Air-ground cooperative monitoring and early warning method and system for buried gas pipeline in landslide hidden danger area

Abstract

The invention discloses an air-ground cooperative monitoring and early warning method and system for a buried gas pipeline in a landslide hidden danger area, and the method comprises the steps: carrying out the ground monitoring through employing distributed optical fiber sensing, and enabling an optical fiber sensor to be embedded in the outer wall of the buried gas pipeline or to be laid with an accompanying optical fiber pipeline; collecting multiple physical quantity data of strain, temperature and vibration of the pipeline; in-air monitoring, a multi-sensor platform carried by an unmanned aerial vehicle is used for carrying out in-air inspection; the air monitoring data and the ground monitoring data are fused, and a landslide hidden danger area pipeline safety state evaluation model is constructed; performing real-time intelligent analysis on the fusion data by using an advanced mixed attention convolutional neural network, and identifying a potential catastrophe hidden danger area of the pipeline; and according to a risk assessment result, early warning grading is carried out, and a corresponding emergency plan is formulated. According to the method, real-time and accurate monitoring of the pipeline state can be realized, the landslide hidden danger can be found in time, accurate early warning can be realized, and safe operation of the buried gas pipeline in the landslide hidden danger area can be effectively guaranteed.

Description

Technical Field

[0001] The present invention relates to the technical field of pipeline safety automatic monitoring, and more particularly to a method and system for coordinated monitoring and early warning of air-ground underground gas pipelines in landslide-prone areas. Background Art

[0002] In the urban infrastructure system, buried gas pipelines occupy an extremely critical position. Their stable operation plays an indispensable role in ensuring the city's energy supply and maintaining the normal life order of residents. However, when these pipelines are in landslide-prone areas, they face many severe and complex safety risks.

[0003] Landslides are a highly destructive geological disaster that can cause large-scale displacement and deformation of the soil. In this case, buried gas pipelines are very susceptible to strong squeezing from the surrounding soil. This squeezing may cause local depressions or bending deformations in the pipeline, seriously affecting the structural integrity of the pipeline. At the same time, the sliding of the soil may also exert tensile force on the pipeline, causing the pipeline to be elongated in the longitudinal direction, and may even cause the pipeline joints to loosen or become disconnected. In addition, the uneven stress distribution caused by the landslide may also cause the pipeline to twist, changing the original direction and shape of the pipeline, thereby hindering the gas transportation inside the pipeline and increasing the risk of gas leakage.

[0004] Once a gas leak occurs, the consequences are disastrous. Gas itself is flammable and explosive. After leaking, it accumulates in the surrounding environment. Once it encounters a fire source, even a very small open flame or static spark, it is very likely to cause a violent explosion. The shock wave generated by the explosion will cause serious damage to surrounding buildings, causing the walls of the buildings to collapse and the structure to be damaged, leaving residents without a safe place to live. Moreover, the fire caused by the explosion will spread rapidly, engulfing all the combustible materials around it, releasing a large amount of harmful gases and thick smoke, posing a direct threat to the lives and safety of surrounding residents. It will also have a long-term negative impact on the ecological environment, such as soil pollution, air pollution, etc., seriously damaging the local ecological balance.

[0005] Traditional monitoring methods have exposed many obvious limitations when dealing with the safety monitoring of buried gas pipelines in landslide-prone areas. In terms of monitoring scope, traditional methods can only cover local areas of the pipeline and cannot conduct comprehensive and systematic monitoring of the overall condition of the pipeline, which makes it difficult to detect many potential safety hazards in a timely manner. For example, for some pipeline sections located in remote areas or areas with complex terrain, traditional monitoring methods may have monitoring blind spots and cannot obtain accurate information about pipelines in these areas. In terms of monitoring accuracy, traditional technologies are difficult to accurately capture information on small deformations of pipelines. When the pipeline is subjected to initial geological stress, it may produce millimeter-level or even smaller deformations, but traditional monitoring equipment and methods cannot accurately measure these subtle changes, and thus cannot detect early damage and potential risks of the pipeline in a timely manner. In addition, the real-time performance of traditional monitoring is also poor, and manual inspections and data collection are usually required on a regular basis, with a long data update interval. In the event of a possible sudden geological disaster such as a landslide, this lagging monitoring method cannot provide real-time status information of the pipeline in a timely manner, resulting in difficulty in providing timely and effective support for pipeline safety protection and emergency response when a disaster occurs, greatly increasing the possibility of pipeline safety accidents and making it difficult to meet the urgent need for safety monitoring of buried gas pipelines in landslide-prone areas.

[0006] Therefore, it has become an urgent task to develop an efficient, accurate and comprehensive monitoring and early warning method and system that can cope with complex geological environments. It is of great significance to ensure the safe operation of urban gas pipelines and the safety of residents' lives and property. Summary of the invention

[0007] In view of this, the present invention aims to provide a method and system for collaborative monitoring and early warning of air-ground buried gas pipelines in landslide-prone areas, which can solve the above-mentioned technical problems, realize real-time and accurate monitoring of pipeline status, timely discover landslide hazards and accurately warn, and effectively ensure the safe operation of buried gas pipelines in landslide-prone areas.

