images and 3D terrain data in the panoramic view of the augmented reality environment. By introducing an adaptive perception network, adjust the parsing accuracy to meet the requirements of scenarios with different complexities, identify the specific entities that cause vibration signals, and extract the visual features, spatial location information, and dynamic behavior characteristics of the specific entities to obtain initial specific entity information and specific entity dynamic attributes. Based on the initial specific entity information and specific entity dynamic attributes, apply geographic information systems and geocoding technologies to analyze the spatial distribution of specific entities, determine the real-time location of each specific entity in the geographic coordinate system, generate a specific entity distribution map with geographic tags, and accelerate the spatial query efficiency by introducing a geographic spatio-temporal index to obtain an optimized specific entity distribution map. According to the optimized specific entity distribution map, use a time series prediction model combined with a long short-term memory network and a variational autoencoder to track and record the behavior trajectories of each specific entity over time, establish a dynamic interaction relationship model, use the dynamic interaction relationship model to describe the interactions and influences between different entities, and predict the behavior trends within a preset time to provide forward-looking analysis results to generate specific entity behavior pattern analysis results. Based on the specific entity behavior pattern analysis results, apply semantic segmentation and object tracking technologies, combined with the Transformer architecture, to refine the specific entity information to generate target specific entity information and target specific entity behavior pattern analysis results. According to the specific entity information and specific entity behavior pattern analysis results, combine machine learning algorithms and Bayesian networks to predict the behavior trends of specific entities, compare and match with the historical vibration signal pattern library, and introduce a causal inference model to explore the potential causal relationships between threat events, establish deep association rules between specific entities and historical vibration signal patterns, and evaluate the geographic features and socio-economic backgrounds of the locations where threat events occur to enhance the accuracy and relevance of predictions, obtaining predicted behavior trends and deep association rules. Evaluate the nature of threat events and the scope of influence of threat events. By comprehensively evaluating the predicted behavior trends, locations, as well as historical data and socio-economic factors, use a risk assessment model to calculate the event priority and dynamically adjust the weight parameters to meet the requirements of different scenarios, integrate a scenario simulation function to generate a threat event impact report and a scenario plan outside the preset goals. Based on the threat event impact report and the scenario plan outside the preset goals, generate an operation guide containing visualization suggestions.

2. The method according to claim 1, characterized in that According to the threat event trigger information, the cloud intelligent UAV scheduling system integrates dynamic meteorological data, geographic obstacle information, and UAV performance parameters, and generates an optimal flight path plan for the threat event trigger information through a reinforcement learning algorithm, including: Determine the threat event location and threat event impact scope in the threat event trigger information, and predict potential event chains based on geographical information systems and historical data analysis to obtain an initialized flight path planning environment; According to the initialized flight path planning environment, integrate dynamic meteorological data, geographical obstacle information, UAV performance parameters, real-time traffic flow data, and temporary control area information to construct a comprehensive environment model; Based on the comprehensive environment model, evaluate different flight path selections through a reinforcement learning algorithm combined with a risk assessment model, calculate flight time, energy consumption, and flight path risk indices, and generate multiple flight path candidate plans; For the multiple flight path candidate plans, conduct simulation tests, calculate the expected effects according to preset evaluation indicators, and select the flight path with a score greater than the preset value as the optimal flight path plan through a multi-objective optimization algorithm.

3. The method according to claim 2, characterized in that According to the initialized flight path planning environment, integrate dynamic meteorological data, geographical obstacle information, UAV performance parameters, real-time traffic flow data, and temporary control area information to construct a comprehensive environment model, including: Utilize the threat event location and threat event impact scope determined in the initialized flight path planning environment to integrate dynamic meteorological data, and analyze the weather change trend over time based on historical meteorological data to obtain environmental basic data; According to the environmental basic data, combine geographical obstacle information, and analyze the physical obstacles encountered by the UAV outside the preset target through three-dimensional modeling technology to generate an obstacle distribution map; Based on the obstacle distribution map and UAV performance parameters, combine the changes in UAV performance under different meteorological conditions to comprehensively evaluate the UAV's ability to perform tasks under various conditions and generate a UAV performance evaluation report; Integrate real-time traffic flow data and temporary control area information, combine the data of the air traffic management system, comprehensively analyze the airspace usage situation, generate an airspace usage situation analysis result, and establish a dynamic update mechanism to cope with changes in meteorological conditions, ground traffic flow, air traffic conditions, temporary control area settings, and the UAV's own state during flight; Combine the environmental basic data, obstacle distribution map, UAV performance evaluation report, and airspace usage situation analysis result, and construct a comprehensive environment model through a multi-source data fusion algorithm.

