Geological disaster real-time monitoring and early warning system based on Internet of Things and artificial intelligence

Abstract

The invention relates to the field of geological disaster prediction, and discloses a geological disaster real-time monitoring and early warning system based on Internet of Things and artificial intelligence, and the system comprises a sensing module which is used for collecting multi-source data of geological disasters in real time through all Internet of Things sensors, and visually displaying potential geological disaster hidden danger signs on the ground surface; the transmission module is used for transmitting various data and guaranteeing the data transmission security by adopting an encryption technology; the processing module is used for sorting and processing the multi-source data and converting the multi-source data into processing data in a standardized data format; and the model module is used for processing data and applying a model fusing a physical principle and a machine learning algorithm. An advanced early warning model is constructed by applying an artificial intelligence technology and fusing a physical principle and a machine learning algorithm, a deep learning optimization unit customizes a strategy for geological disaster images and time sequence data characteristics, and a knowledge graph unit introduces geological domain knowledge to assist modeling.

Description

Technical Field

[0001] The present invention relates to the technical field of geological disaster prediction, and in particular to a real-time monitoring and early warning system for geological disasters based on the Internet of Things and artificial intelligence. Background Art

[0002] As a natural phenomenon with great destructive power to human life, property and ecological environment, geological disasters, such as frequent landslides, mud-rock flows and ground collapse, have brought a heavy burden to society. In mountainous areas with complex geological conditions, highway and railway projects are seriously threatened by geological disasters. Carrying out risk assessment and early warning of geological disasters has important practical significance and is also an important way to effectively reduce the hazards.

[0003] Traditional real-time monitoring and early warning systems for geological disasters mostly rely on simple empirical formulas or statistical models built based on a small amount of historical data. They lack self-learning and optimization capabilities. Faced with the ever-changing geological environment and increasingly complex disaster-inducing factors, they are unable to dynamically adjust prediction strategies, resulting in low accuracy of prediction results and unable to meet the current urgent needs for precise prevention and control of geological disasters. Summary of the invention

[0004] In view of the shortcomings of the existing technology, the present invention provides a real-time monitoring and early warning system for geological disasters based on the Internet of Things and artificial intelligence, which solves the problem that the prediction results of traditional real-time monitoring and early warning systems for geological disasters are of low accuracy and cannot meet the current urgent needs for precise prevention and control of geological disasters.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: a real-time monitoring and early warning system for geological disasters based on the Internet of Things and artificial intelligence, comprising:

[0006] Perception module: used to collect multi-source data of geological disasters in real time through various IoT sensors, and intuitively display potential signs of geological disasters on the surface;

[0007] Transmission module: used to transmit various data and use encryption technology to ensure data transmission security;

[0008] Processing module: used to organize and process multi-source data and convert them into processed data in a standardized data format;

[0009] Model module: It is used to process data, use models that integrate physical principles and machine learning algorithms, and deep learning architecture to explore potential patterns and laws, predict key information on geological disasters, and output disaster risk assessments;

[0010] Application module: used to generate early warning information based on disaster risk assessment and push it through multiple communication channels. It also presents an analysis report of the early warning information in a visual interface, quickly feeds back disaster loss estimates under different strategies based on model predictions, and assists in optimizing decision-making plans;

[0011] Management module: used to remotely monitor the operating status parameters of each device, diagnose potential faults in real time, and initiate remote maintenance or on-site repair processes when abnormalities occur, while archiving and classifying data.

[0012] Preferably, the perception module includes an Internet of Things sensor unit and a satellite remote sensing unit. The Internet of Things sensor unit is used to collect multi-source data of geological disasters in real time through various Internet of Things sensors. The various Internet of Things sensors include optical fiber displacement sensors, tipping bucket rain sensors, groundwater level sensors, and broadband seismic monitors. The multi-source data include displacement data reflecting the deformation state of the geological body, rainfall data characterizing regional precipitation conditions, water level data reflecting the dynamic changes of the groundwater level, seismic data reflecting the strength of crustal movement, and optical image data and radar remote sensing data showing surface coverage and topographic features. The satellite remote sensing unit is used to perform full-area scanning and imaging of the surface, obtain image data, and intuitively display potential signs of geological disasters on the surface.

[0013] Preferably, the transmission module includes a wireless communication unit and a data dedicated line unit. The wireless communication unit is used to transmit various data between various modules, and the data dedicated line unit is used to encrypt various data in accordance with a unified data interface specification.

