Intelligent geological disaster monitoring method and system based on cloud-edge collaboration

Abstract

The present invention relates to the field of intelligent monitoring technology, and specifically to a method and system for intelligent monitoring of geological disasters based on cloud-edge collaboration. The system includes a central intelligent agent and an edge intelligent agent that are communicatively connected to each other. After setting a disaster monitoring task, the central intelligent agent obtains the disaster monitoring task, for example, by monitoring floods, thereby realizing real-time cloud-edge collaborative monitoring of geological disasters such as mountain torrents, mudslides, and reservoir landslides. The disaster monitoring task is automatically decomposed to obtain multiple edge subtasks and central subtasks. The edge subtasks are executed by the edge intelligent agent, and the central subtasks are executed by the central intelligent agent, realizing cloud-edge collaboration and division of labor and cooperation. Each edge intelligent agent can operate independently and collaborate through the network, realizing a control mode that is both decentralized and autonomous and centralized and collaborative, thereby improving robustness and real-time performance. Moreover, by setting multiple edge intelligent agents, the problem of insufficient spatial coverage can be solved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent monitoring technology, and in particular to a method and system for intelligent monitoring of geological disasters based on cloud-edge collaboration. Background Art

[0002] Climate change has led to more frequent extreme weather events, exacerbating the risk of geological disasters such as floods and landslides. At the same time, the aging of water conservancy infrastructure such as dams, rising maintenance costs, and the combination of personnel shortages and extreme weather conditions make it difficult for traditional monitoring methods to provide timely and effective early warnings. In reality, reservoir breaches, urban waterlogging, and landslides continue to occur, causing significant casualties and property losses. Effectively and accurately monitoring and alerting these geological hazards, thereby reducing disaster losses, is an urgent research direction. Existing geological environmental disaster monitoring relies on fixed sites and manual inspections, which suffer from deficiencies such as insufficient spatial coverage, data lags, and delayed human responses. Summary of the Invention

[0003] The main purpose of this invention is to provide a method and system for intelligent monitoring of geological disasters based on cloud-edge collaboration, aiming to solve the problems that existing geological environmental disaster monitoring relies on fixed sites and manual inspections, and has defects such as insufficient spatial coverage, data lag and manual response delay.

[0004] The technical solution proposed by the present invention is:

[0005] A method for intelligent geological disaster monitoring based on cloud-edge collaboration is applied to an intelligent geological disaster monitoring system based on cloud-edge collaboration; the system includes a central intelligent agent and an edge intelligent agent that are communicatively connected to each other; the number of the edge intelligent agents is multiple; the method includes:

[0006] The central agent obtains disaster monitoring tasks for the monitoring area;

[0007] The central agent decomposes the disaster monitoring task to obtain edge subtasks and central subtasks, and sends the edge subtasks to the corresponding edge agents;

[0008] The edge agent executes the corresponding edge subtask to obtain the corresponding edge result information and sends it to the central agent;

[0009] The central agent executes the corresponding central subtask based on the edge result information to obtain the corresponding central result information;

[0010] The central agent generates a disaster risk level corresponding to the disaster monitoring task based on the central result information, where the disaster risk level is any one of high risk, medium risk and low risk;

[0011] When the disaster risk level is high, the edge agent determines the dangerous areas in the monitoring area and monitors in real time whether there are people staying in the dangerous areas;

[0012] When there are people staying in the dangerous area, the central agent will issue risk warnings to the people through the edge agents.

[0013] Preferably, the edge agent includes a first agent in communication with a meteorological sensor, a second agent in communication with a first camera, a third agent in communication with a water level meter, and a fourth agent in communication with a flow meter; the central agent decomposes the disaster monitoring task to obtain edge subtasks and central subtasks, and sends the edge subtasks to corresponding edge agents, including:

[0014] When the disaster monitoring task is flood monitoring, the central agent decomposes the disaster monitoring task based on a predetermined task decomposition template to obtain edge subtasks and central subtasks. The edge subtasks include: obtaining upstream rainfall data, obtaining watershed image data, obtaining water level data, and obtaining flow velocity data; the central subtasks include: obtaining estimated future rainfall data and building a flood risk prediction model;

[0015] The central agent sends the edge subtask of obtaining upstream rainfall data to the first agent;

[0016] The central agent sends the edge subtask of obtaining watershed image data to the second agent;

[0017] The central agent sends the edge subtask of obtaining water level data to the third agent;

[0018] The central agent sends the edge subtask of obtaining flow rate data to the fourth agent.

[0019] Preferably, the first agent is set in the upstream basin of the monitoring area; the second agent, the third agent and the fourth agent are all set in the monitoring area; the number of the first agents is multiple; the edge agent executes the corresponding edge subtask to obtain the corresponding edge result information and sends it to the central agent, including:

[0020] The first agent obtains the rainfall amount within a first preset period of time collected by the meteorological sensor, and each first agent sends the obtained rainfall amount to the first agent closest to the central agent;

[0021] The first agent closest to the central agent averages the rainfall in the past first preset time period obtained by each first agent to obtain the average rainfall in the upstream basin of the monitoring area in the past first preset time period, and sends it to the central agent as edge result information;

[0022] The second agent obtains the watershed images of the monitoring area captured by the first camera at different shooting times within the first preset time period in the past, and performs image analysis on the watershed images to obtain the actual water surface area ratio of the monitoring area at different shooting times within the first preset time period in the past, and sends the obtained values ​​as edge result information to the central agent, wherein the interval between the shooting times is the second preset time period;

[0023] The third agent obtains the actual water level values ​​of the monitoring area collected by the water level meter at different sampling times in the past first preset time period, and sends them to the central agent as edge result information, wherein the interval between the sampling times of the water level meter is the second preset time period;

[0024] The fourth intelligent agent obtains the actual flow rate values ​​of the monitoring area collected by the flow meter at different sampling times within the first preset time period in the past, and sends them to the central intelligent agent as edge result information, wherein the interval duration of the sampling time of the flow meter is the second preset time period.

[0025] Preferably, the central agent performs the corresponding central subtask based on the edge result information to obtain the corresponding central result information, including:

[0026] The central agent calls the meteorological API interface through the Internet to obtain the estimated rainfall in the upstream area of ​​the monitoring area for the first preset time in the future, and uses it as the central result information;

[0027] The central agent constructs a flood risk prediction model and trains it using historical data, where the historical data includes, as input parameters, historical upstream rainfall over a first preset period of time, historical water surface area percentages at each sampling time over the first preset period of time, historical water level values ​​at each sampling time over the first preset period of time, and historical flow velocity values ​​at each sampling time over the first preset period of time, and as an output parameter, the probability of a flood disaster.