4. The method according to claim 1, characterized in that, Use a distributed fiber optic sensor network deployed in a preset terrain to synchronously collect vibration signals caused by external events to obtain time series vibration data, including: Use a distributed fiber optic sensor network deployed in a preset terrain to synchronously collect vibration signals caused by external events, adopt optical time domain reflectometry technology combined with frequency division multiplexing and wavelength division multiplexing technologies to ensure the time and spatial resolution of the vibration signals, and introduce edge computing nodes for real-time data preprocessing to obtain initial time series vibration data; Through the adaptive data preprocessing module built in the distributed fiber optic sensor network, applying a noise filtering algorithm driven by deep learning, intelligently cleaning and formatting the initial time series vibration data, identifying and removing background noise, and retaining vibration signals under preset conditions to generate target time series vibration data; According to the target time series vibration data, applying a multi-scale adaptive filtering algorithm to separate natural background noise from potential threat event signals, capturing the characteristics of vibration signals at different time scales, enhancing the distinguishability of potential threat event signals, obtaining enhanced time series vibration data, and generating feature vectors for subsequent analysis; Using the enhanced time series vibration data and feature vectors, combined with the geographic information system and the global positioning system, mapping the vibration signals to specific geographical locations, determining the real-time location of the vibration occurrence and the distribution pattern of the vibration occurrence through spatio-temporal correlation analysis, and establishing a spatio-temporal database of vibration signals to generate optimal time series vibration data.

5. The method according to claim 1, wherein Based on the optimal flight path planning for threat event trigger information, collecting target image data obtained by the unmanned aerial vehicle and three-dimensional terrain data collected by the airborne lidar, fusing the target image data and the three-dimensional terrain data to generate a panoramic view of the real environment, including: Based on the optimal flight path planning for threat event trigger information, capturing images of the target area and collecting three-dimensional terrain data to obtain target image data and three-dimensional terrain data; Using the target image data obtained by the unmanned aerial vehicle, combined with the three-dimensional terrain data collected by the airborne lidar, performing preliminary processing through an edge computing node to obtain optimized target image data and optimized three-dimensional terrain data; According to the optimized target image data and the optimized three-dimensional terrain data, applying a multi-modal data fusion algorithm to align and fuse the optimized target image data and the optimized three-dimensional terrain data to obtain fused data, and introducing a time synchronization mechanism to ensure the docking of the fused data in the spatio-temporal dimension to generate an intermediate view of the real environment; Integrating the geographic information system and semantic segmentation technology, parsing the intermediate view of the real environment, identifying and annotating the key entities in the intermediate view of the real environment to generate a panoramic view of the real environment.

6. A distributed optical fiber vibration monitoring method and system integrating UAV image recognition, characterized in that Including: An acquisition module for synchronously acquiring vibration signals caused by external events using a distributed fiber optic sensor network deployed in a preset terrain to obtain time series vibration data; An analysis module for analyzing the time series vibration data using a real-time analysis module combined with an adaptive threshold algorithm and a machine learning-driven abnormal pattern detection mechanism to distinguish natural background noise from potential threat events and generate threat event trigger information; An integration module for integrating dynamic meteorological data, geographical obstacle information, and unmanned aerial vehicle performance parameters by the cloud intelligent unmanned aerial vehicle scheduling system according to the threat event trigger information, and generating an optimal flight path planning for the threat event trigger information through a reinforcement learning algorithm; A fusion module, configured to collect target image data obtained by a drone and three-dimensional terrain data collected by an airborne lidar based on the optimal flight path planning for threat event trigger information, and fuse the target image data and the three-dimensional terrain data to generate a panoramic view of the real environment; An identification module, configured to use a 5G communication network in combination with differential privacy protection technology to transmit the panoramic view of the real environment to a central server, analyze the content of the panoramic view of the real environment based on a spatio-temporal context understanding model, identify the specific entity that causes the vibration signal and predict the behavior trend of the specific entity, establish a deep association rule between the specific entity and the historical vibration signal pattern, so as to evaluate the nature of the event and the influence scope of the nature of the event, calculate the event priority, and generate an operation guide including visualization suggestions.

7. A computing device, characterized in that, It includes a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement a distributed optical fiber vibration monitoring method integrating drone image recognition according to any one of claims 1 to 5.

8. A computer storage medium, characterized in that, A computer program is stored, and when the computer program is executed by a computer, it implements a distributed optical fiber vibration monitoring method integrating drone image recognition according to any one of claims 1 to 5.

Images