[0014] Preferably, the processing module includes an edge computing unit and a cloud data center unit. The edge computing unit is used to use preset data quality rules to screen out abnormal data in multi-source data and perform feature extraction to obtain calculated data. The data quality rules include reasonable numerical interval judgment based on historical data statistical analysis and logical consistency verification of multi-sensor data in the same area. The cloud data center unit is used to unify and standardize the calculated data through a standardized data processing process, and use a spatiotemporal registration algorithm to align multi-source heterogeneous data according to geographic coordinates and timestamp information to obtain processed data.

[0015] Preferably, the model module includes an early warning model unit, a deep learning optimization unit, and a knowledge graph unit. The early warning model unit is used to construct a physical model based on the physical mechanism of geological disasters using processed data as input, and to use the output and processing of the physical model as input, to train and learn complex data patterns through a deep learning framework, to predict the probability of disasters, and to construct a machine learning model. The deep learning optimization unit is used to customize optimization strategies for the characteristics of geological disaster images and time series data. When processing image data, a hybrid deep CNN architecture is used, with shallow large convolution kernels extracting macro features, deep small convolution kernels capturing micro details, and combined with void convolution , three-dimensional convolution expands the receptive field, and integrates multimodal data to improve feature expression; when processing time series data, the attention mechanism is introduced to focus on key time periods, the LSTM gating mechanism is improved to strengthen long-term dependency learning, and tensor decomposition is used to reduce the dimension of high-dimensional data. The knowledge graph unit is used to construct a knowledge graph from various data using deep learning entity recognition and relationship extraction technology, and integrate it with the machine learning model to predict the key information of geological disasters and output disaster risk assessment. The various data include geological reports, academic papers, monitoring reports, and disaster databases. The key information includes possibility, time, scope, and degree of harm.

[0016] Preferably, the application module includes an early warning release unit and a decision support unit. The early warning release unit is used to generate early warning information based on disaster risk assessment and push it through multiple communication channels. The early warning information includes disaster type, expected time range, geographical coordinates of affected area, possible degree of harm. The multiple channels include text messages, mobile phone APP push, electronic display screens, and broadcasts. The decision support unit is used to present an in-depth analysis report based on disaster risk assessment in a visual interface. At the same time, when decision makers input different response strategy variables, the system predicts and feedbacks the estimated disaster losses under different strategies based on the model module to assist in optimizing decision plans. The analysis report includes analysis of disaster-inducing factors, review of similar historical cases, and cost-effectiveness evaluation of response strategies. The different response strategy variables include the scope of personnel evacuation and the intensity of engineering reinforcement measures.

[0017] Preferably, the management module includes a system management unit and a data management unit. The system management unit is used to remotely monitor the operating status parameters of each device, diagnose potential faults in real time, and start remote maintenance or on-site emergency repair processes when an abnormality occurs. The operating status parameters include power, signal strength, CPU usage, memory usage, and the data management unit is used to archive and classify each data according to data timeliness and application requirements, and regularly use data quality assessment tools and algorithms to improve data quality.

[0018] A real-time monitoring and early warning method for geological disasters based on the Internet of Things and artificial intelligence includes the following steps:

[0019] S1. Perception: Real-time collection of multi-source data on geological disasters through networked sensor units, and global scanning and imaging of the surface through satellite remote sensing units to obtain image data, intuitively showing potential signs of geological disasters on the surface;

[0020] S2, transmission: Then, the data is encrypted by the dedicated data line unit in accordance with the unified data interface specification, and the data is transmitted between the modules through the wireless communication unit;

[0021] S3, processing: The edge computing unit then uses the preset data quality rules to filter out abnormal data from the multi-source data, and extracts features to obtain the calculated data. The cloud data center unit then uses a standardized data processing process to standardize the calculated data, and uses a spatiotemporal registration algorithm to align the multi-source heterogeneous data based on geographic coordinates and timestamp information to obtain processed data.

[0022] S4, model: Then, the early warning model unit uses the processed data as input to construct a physical model based on the physical mechanism of geological disasters, and uses the output and processing of the physical model as input, and uses the deep learning framework to train and learn complex data patterns, predict the probability of disasters, and build a machine learning model. The deep learning optimization unit is used to customize the optimization strategy for the characteristics of geological disaster images and time series data, and the knowledge graph unit uses deep learning entity recognition and relationship extraction technology from various data to construct a knowledge graph, and integrates it with the machine learning model to predict the key information of geological disasters and output disaster risk assessment;

[0023] S5. Application: The warning release unit is then used to generate warning information based on the disaster risk assessment and push it through multiple communication channels. The decision support unit presents an in-depth analysis report based on the disaster risk assessment in a visual interface. When the decision maker inputs different response strategy variables, the system predicts and feedbacks the disaster loss estimates under different strategies based on the model module to assist in optimizing the decision plan.