[0028] The central intelligent agent obtains the central result information corresponding to the disaster monitoring task based on the trained flood risk prediction model and the edge result information.

[0029] Preferably, the central agent obtains central result information corresponding to the disaster monitoring task based on the trained flood risk prediction model and each edge result information, including:

[0030] The central intelligent agent inputs the average rainfall of the upstream basin of the monitoring area in the past first preset time period, the actual water surface area ratio of the monitoring area at different shooting times in the past first preset time period, the actual water level value of the monitoring area at different sampling times in the past first preset time period, and the actual flow velocity value of the monitoring area at different sampling times in the past first preset time period into the trained flood risk prediction model to obtain the output probability of flood disasters in the monitoring area, and uses the output probability of flood disasters in the monitoring area as the central result information;

[0031] The central agent generates a disaster risk level corresponding to the disaster monitoring task based on the central result information, including:

[0032] The probability of flood disasters occurring in the monitoring area due to the central intelligent agent is , the estimated rainfall in the upstream area of ​​the monitoring area in the first preset time period in the future is The average rainfall in the upstream basin of the monitoring area in the past first preset period is ;

[0033] When the central result information meets any one of the first condition, the second condition, or the third condition, the central agent generates a disaster risk level, and the disaster risk level is high risk, wherein the first condition is: , the second condition is: , the third condition is: ,and ;

[0034] When the central result information meets the fourth or fifth condition, the central agent generates a disaster risk level, and the disaster risk level is medium risk, where the fourth condition is: ,and , the fifth condition is: ,and ;

[0035] When the central result information meets the sixth or seventh condition, the central agent generates a disaster risk level, and the disaster risk level is low risk, where the sixth condition is: , the seventh condition is: ,and .

[0036] Preferably, the edge agent includes a patrol drone; the patrol drone is provided with a second camera and a controller; when the disaster risk level is high, the edge agent is used to determine the dangerous area of ​​the monitoring area and monitor in real time whether there are people staying in the dangerous area, including:

[0037] When the disaster risk level is high, the central agent sends patrol instructions to the patrol drone;

[0038] The inspection drone flies to the monitoring area based on the inspection command, obtains an overhead image of the monitoring area taken by the second camera, and marks it as a target image;

[0039] The controller determines a first dividing line and a second dividing line in the target image, and determines a dangerous area of ​​the monitoring area based on the first dividing line and the second dividing line, wherein the dangerous area is an area between the first dividing line and the second dividing line, the first dividing line is parallel to one edge of the river channel, the second dividing line is parallel to the other edge of the river channel, and the distance between the first dividing line and the one edge of the river channel, and the distance between the second dividing line and the other edge of the river channel are both first preset distance values;

[0040] The controller performs image recognition on the target image to determine whether there is a lingering person in the target image.

[0041] Preferably, the inspection drone further includes an audible and visual alarm; when there are people staying in the dangerous area, the central agent issues a risk warning to the people through the edge agent, including:

[0042] When there are people staying in the dangerous area, the controller performs image recognition on the target image to obtain the number of people staying in the target image;

[0043] When there are multiple people staying, the controller determines the real-time position of each person staying in the target image, where the real-time position of the i-th person staying in the target image is ;

[0044] The controller calculates the following formula:

[0045] ,

[0046] ,

[0047] Where, The variance of the X-axis coordinate values ​​of the real-time positions of all the people staying in the target image can reflect the degree of dispersion of the real-time positions of each person staying in the target image on the X-axis. is the variance of the Y-axis coordinate values ​​of the real-time positions of all the staying persons in the target image, which can reflect the degree of dispersion of the real-time positions of each staying person in the target image on the Y-axis; M is the total number of staying persons;

[0048] when Greater than or equal to the preset variance value, or When the variance is greater than or equal to the preset value, the controller determines that the lingering personnel are in a dispersed state;

[0049] When there are multiple people staying, and the multiple people staying are in a dispersed state, the controller obtains multiple overhead images of the monitoring area taken by the second camera after the target image, and marks them as subsequent images, wherein the interval between the subsequent images is a third preset time length;

[0050] The controller obtains the real-time position of each person staying in the subsequent image, and determines the warning movement route of the drone based on the real-time position of each person staying in the target image and the real-time position of each person staying in the subsequent image;

[0051] When the number of lingering persons is one, or the number of lingering persons is more than one but the multiple lingering persons are not in a dispersed state, the controller controls the patrol drone to fly towards the lingering persons and activates the sound and light alarm during the flight.

[0052] Preferably, the controller obtains the real-time position of each lingering person in the subsequent image, and determines the warning movement route of the drone based on the real-time position of each lingering person in the target image and the real-time position of each lingering person in the subsequent image, including:

[0053] The controller marks the lingering persons whose distance from each other is less than a second preset value in two adjacent subsequent images as the same lingering person;

[0054] The controller obtains the real-time location of the i-th person in the k-th subsequent image , and calculate the average moving speed of the i-th person staying :

[0055] ,

[0056] Where, The third preset duration; , N is the total number of subsequent images (e.g. 20);

[0057] The controller obtains the geometric center point of the real-time position of the i-th lingering person in all subsequent images except the earliest subsequent image and marks it as the average center point;

[0058] The controller uses the average center point of the i-th lingering person and the straight line on which the position points in the earliest subsequent image lie as the movement trend line corresponding to the i-th lingering person;

[0059] The controller determines the distance from the position of the i-th person staying in the latest subsequent image to the danger zone along the corresponding moving trend line , and calculate the estimated moving time of the i-th person from the position in the latest subsequent image to the dangerous area :

[0060] ,

[0061] The controller controls the patrol drones to move over the lingering personnel in order of estimated moving time from small to large to issue an alarm.

[0062] Preferably, the system further comprises a monitoring terminal communicatively connected to the central intelligent agent; the central intelligent agent generates a disaster risk level corresponding to the disaster monitoring task based on the central result information, and then further comprises:

[0063] When the disaster risk level is high or medium, the central intelligent agent sends the disaster risk level corresponding to the disaster monitoring task to the monitoring terminal for display.