[0008] In order to achieve the above object, the present invention adopts the following technical solution:

[0009] In a first aspect, an embodiment of the present invention provides a method for collaborative monitoring and early warning of underground gas pipelines in landslide hazard areas, comprising the following steps:

[0010] S1. Ground monitoring:

[0011] Using distributed optical fiber sensing technology, optical fiber sensors are embedded in the outer wall of buried gas pipelines or accompanying optical fiber pipelines are laid;

[0012] Collect multi-physical quantity data such as strain, temperature and vibration of pipelines;

[0013] The data signal-to-noise ratio is improved through deep learning noise reduction algorithms, and the strain-deformation conversion model of multiple physical fields is integrated to accurately restore the deformation state of the pipeline;

[0014] S2. Aerial monitoring:

[0015] Use multi-sensor platforms carried by drones, including high-resolution lidar and hyperspectral imagers, to conduct aerial inspections;

[0016] Collect 3D terrain data, spectral data and visible light images of landslide risk areas to generate ultra-high precision digital elevation models;

[0017] Through multi-source data fusion and deep learning algorithms, the detailed features of geological disasters such as landslide boundary, sliding surface location, crack depth and direction are identified;

[0018] Real-time monitoring of surface deformation and vegetation changes, and prediction of landslide development trends;

[0019] S3. Air-ground collaborative analysis:

[0020] Fuse aerial monitoring data with ground monitoring data to build a high-precision pipe-soil interaction simulation model;

[0021] Use advanced hybrid attention convolutional neural networks to perform real-time intelligent analysis on fused data to identify potential disaster risk areas in pipelines;

[0022] Based on the risk assessment results, early warning classification is carried out and corresponding emergency plans are formulated.

[0023] Furthermore, the data signal-to-noise ratio is improved through deep learning noise reduction algorithms, including:

[0024] Preprocess the multi-physical quantity data of strain, temperature, and vibration transmitted by optical fiber to remove the noise caused by changes in the external environment;

[0025] Use convolutional neural networks to automatically learn the characteristics of signal noise, and combine them with long short-term memory networks to perform time series modeling to identify and extract noise components in the signal;

[0026] The noise filtering strategy is optimized through deep learning algorithms to gradually improve the signal-to-noise ratio. The processed signals will be used for subsequent pipeline deformation status analysis.

[0027] Furthermore, the multi-physics field strain-deformation conversion model includes:

[0028] The de-noised data is input into a multi-physics field strain-deformation conversion model, which is used to calculate the strain change of the pipeline after being stressed through finite element analysis based on the physical coupling relationship between the pipeline and the surrounding soil;

[0029] Combining real-time strain data with environmental factors, the model estimates the deformation state of the pipeline under different working conditions based on the strain gradient, accurately restores the deformation characteristics of the pipeline under the geological disaster of landslide, and then evaluates the health status of the pipeline;

[0030] The pipeline deformation data output by the model is compared with the actual monitoring data to correct and optimize the accuracy of the strain-deformation conversion model.

[0031] Furthermore, an ultra-high precision digital elevation model is generated, including:

[0032] The multi-beam laser radar carried by the drone sends laser pulses and measures the time of reflected echoes to obtain high-precision point cloud data on the ground. After noise removal and point cloud registration preprocessing, preliminary ground data is obtained.

[0033] The hyperspectral imager carried by the drone obtains spectral data of different bands on the ground to distinguish different ground covers; the spectral data is processed by registration with the geographic location and combined with the point cloud data for elevation correction;

[0034] Combine LiDAR and hyperspectral data, apply Kalman filtering or deep learning data fusion methods to optimize the elevation model, reduce inconsistencies between data, and generate a more accurate digital elevation model;

[0035] After surface smoothing and noise filtering, the digital elevation model is generated into an ultra-high precision digital elevation model.

[0036] Furthermore, through multi-source data fusion and deep learning algorithms, the geological disaster details of landslide boundary, sliding surface location, crack depth and direction are identified, including:

[0037] Use image registration technology to align visible light images and DEM data to ensure that all data are processed in the same coordinate system;

[0038] Use the YOLOv8 model to extract features from visible light images, including the boundaries and cracks of landslides;

[0039] The visual features extracted by YOLOv8 are fused with the spatial information in the laser point cloud and DEM data, and the elevation information and terrain features provided by the laser point cloud and DEM data are used to assist in identifying the boundaries and crack locations of the landslide;

[0040] The YOLOv8 model is trained to identify key geological features in the landslide area, including landslide body boundaries, sliding surfaces, crack depths and their orientations; output accurate landslide area feature information;

[0041] By conducting spatiotemporal sequence analysis on the identified landslide area characteristics, the development trend of the disaster can be predicted to provide decision support for disaster prevention and mitigation.

[0042] Furthermore, the advanced hybrid attention convolutional neural network comprises:

[0043] Convolutional layers, used to extract low-level features of pipeline monitoring data;

[0044] The spatial attention module is used to assign different weights to each input feature map to improve the ability to identify local disaster risk areas;

[0045] Temporal attention module, which is used to capture important moments in time series data;

[0046] The fusion layer is used to fuse the outputs of the spatial attention module and the temporal attention module to further enhance the perception ability of the model;

[0047] The fully connected layer classifies and predicts the fused features to quickly and accurately locate the areas with pipeline disaster risks.

[0048] In a second aspect, an embodiment of the present invention further provides an air-ground coordinated monitoring and early warning system for buried gas pipelines in landslide hazard areas, comprising:

[0049] Monitoring equipment layer, including densely distributed fiber demodulation equipment and multi-functional drone perception platform;

[0050] Data processing and analysis layer, including high-precision pipe-soil interaction simulation model and geological hazard identification and assessment system;

[0051] Early warning release and management level, including intelligent cloud platform early warning system and emergency response command and decision support system.