[0024] S6. Management: Remotely monitor the operating status parameters of each device through the system management unit, diagnose potential faults in real time, and initiate remote maintenance or on-site emergency repair processes when abnormalities occur. Archive and classify various data based on data timeliness and application requirements through the data management unit, and regularly use data quality assessment tools and algorithms to improve data quality.

[0025] The present invention provides a real-time monitoring and early warning system for geological disasters based on the Internet of Things and artificial intelligence. It has the following beneficial effects:

[0026] 1. The present invention uses artificial intelligence technology to integrate physical principles and machine learning algorithms to build an advanced early warning model. The deep learning optimization unit customizes strategies based on the characteristics of geological disaster images and time series data. The knowledge graph unit introduces geological knowledge to assist in modeling, thereby improving the accuracy of predictions on key information such as the possibility, time, scope and degree of hazard of disasters. This solves the problem that the traditional real-time monitoring and early warning system for geological disasters has low prediction accuracy and cannot meet the urgent need for precise prevention and control of geological disasters.

[0027] 2. The present invention enhances the monitoring timeliness of geological disasters by combining the Internet of Things with satellite remote sensing technology. It can quickly capture signs when a disaster first appears, buy critical time for subsequent rescue and relief efforts, effectively protect the safety of personnel and property, and provide early warnings to make responses more relaxed. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 This is a system architecture diagram of a real-time monitoring and early warning system for geological disasters based on the Internet of Things and artificial intelligence proposed by the present invention;

[0029] Figure 2 This is a method flow chart of a method for real-time monitoring and early warning of geological disasters based on the Internet of Things and artificial intelligence proposed by the present invention. DETAILED DESCRIPTION

[0030] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0031] Please see attached Figure 1 The embodiment of the present invention provides a real-time monitoring and early warning system for geological disasters based on the Internet of Things and artificial intelligence, including:

[0032] Perception module: used to collect multi-source data of geological disasters in real time through various IoT sensors, and intuitively display potential signs of geological disasters on the surface;

[0033] Transmission module: used to transmit various data and use encryption technology to ensure data transmission security;

[0034] Processing module: used to organize and process multi-source data and convert them into processed data in a standardized data format;

[0035] Model module: It is used to process data, use models that integrate physical principles and machine learning algorithms, and deep learning architecture to explore potential patterns and laws, predict key information on geological disasters, and output disaster risk assessments;

[0036] Application module: used to generate early warning information based on disaster risk assessment and push it through multiple communication channels. It also presents an analysis report of the early warning information in a visual interface, quickly feeds back disaster loss estimates under different strategies based on model predictions, and assists in optimizing decision-making plans;

[0037] Management module: used to remotely monitor the operating status parameters of each device, diagnose potential faults in real time, and initiate remote maintenance or on-site repair processes when abnormalities occur, while archiving and classifying data.

[0038] Specifically, various IoT sensors are used to collect multi-source data on geological disasters in real time, and to visually display potential signs of geological disaster hazards on the surface, so as to grasp the subtle changes in geological bodies in real time, provide a comprehensive data basis for subsequent precise analysis, and realize the extensive collection of information related to geological disasters.

[0039] By transmitting various data and using encryption technology to ensure the security of data transmission, we can ensure that massive amounts of geological disaster data can flow to the processing center stably and quickly under complex terrain and harsh climatic conditions, thereby ensuring the timeliness and security of data transmission.

[0040] By organizing and processing multi-source data and converting them into processed data in a standardized data format, data from different sensors, different formats, and different temporal and spatial resolutions are uniformly "compiled", achieving data standardization and high-quality conversion, and providing a solid data foundation for subsequent precise modeling.

[0041] By processing data and using models that integrate physical principles and machine learning algorithms, as well as deep learning architectures, we can explore potential patterns and laws, predict key information on the occurrence of geological disasters, and output disaster risk assessments, thereby integrating physical knowledge such as geomechanics and fluid mechanics with big data-driven machine learning algorithms. From the stability coefficient of the rock and soil body calculated by the physical model to the hidden associations of massive data mined by the machine learning model, we can comprehensively analyze the causes of disasters, thereby accurately quantifying and predicting the possibility, time, scope and degree of harm of disasters. This solves the problem that traditional early warning models are simple and rough, can only give fuzzy qualitative judgments, and cannot meet the needs of modern disaster prevention and mitigation fine decision-making.