[0064] The present invention also proposes a geological disaster intelligent monitoring system based on cloud-edge collaboration, which applies a geological disaster intelligent monitoring method based on cloud-edge collaboration; the system includes a central intelligent agent and an edge intelligent agent that are communicatively connected to each other; the number of the edge intelligent agents is multiple, and they are set in the monitoring area.

[0065] The above technical solution can achieve the following beneficial effects:

[0066] The intelligent geological disaster monitoring method based on cloud-edge collaboration proposed in the present invention can solve the problems that the existing geological environment disaster monitoring relies on fixed sites and manual inspections, and has defects such as insufficient spatial coverage, data lag and manual response delay; the system includes a central intelligent agent and an edge intelligent agent that are connected to each other for communication, and the number of edge intelligent agents is multiple; after setting the disaster monitoring task, the central intelligent agent obtains the disaster monitoring task, and automatically decomposes the disaster monitoring task to obtain multiple edge subtasks and central subtasks, and sends the edge subtasks (usually monitoring or data collection tasks) to the corresponding edge intelligent agent for execution by the edge intelligent agent, and the central subtasks (usually data analysis and early warning feedback tasks) are executed by the central intelligent agent, realizing cloud-edge collaboration and division of labor and cooperation. Each edge intelligent agent can operate independently and collaborate through the network, realizing both decentralized autonomy and centralized collaboration. The control mode improves the robustness and real-time performance; and by setting up multiple edge agents, it can solve the problem of insufficient spatial coverage; compared with manual inspections, edge agents can interact with central agents for data in real time to solve the defects of data lag and manual response delay; the central agent can execute the corresponding central subtask based on the edge result information to obtain the corresponding central result information, and then generate the disaster risk level corresponding to the disaster monitoring task based on the central result information; thereby clearly and quickly knowing the geological risk situation of the monitoring area, so as to facilitate timely response; in addition, the solution of this application can also respond to high-risk geological disasters. When the disaster risk level is high, it means that personnel evacuation is required, the dangerous area of ​​the monitoring area is determined, and real-time monitoring is carried out to determine whether there are people staying in the dangerous area, and to warn the people staying there to reduce the occurrence of disaster accidents. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the structures shown in these drawings without paying any creative work.

[0068] Figure 1 This is a flowchart of the first embodiment of a method for intelligent monitoring of geological disasters based on cloud-edge collaboration proposed by the present invention;

[0069] Figure 2 This is a schematic diagram of the target image in the sixth embodiment of the intelligent monitoring method for geological disasters based on cloud-edge collaboration proposed by the present invention. DETAILED DESCRIPTION

[0070] It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0071] The present invention proposes a method and system for intelligent monitoring of geological disasters based on cloud-edge collaboration.

[0072] As attached Figure 1 As shown, in a first embodiment of a method for intelligent monitoring of geological disasters based on cloud-edge collaboration proposed by the present invention, the method is applied to an intelligent monitoring system for geological disasters based on cloud-edge collaboration; the system includes a central intelligent agent (such as a cloud server) and an edge intelligent agent (an intelligent device for performing various monitoring or data collection tasks) that are communicatively connected to each other; the number of the edge intelligent agents is multiple; this embodiment includes the following steps:

[0073] Step S110: The central agent obtains disaster monitoring tasks for the monitoring area.

[0074] Specifically, in this embodiment, the system also includes a monitoring terminal (used by management personnel) communicatively connected to the central agent. The monitoring terminal receives manually inputted disaster monitoring tasks for the monitored area and transmits these tasks to the central agent. These disaster monitoring tasks include, but are not limited to, geological disaster monitoring such as flood monitoring, debris flow monitoring, and landslide monitoring.

[0075] Step S120: The central agent decomposes the disaster monitoring task to obtain edge subtasks and central subtasks, and sends the edge subtasks to the corresponding edge agents.

[0076] Specifically, after receiving the disaster monitoring task, the central intelligent agent can automatically decompose the disaster monitoring task to obtain multiple edge subtasks and central subtasks, and send the edge subtasks (usually monitoring or data collection tasks) to the corresponding edge intelligent agent; that is, the edge subtasks are executed by the edge intelligent agent, and the central subtasks (usually data analysis and early warning feedback tasks) are executed by the central intelligent agent, realizing cloud-edge collaboration and division of labor and cooperation.

[0077] Step S130: The edge agent executes the corresponding edge subtask to obtain the corresponding edge result information and sends it to the central agent.

[0078] Step S140: The central agent executes the corresponding central subtask based on the edge result information to obtain the corresponding central result information.

[0079] Specifically, when the central agent performs central subtasks, it needs to complete them based on the edge result information collected by the edge agent. For example, based on the movement trend data of mountain rocks collected by the edge agent, the risk probability of landslide can be calculated.

[0080] Step S150: The central intelligent agent generates a disaster risk level corresponding to the disaster monitoring task based on the central result information, where the disaster risk level is any one of high risk, medium risk and low risk.

[0081] Step S160: When the disaster risk level is high, the edge agent determines the dangerous area of ​​the monitoring area and monitors in real time whether there are people staying in the dangerous area.

[0082] Specifically, when the disaster risk level is high, it means that people need to be evacuated to prevent safety accidents. To this end, it is necessary to determine the dangerous areas in the monitoring area, monitor in real time whether there are people staying in the dangerous areas, and warn the people staying there to reduce the occurrence of disaster accidents.

[0083] Step S170: When there are people staying in the dangerous area, the central agent issues a risk warning to the people through the edge agents.

[0084] The intelligent geological disaster monitoring method based on cloud-edge collaboration proposed in the present invention can solve the problems that the existing geological environment disaster monitoring relies on fixed sites and manual inspections, and has defects such as insufficient spatial coverage, data lag and manual response delay; the system includes a central intelligent agent and an edge intelligent agent that are connected to each other for communication, and the number of edge intelligent agents is multiple; after setting the disaster monitoring task, the central intelligent agent obtains the disaster monitoring task, and automatically decomposes the disaster monitoring task to obtain multiple edge subtasks and central subtasks, and sends the edge subtasks (usually monitoring or data collection tasks) to the corresponding edge intelligent agent for execution by the edge intelligent agent, and the central subtasks (usually data analysis and early warning feedback tasks) are executed by the central intelligent agent, realizing cloud-edge collaboration and division of labor and cooperation. Each edge intelligent agent can operate independently and collaborate through the network, realizing both decentralized autonomy and centralized collaboration. The control mode improves the robustness and real-time performance; and by setting up multiple edge agents, it can solve the problem of insufficient spatial coverage; compared with manual inspections, edge agents can interact with central agents for data in real time to solve the defects of data lag and manual response delay; the central agent can execute the corresponding central subtask based on the edge result information to obtain the corresponding central result information, and then generate the disaster risk level corresponding to the disaster monitoring task based on the central result information; thereby clearly and quickly knowing the geological risk situation of the monitoring area, so as to facilitate timely response; in addition, the solution of this application can also respond to high-risk geological disasters. When the disaster risk level is high, it means that personnel evacuation is required, the dangerous area of ​​the monitoring area is determined, and real-time monitoring is carried out to determine whether there are people staying in the dangerous area, and to warn the people staying there to reduce the occurrence of disaster accidents.