[0052] Furthermore, it also includes:

[0053] The edge computing terminal uses a distributed parallel computing architecture to process data from optical distributed fiber demodulation equipment and multi-functional drone perception platforms at high speed.

[0054] Furthermore, the multifunctional UAV perception platform includes:

[0055] The UAV automatic take-off and landing and autonomous battery replacement ecosystem enables centimeter-level precision landing and rapid battery replacement of UAVs;

[0056] The real-time data transmission protocol based on 5G communication ensures that the monitoring data is transmitted to the ground control center in a high-speed and stable manner.

[0057] Furthermore, the UAV automatic take-off and landing and autonomous battery replacement ecosystem includes:

[0058] High-precision positioning system, using GPS or GLONASS, ensures centimeter-level precision landing of the drone;

[0059] Automatic control system to realize automatic take-off, landing and inspection of drones;

[0060] Efficient energy management system, intelligent battery management system based on municipal facilities and fast battery replacement platform ensure long-term operation of drones.

[0061] It can be seen from the above technical solution that compared with the prior art, the present invention has the following advantages:

[0062] 1. Traditional monitoring methods are difficult to accurately capture the characteristics of pipeline micro-deformation and landslide hazards. The present invention achieves ultra-high precision monitoring through innovative fusion technology, which can accurately locate millimeter-level pipeline deformation and centimeter-level geological disaster characteristics, provide accurate basis for protection, improve the level of refinement of pipeline safety management, and make up for the traditional accuracy defects;

[0063] 2. Traditional technologies mostly rely on manual experience or simple judgment, which makes it difficult to fully and accurately identify disasters and predict trends. The present invention uses advanced technology to quickly and accurately identify disasters and predict development trends, so that gas pipelines can respond in advance, enhancing initiative and foresight, and overcoming the lag of traditional methods;

[0064] 3. Traditional inspections are limited by manpower, time and environment, with low efficiency and difficulty in ensuring continuous monitoring. The drone technology of the present invention improves the inspection frequency and endurance, realizes 24-hour uninterrupted monitoring, overcomes traditional limitations, and technologies such as quantum encryption ensure data security and system stability, which is superior to traditional monitoring systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0066] Figure 1 This is a diagram of the “cloud-edge-end” integrated intelligent monitoring and early warning framework provided by the present invention.

[0067] Figure 2 This is a correlation diagram of the high-precision pipe-soil interaction simulation model provided by the present invention.

[0068] Figure 3 A flow chart of surface crack detection based on multi-source data fusion provided by the present invention.

[0069] Figure 4 This is a flow chart of data collection and processing of the drone perception platform provided by the present invention. DETAILED DESCRIPTION

[0070] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments 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.

[0071] Embodiment 1:

[0072] The embodiment of the present invention discloses a method for collaborative monitoring and early warning of underground gas pipelines in landslide hazard areas. The overall framework is as follows during implementation: Figure 1 As shown, the following steps S1 to S3 are included:

[0073] S1. Ground monitoring:

[0074] Using distributed optical fiber sensing technology, optical fiber sensors are embedded in the outer wall of buried gas pipelines or accompanying optical fiber pipelines are laid;

[0075] Collect multi-physical quantity data such as strain, temperature and vibration of pipelines;

[0076] The data signal-to-noise ratio is improved through a deep learning noise reduction algorithm, and the strain-deformation conversion model of multiple physical fields is integrated to accurately restore the deformation state of the pipeline.

[0077] The details are as follows:

[0078] 1.1 Pipeline monitoring based on dense distributed optical fiber sensing, in which an integrated externally bonded optical fiber sensing structure is embedded in the outer wall of the pipeline, and a non-excavation directional drilling technology is used to lay the accompanying optical fiber pipeline;

[0079] 1.2 Adopting an adaptive noise reduction algorithm based on deep learning to improve the data signal-to-noise ratio, and integrating the strain-deformation conversion model of multiple physical fields to accurately restore the deformation state of the pipeline;

[0080] The calculation process of the adaptive noise reduction algorithm is as follows:

[0081] First, the original signal transmitted by the optical fiber is preprocessed to remove the noise caused by changes in the external environment (such as temperature fluctuations, mechanical vibrations, etc.). The convolutional neural network (CNN) is used to automatically learn the characteristics of signal noise, and the long short-term memory (LSTM) network is combined for time series modeling to identify and extract the noise components in the signal. The noise filtering strategy is optimized through the deep learning algorithm, and the signal-to-noise ratio of the signal is gradually improved, and the noise suppression amplitude reaches 20-30dB. The processed signal will be used for subsequent pipeline deformation state analysis.

[0082] The specific calculation process of the strain-deformation conversion model integrating multiple physical fields is as follows:

[0083] The de-noised data is input into the multi-physics field strain-deformation conversion model, which first calculates the strain change of the pipeline after being stressed through finite element analysis (FEA) based on the physical coupling relationship between the pipeline and the surrounding soil. Combining real-time strain data with environmental factors (such as geological humidity, temperature, etc.), the model calculates the deformation state of the pipeline under different working conditions based on the strain gradient, accurately restores the deformation characteristics of the pipeline under geological disasters such as landslides, and then evaluates the health of the pipeline. The pipeline deformation data output by the model is compared with the actual monitoring data to further correct and optimize the accuracy of the strain-deformation conversion model.