[0042] By generating early warning information based on disaster risk assessment and pushing it through multiple communication channels, while presenting an analysis report of the early warning information in a visual interface, the disaster loss estimates under different strategies can be quickly fed back based on model predictions to assist in optimizing decision-making plans. As soon as disaster risks emerge, early warning information containing disaster details can be sent directly to people's mobile phones via text messages, electronic display screens can light up dangerous areas, and broadcasts can be heard in surrounding communities, so that all parties can be informed quickly, thereby achieving full coverage and efficient transmission of early warning information, and providing decision makers with one-stop services from disaster analysis to response strategy benefit evaluation, making decisions more scientific and reasonable.

[0043] By remotely monitoring the operating status parameters of each device, diagnosing potential faults in real time, and initiating remote maintenance or on-site emergency repair processes when abnormalities occur, and archiving and classifying data at the same time, we can always keep a close eye on the "health status" of each hardware. Once problems such as low power, weak signal or CPU overheating occur, remote repairs or on-site emergency repairs can be carried out immediately, thereby ensuring stable operation of the equipment and orderly management of data.

[0044] Through the coordinated operation of various modules, an organic whole is formed, which carries out intelligent and refined operation of the entire process from geological disaster data collection to precise prevention and control decision-making, thus solving the problem that the prediction results of traditional geological disaster real-time monitoring and early warning systems are of low accuracy and cannot meet the current urgent needs for precise prevention and control of geological disasters.

[0045] The perception module includes an Internet of Things sensor unit and a satellite remote sensing unit. The Internet of Things sensor unit is used to collect multi-source data on geological disasters in real time through various Internet of Things sensors. Various Internet of Things sensors include optical fiber displacement sensors, tipping bucket rainfall sensors, groundwater level sensors, and broadband seismic monitors. The multi-source data include displacement data reflecting the deformation state of the geological body, rainfall data characterizing the regional precipitation conditions, water level data reflecting the dynamic changes of the groundwater level, seismic data reflecting the strength of crustal movement, as well as optical image data and radar remote sensing data showing the surface cover and topographic features. The satellite remote sensing unit is used to scan and image the entire surface, obtain image data, and intuitively display potential signs of geological disasters on the surface.

[0046] Specifically, various IoT sensors are used to collect multi-source data on geological disasters in real time, and through the collaborative operation of multiple IoT sensors, changes in physical parameters closely related to geological disasters are collected in all directions, providing the system with microscopic and real-time first-hand data. By performing full-area scanning and imaging of the surface, image data is obtained, and potential signs of geological disasters on the surface are intuitively displayed, thereby supplementing the system with large-area, high-resolution image information from a macro perspective, complementing the microscopic data of the IoT sensors, and realizing an integrated space-ground monitoring mode.

[0047] The transmission module includes a wireless communication unit and a data dedicated line unit. The wireless communication unit is used to transmit various data between various modules, and the data dedicated line unit is used to encrypt various data in accordance with a unified data interface specification.

[0048] The processing module includes an edge computing unit and a cloud data center unit. The edge computing unit is used to use preset data quality rules to screen out abnormal data in multi-source data and perform feature extraction to obtain calculation data. The data quality rules include reasonable numerical range judgment based on historical data statistical analysis and logical consistency verification of multi-sensor data in the same area. The cloud data center unit is used to unify and standardize the calculation data through a standardized data processing process, and use a spatiotemporal registration algorithm to align multi-source heterogeneous data based on geographic coordinates and timestamp information to obtain processed data.

[0049] Specifically, by utilizing preset data quality rules, abnormal data in multi-source data is screened out, and feature extraction is performed at the same time to obtain calculated data, thereby judging the reasonable numerical range based on the statistical analysis of historical data. This is like defining a "safety range" for each type of sensor data. Based on past rainfall data in the same area and period, the normal range of hourly rainfall is determined to be 0-100 mm. Once the data from the sensor exceeds this range, it is quickly determined to be abnormal data and screened out, avoiding erroneous or distorted data from being mixed into subsequent analysis processes, thereby ensuring data accuracy.

[0050] Logical consistency verification of multi-sensor data in the same area can give full play to the advantages of multi-sensor collaboration. For example, by comparing the displacement sensors and groundwater level sensor data at different locations on the same hillside, if the displacement feedback of a certain sensor increases dramatically, while the surrounding groundwater level does not fluctuate at all, this obviously goes against the common sense of geological changes. Verification can promptly discover such anomalies, identify faulty sensors, and ensure data reliability.