[0085] Furthermore, the solution proposed in this application utilizes layered collaboration and edge computing. The system employs a three-tiered architecture: cloud (central agent), edge (edge ​​agent), and end (monitoring terminal), incorporating edge computing. At the monitoring frontline, resource-constrained, lightweight agents (edge ​​agents) are responsible for data collection and preliminary analysis. On central servers or in the cloud, high-performance agents (central agents) perform in-depth analysis and overall assessment. This layered design is driven by considerations of network bandwidth and response time. The inclusion of edge computing units allows for local filtering and processing of massive amounts of sensor data, reducing transmission pressure and cloud load. This "cloud-edge collaboration" concept has demonstrated advantages in applications such as water quality monitoring: edge nodes can analyze sensor data and provide preliminary responses in real time, while complex comprehensive analysis is performed in the cloud. For example, in a multi-agent water quality monitoring network, a local node detecting an excessive level of pollutants can immediately trigger action and simultaneously upload the data to the cloud for comprehensive evaluation. Layered collaboration ensures that the system can achieve both real-time local response and comprehensive global decision-making.

[0086] Furthermore, the solution proposed in this application enables autonomous reasoning and learning: each intelligent agent (edge ​​agent) is not only a data collection agent but also incorporates a built-in rule-based expert system, machine learning, and trend prediction capabilities. For example, starting with threshold-based alarm triggering, the system gradually learns mechanisms, enabling the agent to continuously optimize its decision-making strategy as historical data accumulates. The application of advanced algorithms enables the multi-agent system to continuously improve its decision-making based on experience. For example, through reinforcement learning, algorithms such as landslide monitoring, edge agents can gradually calibrate trigger thresholds to reduce false alarms and missed alerts. Autonomous reasoning and learning make the system evolvable, enabling it to adapt to changes in the environment and data characteristics, thereby improving the effectiveness of long-term monitoring.

[0087] Furthermore, this solution emphasizes user-friendly human-computer interaction and explainable decision-making. Ultimately, monitoring and early warning systems serve people, and experts and managers must be able to easily access system information and understand the conclusions drawn by the intelligent agent. Therefore, we implemented interactive interfaces and explanation modules within the monitoring terminal. These terminals provide a visual dashboard and a natural language question-and-answer interface, allowing users to inquire about monitoring status and warning reasons, much like they would with an assistant. For example, a question-and-answer system based on a knowledge graph and large models allows non-expert users to ask questions such as, "What is the current water level in a certain reservoir and the probability of exceeding the warning level?" and receive answers from the intelligent agent based on data and knowledge. The monitoring terminal combines a knowledge base with natural language processing to implement intelligent question-and-answer and explanation capabilities, making the reasons for warnings transparent and traceable, thereby enhancing user trust.

[0088] In summary, this solution achieves both wide-area monitoring and local response through multi-agent collaboration. It combines edge computing to ensure real-time performance, leverages knowledge-based learning to enhance intelligence, and utilizes human-computer interaction to improve usability. This solution aims to build an autonomous, efficient, distributed, coordinated, knowledge-driven, secure, and reliable geological disaster monitoring and early warning system. The following sections will delve into key technical details by functional module and provide a feasible implementation path.

[0089] In a second embodiment of a method for intelligent monitoring of geological disasters based on cloud-edge collaboration proposed by the present invention, based on the first embodiment, the edge agent includes a first agent communicatively connected to a meteorological sensor, a second agent communicatively connected to a first camera, a third agent communicatively connected to a water level meter, and a fourth agent communicatively connected to a flow meter; step S120 includes the following steps:

[0090] Step S210: When the disaster monitoring task is flood monitoring, the central intelligent agent decomposes the disaster monitoring task based on a predetermined task decomposition template to obtain edge subtasks and central subtasks, wherein the edge subtasks obtained include: obtaining upstream rainfall data, obtaining watershed image data, obtaining water level data, and obtaining flow velocity data; the central subtasks include: obtaining estimated future rainfall data, and constructing a flood risk prediction model.

[0091] Specifically, task decomposition templates can be manually preset or automatically decomposed using rules and large models, enabling the agent to flexibly respond to different situations. For example, a central agent driven by a large language model can decompose disaster monitoring tasks.

[0092] Step S220: The central agent sends the edge subtask of obtaining upstream rainfall data to the first agent.

[0093] Step S230: The central agent sends the edge subtask of acquiring watershed image data to the second agent.

[0094] Step S240: The central agent sends the edge subtask of obtaining water level data to the third agent.

[0095] Step S250: The central agent sends the edge subtask of acquiring flow rate data to the fourth agent.

[0096] Specifically, this embodiment provides a specific solution for task decomposition and allocation: a central agent determines which agents are responsible for executing each subtask. In a multi-agent collaborative environment, agents with different roles can be assigned to different tasks: for example, edge agents can be responsible for data collection subtasks, while central agents can be responsible for analysis subtasks. This creates an instructor-executor model, with an "instructor" agent assigning tasks and an "assistant" agent executing and providing feedback.

[0097] Specifically, in this embodiment, the disaster monitoring task scenario is set to flood monitoring. Floods can easily trigger geological disasters such as mountain torrents, mudslides, and reservoir landslides. Therefore, in this embodiment, by monitoring floods, cloud-edge collaborative real-time monitoring of geological disasters is achieved.

[0098] In a third embodiment of a method for intelligent geological disaster monitoring based on cloud-edge collaboration proposed by the present invention, based on the second embodiment, the first intelligent agent is set in the upstream basin of the monitoring area; the second intelligent agent, the third intelligent agent, and the fourth intelligent agent are all set in the monitoring area; there are multiple first intelligent agents; step S130 includes the following steps:

[0099] Step S310: The first agents obtain the rainfall within a first preset time period (e.g., 12 hours) collected by the meteorological sensor, and each first agent sends the obtained rainfall to the first agent closest to the central agent.