[0084] In the specific implementation, for newly built buried gas pipelines, an externally bonded optical fiber sensing structure that is integrated during the pipeline production process is used, and a large-diameter bionic optical fiber locking device is embedded in the outer wall of the pipeline, so that the sensing optical cable is closely combined with the outer wall of the pipeline, ensuring that the pipeline stress and strain can be effectively transmitted to the optical fiber, and realizing efficient sensing signal coupling. For existing pipelines, the accompanying optical fiber pipeline is laid using trenchless directional drilling technology. During the laying process, the distance between the measuring pipe and the existing pipeline is controlled within the range of 0.5-1.5 meters with the help of geological radar and inertial navigation collaborative positioning. This distance has been verified in practice to capture the deformation characteristics of the pipeline well. An adaptive noise reduction algorithm based on deep learning is adopted, which can automatically adjust the filtering parameters according to the noise characteristics under different geological environments, thereby increasing the signal-to-noise ratio of the data by 20-30dB. At the same time, by integrating the strain-deformation conversion model of multiple physical fields, the deformation state of the pipeline under complex stress conditions can be accurately restored, and high-sensitivity monitoring of small deformations of the pipeline can be achieved. The strain monitoring accuracy can reach ±3μm, and the spatiotemporal characteristics of pipeline deformation can be accurately obtained. Indoor unit tests and large-scale physical model tests were conducted to simulate various landslide types (such as translational landslides, rotational landslides, traction landslides, etc.), pipeline-soil interaction processes under different working conditions such as sliding rates in the range of 0.1-10mm / s and changes in soil properties. Through these tests, the soil stress-strain transmission mechanism and the stress-deformation law of the pipeline were deeply analyzed, and a pipe-soil interface constitutive model was constructed based on the test data. In combination with the optimized material point method (MPM), an efficient numerical simulation model was developed, which can accurately simulate the mechanical behavior of the pipeline during the disaster process, thereby accurately determining the disaster threshold range. The field monitoring data is analyzed in real time by an advanced hybrid attention convolutional neural network (HACNN) with dynamic weight allocation, and dynamically compared with the simulation model threshold, thereby realizing the rapid and accurate positioning of the pipeline disaster risk area, with a positioning accuracy of less than 0.5 meters, and the initial judgment accuracy of the disaster type exceeds 85%.

[0085] S2. Aerial monitoring:

[0086] Use multi-sensor platforms carried by drones, including high-resolution lidar and hyperspectral imagers, to conduct aerial inspections;

[0087] Collect 3D terrain data, spectral data and visible light images of landslide risk areas to generate ultra-high precision digital elevation models;

[0088] Through multi-source data fusion and deep learning algorithms, the detailed features of geological disasters such as landslide boundary, sliding surface location, crack depth and direction are identified;

[0089] Monitor surface deformation and vegetation changes in real time and predict landslide development trends.

[0090] The details are as follows:

[0091] 2.1 Generate an ultra-high precision digital elevation model using sensors such as multi-beam lidar and hyperspectral imagers carried by drones: The calculation process of building an "ultra-high precision digital elevation model" based on radar data and spectral data includes:

[0092] LiDAR data processing: The multi-beam LiDAR carried by the drone obtains high-precision ground point cloud data by sending laser pulses and measuring the time of reflected echoes. The scanning data of LiDAR covers a wide area and can obtain high-density ground reflection points. After pre-processing (such as noise removal, point cloud registration, etc.), the radar data is combined with the ground height information and the ground elevation information is calculated through a three-dimensional reconstruction algorithm to generate a preliminary digital elevation model (DEM).

[0093] Hyperspectral data processing: The hyperspectral imager carried by the drone captures the spectral reflectance characteristics of ground objects and obtains ground spectral data in different bands. These data can be used to distinguish different types of ground covers, such as soil, vegetation, buildings, etc. Through the registration processing of spectral data and geographic location, combined with ground point cloud data, the surface details in the elevation model can be further accurately corrected, especially when hyperspectral information is available, which helps to improve the accuracy of the model and the resolution of ground features.

[0094] Data fusion and elevation model optimization: LiDAR and hyperspectral data provide different types of information. The former mainly provides ground elevation and morphology, while the latter provides object classification and texture features. In order to build a more accurate ultra-high precision digital elevation model, it is key to fuse the data of the two. Through advanced data fusion technology, such as Kalman filtering or deep learning models, combined with the advantages of the two data sources, the accuracy of the elevation model can be effectively improved and the inconsistency between data can be reduced.

[0095] Model output and analysis: The fused elevation model is post-processed (such as ground surface smoothing, noise filtering, etc.) to ultimately generate an ultra-high-precision digital elevation model (DEM).

[0096] The process of data transmission and elevation model construction:

[0097] Transmission process: The data collected by the drone is first transmitted to the "data processing and analysis layer" through the wireless communication module. This layer includes efficient computing and analysis units that can process and generate high-precision digital elevation models in real time. During the transmission process, the data can be compressed and optimized to improve transmission efficiency and stability.

[0098] Model construction: Once the data is transferred to the data processing and analysis layer, after rigorous preprocessing and analysis, the above-mentioned algorithms and models are used to construct the final elevation model.

[0099] 2.2 Through multimodal fusion image registration and interpretation algorithms, the detailed features of geological disasters such as landslide boundaries, sliding surface locations, crack depths and directions can be accurately identified.

[0100] The multimodal data includes: visible light images and DEM digital elevation models, wherein: visible light images use drones equipped with high-definition cameras to obtain high-resolution visible light images of the landslide area. The images are first pre-processed by denoising, contrast enhancement, and other operations to improve the accuracy of subsequent analysis. DEM data is digital elevation model (DEM) data generated from lidar point clouds, providing elevation information of the area and helping to determine the topographic features of the landslide area.