[0051] While screening out anomalies, feature extraction is performed, such as calculating the slope of rainfall change in a short period of time and the rate of groundwater level rise. These features can more intuitively reflect the dynamic development trend of geological disasters, provide key clues for subsequent rapid judgment of disaster risks, and achieve preliminary "purification" and "processing" of the original data.

[0052] The computing data is standardized through a standardized data processing process, and the spatiotemporal registration algorithm is used to align multi-source heterogeneous data according to geographic coordinates and timestamp information to obtain processed data, thereby systematically integrating the scattered and heterogeneous data after preliminary processing by the edge computing unit. The standardized data processing process converts the displacement, rainfall, water level and other data from sensors of different manufacturers and models into a unified data format and encoding rules so that they can be smoothly identified and processed by the system.

[0053] By using the spatiotemporal registration algorithm, sensor data distributed in different geographical locations are accurately mapped to actual spatial locations based on geographic coordinates. Combined with timestamp information, this ensures that different data collected at the same time can be accurately matched, so that the rainfall data at point A at a certain moment and the displacement data at the adjacent point B are perfectly aligned in the spatiotemporal dimension. In this way, the disorganized data can be integrated into orderly, standardized and spatiotemporally coherent processing data, and high-quality data can be input into the model module to ensure that the model can efficiently mine the value of the data.

[0054] The model module includes an early warning model unit, a deep learning optimization unit, and a knowledge graph unit. The early warning model unit is used to construct a physical model based on the physical mechanism of geological disasters with processed data as input, and the output and processing of the physical model as input. It trains and learns complex data patterns through a deep learning framework, predicts the probability of disasters, and constructs a machine learning model. The deep learning optimization unit is used to customize optimization strategies for the characteristics of geological disaster images and time series data. When processing image data, a hybrid deep CNN architecture is adopted, shallow large convolution kernels are used to extract macro features, deep small convolution kernels are used to capture micro details, and hollow convolution and three-dimensional convolution are combined to expand the receptive field, and multimodal data are integrated to enhance feature expression; when processing time series data, an attention mechanism is introduced to focus on key periods, the LSTM gating mechanism is improved to strengthen long-term dependency learning, and tensor decomposition is used to reduce dimensionality and process high-dimensional data. The knowledge graph unit is used to construct a knowledge graph from various data using deep learning entity recognition and relationship extraction technology, and integrate it with the machine learning model to predict key information on the occurrence of geological disasters and output disaster risk assessments. Various data include geological reports, academic papers, monitoring reports, and disaster databases. Key information includes possibility, time, scope, and degree of harm.

[0055] Specifically, by taking the processed data as input to construct a physical model based on the physical mechanism of geological disasters, and taking the output and processing of the physical model as input, the deep learning framework is used to train and learn complex data patterns, predict the probability of disasters, and build a machine learning model, so as to deeply integrate the occurrence mechanism of geological disasters with massive measured data. Taking landslide disasters as an example, the model based on the physical mechanism is based on the cohesion, internal friction angle, and gravity of the rock and soil body, combined with the terrain slope angle, slope length, pore water pressure, seismic force and other external factors, to construct a safety factor calculation formula such as the limit equilibrium theory, and accurately quantify the mechanical conditions for landslide occurrence, which is equivalent to building a solid "theoretical skeleton" for disaster prediction; then the physical model output and processed data are used as nutrients and input into the deep learning framework, using its powerful nonlinear fitting ability to let the model learn from a large amount of historical monitoring data and real-time processing data The hidden patterns of disaster breeding and development under complex and changeable geological environments, such as the correlation between rock and soil deformation and disaster triggering under different rainfall intensities and seismic activity backgrounds, in this way, an accurate prediction of the probability of disaster occurrence can be achieved, providing a solid core basis for disaster warning.

[0056] By customizing optimization strategies based on the characteristics of geological disaster images and time series data, the model is endowed with the ability to keenly perceive the complex characteristics of geological disasters. When processing image data, a hybrid deep CNN architecture is adopted. The shallow large convolution kernel is like a wide-angle lens, which quickly scans satellite remote sensing images, drone shots, etc., and extracts macro-geomorphological features such as mountain trends and watershed morphology, so as to frame the large geological background for subsequent analysis; the deep small convolution kernel is like a microscope, focusing on micro-hazard signs such as rock cracks and loose soil. Combined with void convolution and three-dimensional convolution, it further expands the perception of complex geological structures and dynamic changes of disasters, integrates multimodal data, complements the advantages of optical imaging and radar data, and comprehensively improves the feature expression ability, making the model more accurate in image recognition of geological disaster risks.