[0100] Step S320: The first agent closest to the central agent takes the average of the rainfall obtained by each first agent within the past first preset time period to obtain the average rainfall in the upstream basin of the monitoring area in the past first preset time period, and sends it to the central agent as edge result information.

[0101] Step S330: The second intelligent agent obtains the watershed images of the monitoring area taken by the first camera at different shooting times in the past first preset time period, and performs image analysis on the watershed images to obtain the actual water surface area ratio of the monitoring area at different shooting times in the past first preset time period, and sends it to the central intelligent agent as edge result information, wherein the interval between the shooting moments is the second preset time period (for example, 10 minutes).

[0102] Specifically, the water surface area ratio here is the ratio of the area of ​​the water surface area in the basin image to the total area of ​​the entire image. The larger the water surface area ratio, the higher the water level of the river in the basin and the higher the risk of flood disasters.

[0103] Step S340: The third intelligent agent obtains the actual water level values ​​of the monitoring area collected by the water level meter at different sampling times in the past first preset time period, and sends them to the central intelligent agent as edge result information, wherein the interval duration of the sampling time of the water level meter is the second preset time period (for example, 10 minutes).

[0104] Specifically, the water level value can directly reflect the flood disaster risk of the rivers in the monitoring area. The higher the water level value, the higher the risk of flood disaster.

[0105] Step S350: The fourth intelligent agent obtains the actual flow rate values ​​of the monitoring area collected by the flow meter at different sampling times within the first preset time period in the past, and sends them to the central intelligent agent as edge result information, wherein the interval duration of the sampling time of the flow meter is the second preset time period.

[0106] Specifically, the river flow velocity value combined with the water level value can reflect the flood disaster risk of the river in the monitoring area. The higher the water level value, the slower the flow velocity value, which means that the river water discharges slower and the risk of flood disaster is higher.

[0107] Specifically, this embodiment provides a specific plan for each edge agent to execute the corresponding edge subtask, which reflects the execution and collaboration of this system: each edge agent executes the subtask in parallel or sequentially. During the execution process, the edge agent and the central agent can also exchange information and intermediate results through a message mechanism, and realize multiple rounds of interaction to gradually approach the task goal. For example, after the edge agent completes rainfall collection, it sends the rainfall data to the central agent to continue the prediction calculation. If the edge result information of a certain edge subtask is uncertain, it triggers further refinement or repeated execution until the conditions are met. Through multiple rounds of iterative interaction, this system ultimately generates high-quality comprehensive results.

[0108] In a fourth embodiment of a method for intelligent geological disaster monitoring based on cloud-edge collaboration proposed by the present invention, based on the third embodiment, step S140 includes the following steps:

[0109] Step S410: The central intelligent agent calls the meteorological API interface through the Internet to obtain the estimated rainfall in the upstream area of ​​the monitoring area within the first preset time period in the future, and uses it as the central result information.

[0110] Specifically, the central agent has the ability to call external APIs to obtain data (for example, meteorological data in this example). This includes, but is not limited to, obtaining weather, hydrological, geological, and other related information, expanding the system's cognitive scope. Through standardized API calls, the central agent can regularly pull the latest information from authoritative data sources or query it on demand when needed, providing a basis for autonomous decision-making.

[0111] During implementation, different external data can be retrieved based on different disaster monitoring tasks, such as the basin management agency's hydrological station data API and remote sensing imagery service API. The central agent supports multiple interface protocols (RESTful, WebSocket, etc.) and asynchronous data acquisition mechanisms, allowing for reliable and efficient integration of external data. Specific implementation paths: The central agent supports common open data interface formats and has authentication and frequency control mechanisms. For example, meteorological services can be used to obtain future rainfall forecasts, and remote sensing imagery APIs can be used to obtain the latest satellite rainfall or surface deformation data. The central agent can also periodically call similar interfaces to obtain the latest observation data. For example, by accessing real-time data from various water conservancy stations, the latest water level values ​​at all stations can be obtained at regular intervals and cached locally.

[0112] In addition, in order to cope with possible service interruptions or throttling issues with external APIs, the central agent has a fault-tolerant and caching strategy: when the API is unavailable, the most recently cached data is used, and it is automatically updated after the service is restored. Retry and degradation mechanisms are added, and the request frequency is optimized. For example, multiple site data requests are consolidated into a single batch call to reduce the frequency. Finally, metadata configuration is used to flexibly add and delete communication connections with different external APIs, making it easy to expand to new data sources. This iterative configurable design ensures that the agent can continuously integrate the latest data services and maintain comprehensive access to external information under existing technical conditions.

[0113] Step S420: The central intelligent agent constructs a flood risk prediction model and trains it through historical data, wherein the historical data includes the historical upstream rainfall in the past first preset time period as input parameters, the historical water surface area ratio value at each shooting time in the past first preset time period, the historical water level value at each sampling time in the past first preset time period, and the historical flow velocity value at each sampling time in the past first preset time period, as well as the probability of flood disaster occurrence used as an output parameter.

[0114] Specifically, the flood risk prediction model in this embodiment adopts the CNN-LSTM-Attention model, which can output the probability of flood disasters occurring in the basin based on upstream rainfall, river water surface area ratio, water level value and flow rate value.

[0115] Step S430: The central intelligent agent obtains the central result information corresponding to the disaster monitoring task based on the trained flood risk prediction model and the edge result information.

[0116] In a fifth embodiment of a method for intelligent monitoring of geological disasters based on cloud-edge collaboration proposed by the present invention, based on the fourth embodiment, step S430 includes the following steps:

[0117] Step S510: The central intelligent body inputs the average rainfall of the upstream basin of the monitoring area in the past first preset time period, the actual water surface area ratio of the monitoring area at different shooting times in the past first preset time period, the actual water level value of the monitoring area at different sampling times in the past first preset time period, and the actual flow rate value of the monitoring area at different sampling times in the past first preset time period into the trained flood risk prediction model to obtain the output probability of flood disasters in the monitoring area, and uses the output probability of flood disasters in the monitoring area as the central result information.