[0101] Image registration technology is used to align the data of different modalities (visible light image and DEM data) to ensure that all data are processed in the same coordinate system. Image registration is achieved through feature point matching automatic registration algorithm to ensure spatial consistency between different data sources.

[0102] Feature extraction and fusion: The Yolov8 model is used to extract features from visible light images (such as the boundaries and cracks of landslide bodies), and these features are fused with the spatial information in the laser point cloud and DEM data. The laser point cloud and DEM data provide topographic and elevation information of the landslide area, and the visual features extracted by CNN help identify the boundaries and crack locations of the landslide.

[0103] The YOLOv8 model is trained to identify key geological features in the landslide area, such as the landslide boundary, sliding surface, crack depth and its direction. Its input data includes visible light images and DEM data. The visible light images are used to capture the surface features of the landslide, and the DEM data is used to extract elevation information to assist in identifying the sliding surface and cracks of the landslide.

[0104] Ultimately, the precise landslide regional features generated include the landslide body boundary, the location of the sliding surface, the depth and direction of the cracks, etc., which will be used for subsequent landslide monitoring, disaster assessment, and disaster prevention and mitigation decision support.

[0105] Combining the fusion analysis results of different modal data can improve the accuracy of landslide area identification, especially in complex terrain and extreme weather conditions, and can also effectively ensure the robustness of the model.

[0106] In the specific implementation, a high-resolution, long-endurance UAV platform is selected, and it is equipped with advanced sensors such as multi-beam laser radar and hyperspectral imager to form a full-range perception system. The three-dimensional point cloud data obtained by laser radar scanning is quickly processed using the point cloud intelligent classification and adaptive filtering algorithm based on deep learning to generate an ultra-high-precision digital elevation model (DEM) with an average resolution of 2-3 cm. Combining the spectral feature analysis of hyperspectral images and the texture feature extraction technology of optical images, the image registration and interpretation algorithm of multimodal fusion can accurately identify the detailed features of geological disasters such as landslide body boundaries, sliding surface positions, crack depths and directions, and the accuracy of geological disaster feature recognition is improved by 30%-40% compared with traditional methods. By analyzing the spatiotemporal sequence of these features, the development trend of disasters can also be predicted. Multimodal data fusion of visible light images and DEM elevation data, and Yolov8 for landslide crack identification, can be used to perform ultra-precise positioning (positioning accuracy can reach within 0.3 meters) and quantitative evaluation (crack size evaluation error is less than 5 cm) of ground collapse and ground fissures. By establishing a pipeline leakage gas diffusion model and a surface deformation correlation model, combined with multi-sensor data collaborative analysis, including DEM data obtained by drone lidar processing, visible light image data, and strain, temperature, and vibration data of ground optical fibers, the potential geological disaster risks caused by pipeline leakage can be effectively identified. The low-power, high-performance edge computing platform on the drone adopts a real-time data transmission protocol based on 5G communication, which can ensure that the monitoring data is transmitted to the ground control center at a high speed and stability of more than 1Gbps, and realize real-time analysis and comprehensive evaluation of the data. Existing municipal street lights, traffic signal poles and other facilities are transformed, integrated with edge computing and control modules, and an intelligent drone automatic take-off and landing and autonomous battery replacement ecosystem is built. Allow drones to achieve centimeter-level (average positioning accuracy of 3-5 cm) precise landing in complex urban environments. The designed modular, fast-plug and efficient battery replacement platform and intelligent battery management system shorten the battery replacement time to less than 30 seconds, ensuring that drones can inspect pipelines 24 hours a day.

[0107] S3. Air-ground collaborative analysis:

[0108] The aerial monitoring data and the ground monitoring data are integrated to build a pipeline safety status assessment model in landslide-prone areas;

[0109] Use advanced hybrid attention convolutional neural networks to perform real-time intelligent analysis on fused data to identify potential disaster risk areas in pipelines;

[0110] Based on the risk assessment results, early warning classification is carried out and corresponding emergency plans are formulated.

[0111] 3.1 Real-time intelligent analysis is performed through advanced hybrid attention convolutional neural networks to achieve rapid and accurate positioning of pipeline disaster risk areas.

[0112] The advanced hybrid attention convolutional neural network (HACNN) first receives input from pipeline monitoring data, which includes multiple monitoring signals provided by fiber optic sensors, GNSS systems, drones, etc. The network performs deep feature extraction on these multi-dimensional data, and can quickly identify potential disaster risk areas of the pipeline.

[0113] The specific structure of the "Advanced Hybrid Attention Convolutional Neural Network" includes:

[0114] 1). Convolutional layer (CNN part):

[0115] This part is responsible for extracting low-level features of pipeline monitoring data. The input data is filtered and feature extracted step by step through multiple convolutional layers. Through multi-layer convolution operations, the network can identify basic features such as pipeline deformation, stress, and displacement.

[0116] 2) Spatial Attention Module:

[0117] In order to improve the expressiveness of spatial features, a spatial attention mechanism is introduced. This module assigns different weights to each input feature map according to the importance of the spatial features. By calculating the attention distribution in the spatial dimension, the feature representation of key areas is strengthened, and the ability to identify local disaster risk areas is improved.

[0118] 3) Temporal Attention Module:

[0119] Used to capture important moments in time series data. This module combines the long short-term memory (LSTM) network to automatically adjust the attention weights of different time steps in the time dimension, identify the time trend of pipeline deformation changes, and accurately capture the key period when the pipeline is affected by the disaster.