[0057] When processing time series data, an attention mechanism is introduced to automatically focus on key time nodes such as concentrated periods of heavy rain, periods of rapid water level changes, and stages of active earthquakes, accurately capture precursor information of disasters, and improve the LSTM gating mechanism to strengthen long-term dependent learning, so that the model can better understand the gradual change laws of geological bodies contained in long-term series data. Tensor decomposition is used to reduce the dimensionality of high-dimensional data, which retains key information while reducing computing costs. This allows the model to handle massive, high-dimensional time series monitoring data with ease, greatly improving the model's ability to parse and predict geological disaster data.

[0058] By using deep learning entity recognition and relationship extraction technology to build a knowledge graph from various data, and integrating it with the machine learning model, we can predict the key information of geological disasters and output disaster risk assessment, thereby injecting rich domain knowledge into the model. From various data such as geological reports, academic papers, monitoring reports, disaster databases, etc., we use the pre-trained BERT model to perform entity recognition, accurately lock in key entities such as geological disaster types, geological structures, monitoring points, sensors, etc., and then determine the relationships between them such as "occurrence", "monitoring", and "impact" through sequence labeling or dependency syntax analysis, and build a huge and detailed knowledge graph, which is stored in the graph database Neo4j.

[0059] During the prediction process, the knowledge features in the knowledge graph are added to the input of the machine learning model. For example, if the geological structure of a certain area is known to be a fractured zone, the knowledge graph can associate the information that the structural area is prone to landslides under heavy rainfall in history, providing key clues for the model to judge the current disaster risk; or reasoning is performed based on the graph. When the model predicts that a certain area has a high landslide risk, the geological causes and experience in handling similar cases can be traced back from the knowledge graph, so that the prediction results can be traced, enhancing the interpretability and credibility of the model. In this way, the organic combination of knowledge-driven and data-driven is realized, and a risk assessment that includes the possibility, time, scope and degree of harm of disasters is accurately output.

[0060] The application module includes an early warning release unit and a decision support unit. The early warning release unit is used to generate early warning information based on disaster risk assessment and push it through multiple communication channels. The early warning information includes the type of disaster, expected time range, geographical coordinates of the affected area, and possible degree of harm. The multiple channels include text messages, mobile phone APP push, electronic display screens, and broadcasts. The decision support unit is used to present an in-depth analysis report based on disaster risk assessment in a visual interface. At the same time, when decision makers input different response strategy variables, the system predicts and feedbacks the estimated disaster losses under different strategies based on the model module, and assists in optimizing decision plans. The analysis report includes analysis of disaster-inducing factors, review of similar historical cases, and cost-effectiveness evaluation of response strategies. Different response strategy variables include the scope of personnel evacuation and the intensity of engineering reinforcement measures.

[0061] The management module includes a system management unit and a data management unit. The system management unit is used to remotely monitor the operating status parameters of each device, diagnose potential faults in real time, and initiate remote maintenance or on-site repair processes when an abnormality occurs. The operating status parameters include power, signal strength, CPU usage, memory usage, and the data management unit is used to archive and classify various data based on data timeliness and application requirements, and regularly use data quality assessment tools and algorithms to improve data quality.

[0062] Please see attached Figure 2, a real-time monitoring and early warning method for geological disasters based on the Internet of Things and artificial intelligence, comprising the following steps:

[0063] S1. Perception: Real-time collection of multi-source data on geological disasters through networked sensor units, and global scanning and imaging of the surface through satellite remote sensing units to obtain image data, intuitively showing potential signs of geological disasters on the surface;

[0064] S2, transmission: Then, the data is encrypted by the dedicated data line unit in accordance with the unified data interface specification, and the data is transmitted between the modules through the wireless communication unit;

[0065] S3, processing: The edge computing unit then uses the preset data quality rules to filter out abnormal data from the multi-source data, and extracts features to obtain the calculated data. The cloud data center unit then uses a standardized data processing process to standardize the calculated data, and uses a spatiotemporal registration algorithm to align the multi-source heterogeneous data based on geographic coordinates and timestamp information to obtain processed data.