[0118] Specifically, by completing the trained flood risk prediction model and the real-time data collected by each edge intelligent agent (the average rainfall in the upstream basin of the monitoring area in the past first preset time, the actual water surface area ratio of the monitoring area at different shooting times in the past first preset time, the actual water level value of the monitoring area at different sampling times in the past first preset time, and the actual flow rate value of the monitoring area at different sampling times in the past first preset time), the probability of flood disasters in the monitoring area can be obtained and used as the central result information.

[0119] Step S150 includes the following steps:

[0120] Step S520: The central agent determines the probability of flood disaster in the monitoring area to be , the estimated rainfall in the upstream area of ​​the monitoring area in the first preset time period in the future is The average rainfall in the upstream basin of the monitoring area in the past first preset period is .

[0121] Step S530: When the central result information satisfies any one of the first condition, the second condition, or the third condition, the central agent generates a disaster risk level, and the disaster risk level is high risk, wherein the first condition is: , the second condition is: , the third condition is: ,and .

[0122] Step S540: When the central result information satisfies the fourth condition or the fifth condition, the central agent generates a disaster risk level, and the disaster risk level is medium risk, wherein the fourth condition is: ,and , the fifth condition is: ,and .

[0123] Step S550: When the central result information satisfies the sixth or seventh condition, the central agent generates a disaster risk level, and the disaster risk level is low risk, wherein the sixth condition is: , the seventh condition is: ,and .

[0124] Specifically, in this embodiment, the disaster risk level of the monitoring area (i.e., the risk level of flood disaster) is determined by comprehensively considering the probability of flood disasters occurring in the monitoring area and the estimated rainfall in the upstream area of ​​the monitoring area within the first preset time period in the future.

[0125] In a sixth embodiment of a method for intelligent monitoring of geological disasters based on cloud-edge collaboration proposed by the present invention, based on the first embodiment, the edge agent includes a patrol drone; the patrol drone is provided with a second camera and a controller; step S160 includes the following steps:

[0126] Step S610: When the disaster risk level is high, the central intelligent agent sends a patrol instruction to the patrol drone.

[0127] Step S620: The inspection drone flies to the monitoring area based on the inspection instruction, and obtains an overhead image of the monitoring area taken by the second camera, and marks it as a target image.

[0128] Step S630: The controller determines a first dividing line and a second dividing line in the target image, and determines a dangerous area of ​​the monitoring area based on the first dividing line and the second dividing line, wherein the dangerous area is the area between the first dividing line and the second dividing line, the first dividing line is parallel to one edge of the river channel, the second dividing line is parallel to the other edge of the river channel, and the distance between the first dividing line and one edge of the river channel, and the distance between the second dividing line and the other edge of the river channel are both a first preset distance value (for example, 10 meters).

[0129] Specifically, as attached Figure 2 As shown in the figure, when the disaster risk level is high, the area within 10 meters from the edge of the river is a dangerous area.

[0130] Step S640: The controller performs image recognition on the target image to determine whether there is a lingering person in the target image.

[0131] In the seventh embodiment of the intelligent geological disaster monitoring method based on cloud-edge collaboration proposed by the present invention, based on the sixth embodiment, the inspection drone further includes an audible and visual alarm; step S170 includes the following steps:

[0132] Step S710: When there are people staying in the dangerous area, the controller performs image recognition on the target image to obtain the number of people staying in the target image.

[0133] Step S720: When there are multiple people staying, the controller determines the real-time position of each person staying in the target image, wherein the real-time position of the i-th person staying in the target image is ;

[0134] The controller calculates the following formula:

[0135] ,

[0136] ,

[0137] Where, The variance of the X-axis coordinate values ​​of the real-time positions of all the people staying in the target image can reflect the degree of dispersion of the real-time positions of each person staying in the target image on the X-axis. is the variance of the Y-axis coordinate values ​​of the real-time positions of all lingering persons in the target image, which can reflect the degree of dispersion of the real-time positions of each lingering person in the target image on the Y-axis; M is the total number of lingering persons.

[0138] Step S730: When Greater than or equal to the preset variance value, or When the variance is greater than or equal to the preset value, the controller determines that the lingering personnel are in a dispersed state.

[0139] Specifically, when Greater than or equal to the preset variance value, or If the variance is greater than or equal to the preset variance value, it means that the people staying in the target image are relatively scattered. The inspection drone needs to fly over each person to issue an alarm. Therefore, the order of alarms needs to be arranged according to the dangerous situation of each person.

[0140] Step S740: When there are multiple people staying there and the multiple people staying there are in a dispersed state, the controller obtains multiple overhead images of the monitoring area after the target image taken by the second camera and marks them as subsequent images, wherein the interval between the subsequent images is a third preset time length (for example, 0.2 seconds).

[0141] Step S750: The controller obtains the real-time position of each lingering person in the subsequent image, and determines the warning movement route of the drone based on the real-time position of each lingering person in the target image and the real-time position of each lingering person in the subsequent image.

[0142] Specifically, based on the real-time position of the lingering person in the target image and the real-time position in each subsequent image, the movement direction trend and movement speed of the lingering person can be determined, thereby determining the warning movement route of the drone (the drone will preferentially fly to the lingering person who is closer to the danger zone and moves faster).

[0143] Step S760: When the number of the lingering persons is one, or the number of the lingering persons is more than one but the multiple lingering persons are not in a dispersed state, the controller controls the inspection drone to fly toward the lingering persons and activates the sound and light alarm during the flight.

[0144] Specifically, when the number of lingering persons is single, or even if the number of lingering persons is multiple but the multiple lingering persons are not in a dispersed state, the patrol drone is directly controlled to fly towards the single lingering person or the multiple lingering persons gathered, and the sound and light alarm is activated during the flight to warn the lingering person and remind him to stay away from danger.

[0145] In the eighth embodiment of the intelligent geological disaster monitoring method based on cloud-edge collaboration proposed by the present invention, based on the seventh embodiment, step S750 includes the following steps:

[0146] Step S810: The controller marks two lingering persons whose distance from each other is less than a second preset distance in two adjacent subsequent images as the same lingering person.

[0147] Specifically, since the time interval between adjacent subsequent images is short (0.2 seconds), the moving distance of the lingering person between the two adjacent subsequent images is not long (less than the second preset distance value). Therefore, the lingering persons whose distance from each other is less than the second preset distance value in the two adjacent subsequent images can be marked as the same lingering person.