[0120] 4). Fusion layer:

[0121] The outputs of the spatial attention module and the temporal attention module are combined through a fusion layer, which fuses the features extracted from the spatial and temporal dimensions into a unified high-dimensional feature representation, further enhancing the perception ability of the model.

[0122] 5). Fully connected layer:

[0123] Finally, the fused features are classified and predicted through multiple fully connected layers. Through the layer-by-layer learning of these layers, the network can extract the final disaster risk area information from complex multi-physical data and locate it quickly and accurately.

[0124] Through this advanced hybrid attention mechanism, the model can efficiently identify potential catastrophic risks of pipelines from complex time-space data and provide fast and accurate positioning results, thereby effectively supporting real-time monitoring and safety management of pipelines.

[0125] In the specific implementation, a distributed parallel computing architecture is used in the edge computing terminal to process the data from the densely distributed optical fiber sensing system and the data collected by the multi-sensor of the drone at high speed, including nanosecond time registration, spatial registration and multi-scale fusion processing, and then extract multi-dimensional feature vectors. Blockchain technology is combined to store and trace the data for trustworthy storage and traceability management to ensure the integrity and reliability of the data. The processed data is uploaded to the cloud server through the low-latency, high-bandwidth 6G communication network, and the data upload delay is controlled within 10 milliseconds. A risk assessment model based on deep reinforcement learning is deployed on the cloud server. The model integrates geological big data, historical pipeline operation and maintenance data and real-time monitoring data. By dynamically learning different geological disasters and pipeline failure modes, an adaptive buried pipeline health status characterization indicator system and dynamic early warning criteria are established. An adaptive frequency conversion acquisition, transmission and control algorithm based on edge intelligence is developed on the edge computing terminal to realize intelligent interconnection and collaborative work with various monitoring sensors, drones and municipal facilities, effectively reducing system power consumption and delay. The power consumption of the edge computing terminal is reduced by 30%-40%, and the data processing delay is shortened to less than 50 milliseconds. Relying on the urban Internet of Things (IoT) infrastructure, wireless energy transmission and smart microgrid technologies are used to solve the power supply problem of monitoring equipment. At the same time, software-defined networking (SDN) and network function virtualization (NFV) technologies are used to realize flexible configuration and optimization of communication networks to ensure efficient and stable data transmission. Large-scale demonstration applications are carried out in typical landslide hazard areas. Based on detailed local geological survey reports and pipeline geographic information system (GIS) data, high-risk pipeline sections are carefully selected for systematic and refined layout. In the application process, big data analysis and artificial intelligence technologies are used to continuously explore the value of monitoring data, deeply optimize early warning algorithms and model parameters, and according to actual statistics, the accuracy of system early warnings reaches more than 90%, and the false alarm rate is less than 5%.

[0126] The embodiment of the present invention provides a method for collaborative monitoring and early warning of underground gas pipelines in landslide-prone areas, which aims to achieve real-time and accurate monitoring of pipeline status, timely discover landslide hazards and accurately warn, and effectively ensure the safe operation of buried gas pipelines in landslide-prone areas. The method includes pipeline monitoring based on dense distributed fiber optic sensing, using an integrated external bonding fiber optic sensing structure to embed into the outer wall of the pipeline, and laying a companion fiber optic pipeline using trenchless directional drilling technology. It uses an adaptive noise reduction algorithm based on deep learning to improve the data signal-to-noise ratio, and integrates the strain-deformation conversion model of multiple physical fields to accurately restore the deformation state of the pipeline. The system uses an advanced hybrid attention convolutional neural network to perform real-time intelligent analysis to achieve rapid and accurate positioning of pipeline disaster risk areas.

[0127] Embodiment 2:

[0128] The embodiment of the present invention further provides an air-ground coordinated monitoring and early warning system for buried gas pipelines in landslide hazard areas, comprising:

[0129] Monitoring equipment layer, including densely distributed fiber demodulation equipment and multi-functional drone perception platform;

[0130] Data processing and analysis layer, including high-precision pipe-soil interaction simulation model and geological hazard identification and assessment system;

[0131] Early warning release and management level, including intelligent cloud platform early warning system and emergency response command and decision support system.

[0132] The edge computing terminal uses a distributed parallel computing architecture to process data from optical distributed fiber demodulation equipment and multi-functional drone perception platforms at high speed.