[0066] S4, Model: Then, the early warning model unit uses the processed data as input to build a physical model based on the physical mechanism of geological disasters, and uses the output and processing of the physical model as input to train and learn complex data patterns through a deep learning framework to predict the probability of disasters and build a machine learning model. The deep learning optimization unit is used to customize optimization strategies for the characteristics of geological disaster images and time series data, and the knowledge graph unit uses deep learning entity recognition and relationship extraction technology from various data to build a knowledge graph, which is integrated with the machine learning model to predict key information on the occurrence of geological disasters and output disaster risk assessments;

[0067] S5. Application: The warning release unit is then used to generate warning information based on the disaster risk assessment and push it through multiple communication channels. The decision support unit presents an in-depth analysis report based on the disaster risk assessment in a visual interface. When the decision maker inputs different response strategy variables, the system predicts and feedbacks the disaster loss estimates under different strategies based on the model module to assist in optimizing the decision plan.

[0068] S6. Management: Remotely monitor the operating status parameters of each device through the system management unit, diagnose potential faults in real time, and initiate remote maintenance or on-site emergency repair processes when abnormalities occur. Archive and classify various data based on data timeliness and application requirements through the data management unit, and regularly use data quality assessment tools and algorithms to improve data quality.

[0069] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A real-time monitoring and early warning system for geological disasters based on the Internet of Things and artificial intelligence, characterized in that: include: Perception module: used to collect multi-source data of geological disasters in real time through various IoT sensors, and intuitively display potential signs of geological disasters on the surface; Transmission module: used to transmit various data and use encryption technology to ensure data transmission security; Processing module: used to organize and process multi-source data and convert them into processed data in a standardized data format; Model module: It is used to process data, use models that integrate physical principles and machine learning algorithms, and deep learning architecture to explore potential patterns and laws, predict key information on geological disasters, and output disaster risk assessments; Application module: used to generate early warning information based on disaster risk assessment and push it through multiple communication channels. It also presents an analysis report of the early warning information in a visual interface, quickly feeds back disaster loss estimates under different strategies based on model predictions, and assists in optimizing decision-making plans; Management module: used to remotely monitor the operating status parameters of each device, diagnose potential faults in real time, and initiate remote maintenance or on-site repair processes when abnormalities occur, while archiving and classifying data.

2. The real-time monitoring and early warning system for geological disasters based on the Internet of Things and artificial intelligence according to claim 1 is characterized by: The perception module includes an Internet of Things sensor unit and a satellite remote sensing unit. The Internet of Things sensor unit is used to collect multi-source data of geological disasters in real time through various Internet of Things sensors. The various Internet of Things sensors include optical fiber displacement sensors, tipping bucket rainfall sensors, groundwater level sensors, and broadband seismic monitors. The multi-source data include displacement data reflecting the deformation state of the geological body, rainfall data characterizing regional precipitation conditions, water level data reflecting the dynamic changes of the groundwater level, seismic data reflecting the strength of crustal movement, and optical image data and radar remote sensing data showing surface coverage and topographic features. The satellite remote sensing unit is used to perform full-area scanning and imaging of the surface, obtain image data, and intuitively display potential signs of geological disasters on the surface.

3. The real-time monitoring and early warning system for geological disasters based on the Internet of Things and artificial intelligence according to claim 1 is characterized in that: The transmission module includes a wireless communication unit and a data dedicated line unit. The wireless communication unit is used to transmit various data between various modules, and the data dedicated line unit is used to encrypt various data in accordance with a unified data interface specification.

4. The real-time monitoring and early warning system for geological disasters based on the Internet of Things and artificial intelligence according to claim 1 is characterized in that: The processing module includes an edge computing unit and a cloud data center unit. The edge computing unit is used to use preset data quality rules to screen out abnormal data in multi-source data and perform feature extraction to obtain calculation data. The data quality rules include reasonable numerical interval judgment based on historical data statistical analysis and logical consistency verification of multi-sensor data in the same area. The cloud data center unit is used to unify and standardize the calculation data through a standardized data processing process, and use a spatiotemporal registration algorithm to align multi-source heterogeneous data based on geographic coordinates and timestamp information to obtain processed data.