[0148] Step S820: The controller obtains the real-time position of the i-th person staying in the k-th subsequent image , and calculate the average moving speed of the i-th person staying :

[0149] ,

[0150] Where, The third preset duration; , N is the total number of subsequent images.

[0151] Step S830: The controller obtains the geometric center point of the real-time position of the i-th lingering person in all subsequent images except the earliest subsequent image, and marks it as the average center point.

[0152] Step S840: The controller uses the straight line on which the average center point of the i-th lingering person and the position points in the earliest subsequent image are located as the moving trend line corresponding to the i-th lingering person.

[0153] Step S850: The controller determines the distance from the position of the i-th person staying in the latest subsequent image to the danger zone along the corresponding moving trend line. , and calculate the estimated moving time of the i-th person from the position in the latest subsequent image to the dangerous area :

[0154] ,

[0155] Step S860: The controller controls the inspection drones to move to the top of the lingering personnel in order of estimated moving time from small to large to issue an alarm.

[0156] Specifically, the estimated moving time in this embodiment is the estimated time it takes for each lingering person to move to the dangerous area. The shorter the estimated moving time, the more dangerous the corresponding lingering person is. Therefore, it is necessary to control the patrol drone to fly to the lingering person first to issue an alert.

[0157] In a ninth embodiment of a method for intelligent monitoring of geological disasters based on cloud-edge collaboration proposed by the present invention, based on the first embodiment, the system further includes a monitoring terminal communicatively connected to the central intelligent body; step S150, followed by the following steps:

[0158] Step S910: When the disaster risk level is high risk or medium risk, the central intelligent agent sends the disaster risk level corresponding to the disaster monitoring task to the monitoring terminal for display.

[0159] The present invention also proposes a geological disaster intelligent monitoring system based on cloud-edge collaboration, which applies a geological disaster intelligent monitoring method based on cloud-edge collaboration; the system includes a central intelligent agent and an edge intelligent agent that are communicatively connected to each other; the number of the edge intelligent agents is multiple, and they are set in the monitoring area.

[0160] The serial numbers of the above embodiments of the present invention are for description only and do not represent the advantages or disadvantages of the embodiments.

[0161] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present invention, ordinary technicians in this field can also make many forms without departing from the scope of protection of the present invention and the claims, all of which are protected by the present invention.

Claims

1. A method for intelligent monitoring of geological disasters based on cloud-edge collaboration, characterized in that: Applied to a geological disaster intelligent monitoring system based on cloud-edge collaboration; the system includes a central intelligent agent and an edge intelligent agent that are communicatively connected to each other; The number of the edge agents is multiple; the method includes: The central agent obtains disaster monitoring tasks for the monitoring area; The central agent decomposes the disaster monitoring task to obtain edge subtasks and central subtasks, and sends the edge subtasks to the corresponding edge agents; The edge agent executes the corresponding edge subtask to obtain the corresponding edge result information and sends it to the central agent; The central agent executes the corresponding central subtask based on the edge result information to obtain the corresponding central result information; The central agent generates a disaster risk level corresponding to the disaster monitoring task based on the central result information, where the disaster risk level is any one of high risk, medium risk and low risk; When the disaster risk level is high, the edge agent determines the dangerous areas in the monitoring area and monitors in real time whether there are people staying in the dangerous areas; When there are people staying in the dangerous area, the central agent will issue risk warnings to the people through the edge agents; The edge agent includes a first agent in communication with a meteorological sensor, a second agent in communication with a first camera, a third agent in communication with a water level meter, and a fourth agent in communication with a flow meter. The central agent decomposes the disaster monitoring task to obtain edge subtasks and central subtasks, and sends the edge subtasks to corresponding edge agents, including: When the disaster monitoring task is flood monitoring, the central agent decomposes the disaster monitoring task based on a predetermined task decomposition template to obtain edge subtasks and central subtasks. The edge subtasks include: obtaining upstream rainfall data, obtaining watershed image data, obtaining water level data, and obtaining flow velocity data; the central subtasks include: obtaining estimated future rainfall data and building a flood risk prediction model; The central agent sends the edge subtask of obtaining upstream rainfall data to the first agent; The central agent sends the edge subtask of obtaining watershed image data to the second agent; The central agent sends the edge subtask of obtaining water level data to the third agent; The central agent sends the edge subtask of obtaining flow rate data to the fourth agent.

2. The method for intelligent geological disaster monitoring based on cloud-edge collaboration according to claim 1 is characterized in that: The first agent is set in the upstream basin of the monitoring area; the second agent, the third agent and the fourth agent are all set in the monitoring area; there are multiple first agents; the edge agent executes the corresponding edge subtask to obtain corresponding edge result information and sends it to the central agent, including: The first agent obtains the rainfall amount within a first preset period of time collected by the meteorological sensor, and each first agent sends the obtained rainfall amount to the first agent closest to the central agent; The first agent closest to the central agent averages the rainfall in the past first preset time period obtained by each first agent to obtain the average rainfall in the upstream basin of the monitoring area in the past first preset time period, and sends it to the central agent as edge result information; The second agent obtains the watershed images of the monitoring area captured by the first camera at different shooting times within the first preset time period in the past, and performs image analysis on the watershed images to obtain the actual water surface area ratio of the monitoring area at different shooting times within the first preset time period in the past, and sends the obtained values ​​as edge result information to the central agent, wherein the interval between the shooting times is the second preset time period; The third agent obtains the actual water level values ​​of the monitoring area collected by the water level meter at different sampling times in the past first preset time period, and sends them to the central agent as edge result information, wherein the interval between the sampling times of the water level meter is the second preset time period; The fourth intelligent agent obtains the actual flow rate values ​​of the monitoring area collected by the flow meter at different sampling times within the first preset time period in the past, and sends them to the central intelligent agent as edge result information, wherein the interval duration of the sampling time of the flow meter is the second preset time period.

3. The method for intelligent geological disaster monitoring based on cloud-edge collaboration according to claim 2 is characterized in that: The central agent executes the corresponding central subtask based on the edge result information to obtain the corresponding central result information, including: The central agent calls the meteorological API interface through the Internet to obtain the estimated rainfall in the upstream area of ​​the monitoring area for the first preset time in the future, and uses it as the central result information; The central agent constructs a flood risk prediction model and trains it using historical data, where the historical data includes, as input parameters, historical upstream rainfall over a first preset period of time, historical water surface area percentages at each sampling time over the first preset period of time, historical water level values ​​at each sampling time over the first preset period of time, and historical flow velocity values ​​at each sampling time over the first preset period of time, and as an output parameter, the probability of a flood disaster. The central intelligent agent obtains the central result information corresponding to the disaster monitoring task based on the trained flood risk prediction model and the edge result information.