[0133] The monitoring and early warning system includes a monitoring equipment layer, a data processing and analysis layer, and an early warning release and management layer. The monitoring equipment layer includes densely distributed optical fiber sensors and intelligent drones. The densely distributed optical fiber sensors mainly collect strain, temperature, and vibration data, and can perform ultra-high-precision monitoring of multiple physical quantities such as strain, temperature, and vibration of buried gas pipelines at the micro-strain level (the strain monitoring accuracy can reach ±3μm), which can effectively capture pipeline micro-deformation and potential fault signals; the multifunctional drone perception platform integrates a variety of advanced sensors, including high-resolution laser radar (resolution can reach 5mm), hyperspectral camera (spectral resolution can reach 5nm), infrared thermal imager (temperature resolution can reach 0.1℃), ultraviolet imager and synthetic aperture radar, etc., so that it has the ability to perform all-round and multi-scale perception of geological disasters in landslide hazard areas under complex meteorological conditions, and can obtain rich and accurate monitoring data. The intelligent drone automatic take-off and landing and battery replacement platform is based on the intelligent transformation of existing municipal facilities, integrating high-precision positioning, automatic control and efficient energy management technology to ensure that drones can take off and land safely and quickly and efficiently replace batteries in various environments, realizing long-term, uninterrupted autonomous inspection operations. The data processing and analysis layer is mainly implemented in the edge computing module, including a high-precision pipe-soil interaction simulation model, which takes into account the nonlinear mechanical properties of soil, the constitutive relationship of pipeline materials, and complex geological environmental factors. It is verified and optimized by accumulating more than 100 sets of test data and actual engineering cases under different working conditions, and accurately simulates the mechanical process of pipeline disasters under different landslide conditions. The efficient geological disaster identification and assessment system integrates deep learning, computer vision and geographic information system (GIS) technology, and can quickly process and deeply analyze multi-source heterogeneous data collected by drones, and realize automatic identification, precise positioning, quantitative evaluation and dynamic prediction of geological disasters such as landslides, ground collapses, and ground fissures, providing decision support for pipeline safety protection. The safe and reliable data fusion and encryption transmission module uses hash algorithms to ensure the security, integrity and accuracy of monitoring data during collection, transmission and storage, effectively prevent data leakage and tampering, and ensure stable operation of the system. The warning release and management level are mainly implemented in the intelligent cloud platform warning system, which processes and deeply analyzes massive monitoring data in real time, and automatically generates accurate and detailed warning information based on preset dynamic warning thresholds and risk level standards. It is also released to pipeline operation management departments, emergency rescue agencies, surrounding residents and relevant responsible persons in a timely manner through 5G SMS, customized APP push, intelligent voice broadcast, electronic map pop-up windows and other communication methods to ensure the efficiency and reliability of information transmission.The emergency response command and decision support system integrates geographic information system (GIS), emergency resource management and intelligent scheduling algorithms. It can quickly activate the emergency plan after the early warning is triggered, and provide emergency rescue commanders with real-time situation awareness at the disaster site, optimal allocation of rescue resources and intelligent decision-making support for action plans, so as to achieve rapid, scientific and orderly emergency response and minimize disaster losses.

[0134] The system also includes disaster identification and assessment based on drone multi-sensor perception, drone automatic take-off and landing and autonomous battery replacement ecosystem, and air-ground collaborative monitoring and early warning system integration and application. The system can achieve 24-hour uninterrupted monitoring, using technologies such as hash algorithms to ensure data security and system stability, with high early warning accuracy and low false alarm rate, effectively improving the level of refined pipeline safety management, enhancing initiative and foresight, and overcoming the limitations of traditional monitoring methods.

[0135] Reference Figure 1 :It is a framework diagram of intelligent monitoring and early warning integrated with “cloud-edge-end”;

[0136] This figure shows the architecture of the entire monitoring and early warning system, including the buried pipeline monitoring and early warning cloud platform in urban geological disaster risk areas, edge multi-network fusion gateway equipment and multi-source multi-physical sensing equipment. The cloud platform is responsible for equipment monitoring, data storage, risk analysis, early warning calculation and release and other functions. The edge multi-network fusion gateway device performs data reception, forwarding, model synchronization, data display, storage, distribution, early warning calculation and release, and issues instructions to the next level of equipment, connecting with other devices through multiple networks (such as LoRa, mobile 5G / 4G, multi-network fusion, etc.). Multi-source multi-physical sensing equipment includes distributed fiber optic sensors, municipal facility sensing equipment, drone-mounted sensors, geological disaster monitoring equipment, etc., which are distributed around underground pipelines and on the surface to collect various data.

[0137] Figure 2 :Correlation diagram of high-precision pipe-soil interaction simulation model

[0138] It includes pipeline deformation monitoring, artificial intelligence analysis, and pipe-soil interaction constitutive model. The pipeline deformation monitoring diagram shows the deformation of the pipeline under stress, and provides a basis for subsequent research through the collection and analysis of pipeline deformation data. The artificial intelligence analysis diagram shows how to use advanced algorithms to process and analyze the collected data and mine the potential information in the data, such as judging the health status of the pipeline and the risks it may face by analyzing the pipeline deformation data. The pipe-soil interaction constitutive model diagram presents the mechanical relationship model between the pipe and the soil, considering factors such as matrix suction, the influence of temperature on the critical friction angle, and the coupling of temperature and matrix suction. This model can more accurately simulate the stress of the pipeline under different geological conditions, providing a theoretical basis for pipeline disaster prediction.

[0139] Figure 3 :Flowchart of surface crack detection based on multi-source data fusion

[0140] Starting with multi-source data such as digital elevation model, slope grayscale map, visible light and infrared images, these data are first interpolated and filtered to improve data quality. Then image registration is performed to align data from different sources in space. The registered images are then fused to obtain a fused image containing more information. Finally, the fused image is analyzed through a crack detection and quantification network to obtain the crack detection results. This process shows the detailed process of how to use multi-source data for surface crack detection, which can more accurately identify and evaluate the surface crack situation, and then determine the potential threat to the buried gas pipeline.

[0141] Figure 4 :Data collection and processing flow chart of drone perception platform

[0142] After the drone is equipped with multiple sensors to collect data, the collected point cloud data is first processed using the point cloud preprocessing algorithm, and the optical image and other data are processed using the image processing algorithm. The processed data is combined with the digital elevation model and analyzed through the disaster area boundary and feature analysis network to finally obtain the disaster area identification result.