5. The real-time monitoring and early warning system for geological disasters based on the Internet of Things and artificial intelligence according to claim 1 is characterized in that: The model module includes an early warning model unit, a deep learning optimization unit, and a knowledge graph unit. The early warning model unit is used to construct a physical model based on the physical mechanism of geological disasters with processed data as input, and the output and processing of the physical model as input, and to train and learn complex data patterns through a deep learning framework, predict the probability of disaster occurrence, and construct a machine learning model. The deep learning optimization unit is used to customize optimization strategies for the characteristics of geological disaster images and time series data. When processing image data, a hybrid deep CNN architecture is used, shallow large convolution kernels are used to extract macro features, deep small convolution kernels are used to capture micro details, and hollow convolution and three-dimensional convolution are combined to expand the receptive field, and multi-modal data are integrated to enhance feature expression; When processing time series data, an attention mechanism is introduced to focus on key time periods, the LSTM gating mechanism is improved to strengthen long-term dependency learning, and tensor decomposition is used to reduce the dimension to process high-dimensional data. The knowledge graph unit is used to construct a knowledge graph from various data using deep learning entity recognition and relationship extraction technology, and integrate it with the machine learning model to predict key information on the occurrence of geological disasters and output disaster risk assessments. The various data include geological reports, academic papers, monitoring reports, and disaster databases, and the key information includes possibility, time, scope, and degree of harm.

6. The real-time monitoring and early warning system for geological disasters based on the Internet of Things and artificial intelligence according to claim 1 is characterized in that: The application module includes an early warning release unit and a decision support unit. The early warning release unit is used to generate early warning information based on disaster risk assessment and push it through multiple communication channels. The early warning information includes disaster type, expected time range, geographical coordinates of affected area, and possible degree of harm. The multiple channels include text messages, mobile phone APP push, electronic display screens, and broadcasts. The decision support unit is used to present an in-depth analysis report based on disaster risk assessment in a visual interface. At the same time, when decision makers input different response strategy variables, the system predicts and feedbacks the estimated disaster losses under different strategies based on the model module to assist in optimizing decision plans. The analysis report includes analysis of disaster-inducing factors, review of similar historical cases, and cost-effectiveness evaluation of response strategies. The different response strategy variables include the scope of personnel evacuation and the intensity of engineering reinforcement measures.

7. The real-time monitoring and early warning system for geological disasters based on the Internet of Things and artificial intelligence according to claim 1 is characterized by: The management module includes a system management unit and a data management unit. The system management unit is used to remotely monitor the operating status parameters of each device, diagnose potential faults in real time, and start remote maintenance or on-site emergency repair processes when an abnormality occurs. The operating status parameters include power, signal strength, CPU usage, memory usage, and the data management unit is used to archive and classify various data according to data timeliness and application requirements, and regularly use data quality assessment tools and algorithms to improve data quality.

8. A real-time monitoring and early warning method for geological disasters based on the Internet of Things and artificial intelligence, characterized by: A real-time monitoring and early warning system for geological disasters based on the Internet of Things and artificial intelligence as described in any one of claims 1 to 7, comprising the following steps: S1. Perception: Real-time collection of multi-source data on geological disasters through networked sensor units, and global scanning and imaging of the surface through satellite remote sensing units to obtain image data, intuitively showing potential signs of geological disasters on the surface; S2, transmission: Then, the data is encrypted by the dedicated data line unit in accordance with the unified data interface specification, and the data is transmitted between the modules through the wireless communication unit; S3, processing: The edge computing unit then uses the preset data quality rules to filter out abnormal data from the multi-source data, and extracts features to obtain the calculated data. The cloud data center unit then uses a standardized data processing process to standardize the calculated data, and uses a spatiotemporal registration algorithm to align the multi-source heterogeneous data based on geographic coordinates and timestamp information to obtain processed data. S4, Model: Then, the early warning model unit uses the processed data as input to build a physical model based on the physical mechanism of geological disasters, and uses the output and processing of the physical model as input, and uses the deep learning framework to train and learn complex data patterns, predict the probability of disasters, and build a machine learning model. The deep learning optimization unit customizes the optimization strategy for the characteristics of geological disaster images and time series data, and the knowledge graph unit uses deep learning entity recognition and relationship extraction technology from various data to build a knowledge graph, and integrates it with the machine learning model to predict the key information of geological disasters and output disaster risk assessment; S5. Application: The warning release unit is then used to generate warning information based on the disaster risk assessment and push it through multiple communication channels. The decision support unit presents an in-depth analysis report based on the disaster risk assessment in a visual interface. When the decision maker inputs different response strategy variables, the system predicts and feedbacks the disaster loss estimates under different strategies based on the model module to assist in optimizing the decision plan. S6. Management: Remotely monitor the operating status parameters of each device through the system management unit, diagnose potential faults in real time, and initiate remote maintenance or on-site emergency repair processes when abnormalities occur. Archive and classify various data based on data timeliness and application requirements through the data management unit, and regularly use data quality assessment tools and algorithms to improve data quality.

Images