4. The method for intelligent geological disaster monitoring based on cloud-edge collaboration according to claim 3 is characterized in that: The central agent obtains central result information corresponding to the disaster monitoring task based on the trained flood risk prediction model and the edge result information, including: The central intelligent agent inputs the average rainfall of the upstream basin of the monitoring area in the past first preset time period, the actual water surface area ratio of the monitoring area at different shooting times in the past first preset time period, the actual water level value of the monitoring area at different sampling times in the past first preset time period, and the actual flow velocity value of the monitoring area at different sampling times in the past first preset time period into the trained flood risk prediction model to obtain the output probability of flood disasters in the monitoring area, and uses the output probability of flood disasters in the monitoring area as the central result information; The central agent generates a disaster risk level corresponding to the disaster monitoring task based on the central result information, including: The probability of flood disasters occurring in the monitoring area due to the central intelligent agent is , the estimated rainfall in the upstream area of ​​the monitoring area in the first preset time period in the future is The average rainfall in the upstream basin of the monitoring area in the past first preset period is ; When the central result information meets any one of the first condition, the second condition, or the third condition, the central agent generates a disaster risk level, and the disaster risk level is high risk, wherein the first condition is: , the second condition is: , the third condition is: ,and ; When the central result information meets the fourth or fifth condition, the central agent generates a disaster risk level, and the disaster risk level is medium risk, where the fourth condition is: ,and , the fifth condition is: ,and ; When the central result information meets the sixth or seventh condition, the central agent generates a disaster risk level, and the disaster risk level is low risk, where the sixth condition is: , the seventh condition is: ,and .

5. The method for intelligent geological disaster monitoring based on cloud-edge collaboration according to claim 1 is characterized in that: The edge agent includes a patrol drone; the patrol drone is provided with a second camera and a controller; When the disaster risk level is high, the edge agent determines the dangerous area of ​​the monitoring area and monitors in real time whether there are people staying in the dangerous area, including: When the disaster risk level is high, the central agent sends patrol instructions to the patrol drone; The inspection drone flies to the monitoring area based on the inspection command, obtains an overhead image of the monitoring area taken by the second camera, and marks it as a target image; The controller determines a first dividing line and a second dividing line in the target image, and determines a dangerous area of ​​the monitoring area based on the first dividing line and the second dividing line, wherein the dangerous area is an area between the first dividing line and the second dividing line, the first dividing line is parallel to one edge of the river channel, the second dividing line is parallel to the other edge of the river channel, and the distance between the first dividing line and the one edge of the river channel, and the distance between the second dividing line and the other edge of the river channel are both first preset distance values; The controller performs image recognition on the target image to determine whether there is a lingering person in the target image.

6. The method for intelligent geological disaster monitoring based on cloud-edge collaboration according to claim 5 is characterized in that: The inspection drone also includes an audible and visual alarm. When there are people staying in the dangerous area, the central agent issues a risk warning to the people through the edge agent, including: When there are people staying in the dangerous area, the controller performs image recognition on the target image to obtain the number of people staying in the target image; When there are multiple people staying, the controller determines the real-time position of each person staying in the target image, where the real-time position of the i-th person staying in the target image is ; The controller calculates the following formula: ， ， Where, is the variance of the X-axis coordinate values ​​of the real-time positions of all the people staying in the target image; is the variance of the Y-axis coordinate values ​​of the real-time positions of all the people staying in the target image; M is the total number of people staying; when Greater than or equal to the preset variance value, or When the variance is greater than or equal to the preset value, the controller determines that the lingering personnel are in a dispersed state; When there are multiple people staying, and the multiple people staying are in a dispersed state, the controller obtains multiple overhead images of the monitoring area taken by the second camera after the target image, and marks them as subsequent images, wherein the interval between the subsequent images is a third preset time length; The controller obtains the real-time position of each person staying in the subsequent image, and determines the warning movement route of the drone based on the real-time position of each person staying in the target image and the real-time position of each person staying in the subsequent image; When the number of lingering persons is one, or the number of lingering persons is more than one but the multiple lingering persons are not in a dispersed state, the controller controls the patrol drone to fly towards the lingering persons and activates the sound and light alarm during the flight.

7. The method for intelligent geological disaster monitoring based on cloud-edge collaboration according to claim 6 is characterized in that: The controller obtains the real-time position of each lingering person in the subsequent image, and determines the warning movement route of the drone based on the real-time position of each lingering person in the target image and the real-time position of each lingering person in the subsequent image, including: The controller marks the lingering persons whose distance from each other is less than a second preset value in two adjacent subsequent images as the same lingering person; The controller obtains the real-time location of the i-th person in the k-th subsequent image , and calculate the average moving speed of the i-th person staying : ， Where, The third preset duration; , N is the total number of subsequent images; The controller obtains the geometric center point of the real-time position of the i-th lingering person in all subsequent images except the earliest subsequent image and marks it as the average center point; The controller uses the average center point of the i-th lingering person and the straight line on which the position points in the earliest subsequent image lie as the movement trend line corresponding to the i-th lingering person; The controller determines the distance from the position of the i-th person staying in the latest subsequent image to the danger zone along the corresponding moving trend line , and calculate the estimated moving time of the i-th person from the position in the latest subsequent image to the dangerous area : ， The controller controls the patrol drones to move over the lingering personnel in order of estimated moving time from small to large to issue an alarm.

8. The method for intelligent geological disaster monitoring based on cloud-edge collaboration according to claim 1 is characterized in that: The system further includes a monitoring terminal communicatively connected to a central agent; the central agent generates a disaster risk level corresponding to the disaster monitoring task based on the central result information, and then further includes: When the disaster risk level is high or medium, the central intelligent agent sends the disaster risk level corresponding to the disaster monitoring task to the monitoring terminal for display.

9. A geological disaster intelligent monitoring system based on cloud-edge collaboration, characterized in that: A method for intelligent monitoring of geological disasters based on cloud-edge collaboration as described in any one of claims 1 to 8 is applied; the system includes a central intelligent agent and an edge intelligent agent that are communicatively connected to each other; the number of the edge intelligent agents is multiple and they are arranged in the monitoring area.

Images