[0143] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0144] For example, the present invention involves the following four steps when it is implemented:

[0145] 1: System installation and debugging

[0146] Along the buried gas pipelines selected in the landslide hazard area, a fiber optic sensor installation plan is formulated based on the pipeline material, diameter, burial depth and surrounding geological conditions. For newly built pipelines, the integrated external bonded fiber optic sensing structure is precisely embedded in the outer wall of the pipeline during the factory prefabrication stage to ensure the coordinated deformation of the optical fiber and the pipeline. For existing pipelines, the accompanying fiber optic pipeline is laid using trenchless directional drilling technology, combined with geological radar to detect the underground geological structure, and the inertial navigation system is used to determine the position of the measuring pipe to ensure that the distance between the measuring pipe and the existing pipeline is appropriate. After laying, the fiber optic pipeline is sealed and fixed. After installation, high-precision calibration equipment is used to calibrate the fiber optic sensors one by one, and the accuracy of the sensor output data is checked by applying known stress and strain.

[0147] Install intelligent automatic take-off and landing platforms and efficient battery exchange equipment at appropriate locations of surrounding municipal facilities. Use a total station to calibrate the horizontal positioning of the take-off and landing platform, and use a laser level to calibrate the vertical positioning to ensure that the drone takes off and lands smoothly and safely. Perform charging and discharging tests on the battery exchange equipment and debug the battery management system to ensure stable battery performance and efficient battery exchange. After the installation of various sensors carried by the drone is completed, use a standard calibration board to calibrate the sensor parameters, and use professional calibration software to test the sensor performance to ensure high-precision and reliability of data collection.

[0148] 2: Data collection and transmission

[0149] Start the densely distributed fiber optic monitoring system and use distributed clock synchronization technology to ensure the time consistency of data collected by each fiber optic sensor. Collect the strain, temperature, vibration and other physical quantity data of the pipeline in real time, and perform rapid preprocessing through the multi-mode interference enhanced edge computing chip built into the demodulation device. Adopt an adaptive noise reduction algorithm based on deep learning to remove noise interference, extract key feature information, and use a wireless communication module based on quantum encryption to securely transmit data to the edge computing gateway. Data redundancy check and error correction coding technology are used during transmission to ensure data integrity.

[0150] According to the scheduled inspection plan, the drone obtains weather, pipeline location information and mission instructions through a two-way communication link with the ground control center before taking off. After taking off, it uses a variety of advanced sensors to collect data from the landslide hazard area in an all-round scanning manner, and adjusts sensor parameters in real time according to environmental changes during collection. The collected data is pre-processed and preliminarily analyzed in real time by the onboard high-performance edge computing platform, and encrypted and transmitted back to the ground control center using a low-latency, high-bandwidth data transmission protocol based on 5G communication to ensure data immediacy and security.

[0151] 3: Data analysis and early warning decision-making

[0152] After receiving the data from the fiber optic monitoring system and the drone inspection system, the ground control center uses a hash algorithm to verify the integrity and then transmits the data to the cloud server. The cloud server uses a pre-established high-precision pipe-soil interaction simulation model and a multivariate data fusion algorithm to deeply fuse and analyze the data, combines the historical operation and maintenance data of the pipeline, the long-term monitoring data of the surrounding geological environment, and geological big data, and uses a risk assessment model based on deep reinforcement learning to comprehensively assess the safety status of the pipeline, and dynamically update the pipeline health status characterization indicators and early warning criteria.

[0153] According to the real-time risk assessment results and dynamic warning thresholds, when the system determines that there is a safety risk in the pipeline, the intelligent warning decision-making mechanism is immediately activated. The warning information generation module generates detailed and accurate warning information based on the risk level, disaster type, possible impact range, etc., including text description, geographic coordinates, risk level icon, on-site real-time images or videos, etc., and issues warning information to pipeline operation management departments, emergency rescue agencies, surrounding residents and relevant responsible persons through 5G SMS, customized APP push, intelligent voice broadcast, electronic map pop-up windows and other communication methods. At the same time, the intelligent voice interaction system automatically dials the preset emergency contact number for voice notification to ensure that the information is fully and timely conveyed.

[0154] 4: Emergency response and system maintenance

[0155] After relevant departments and personnel receive the early warning information, the emergency response command and decision support system automatically activates the emergency plan. According to the disaster type, location and risk level, emergency rescue resources such as rescue equipment, rescue personnel, firefighting equipment, etc. are quickly deployed, and the Geographic Information System (GIS) is used to provide the rescue team with the best action route planning and on-site situation awareness information. During the emergency rescue process, the real-time data provided by the monitoring and early warning system is used to dynamically adjust the rescue action and support decision-making to ensure that the rescue work is carried out efficiently and safely.

[0156] Regularly carry out comprehensive maintenance and overhaul of the monitoring and early warning system, and establish a strict inspection system and equipment maintenance log. Check the working status of fiber optic sensors, the stability of connection lines, the performance of drone equipment and the wear of key components, the signal strength and transmission quality of the communication network, etc. Regularly update and optimize the system software algorithms and models, adjust the warning thresholds and optimize the risk assessment model parameters according to new monitoring data and actual application feedback, and improve the accuracy and reliability of the system. At the same time, use big data analysis technology to deeply mine and discover knowledge of the massive data generated by the operation of the system, and provide data support and decision-making reference for pipeline operation and maintenance management and disaster prevention and control. In addition, organize professionals to train system operators to familiarize them with the system operation process and emergency response methods to ensure that the system can operate normally and maximize its efficiency at critical moments.

[0157] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for collaborative monitoring and early warning of underground gas pipelines and open spaces in landslide hazard areas, characterized in that: The following steps are involved: S1. Ground monitoring: Using distributed optical fiber sensing technology, optical fiber sensors are embedded in the outer wall of buried gas pipelines or accompanying optical fiber pipelines are laid; Collect multi-physical quantity data such as strain, temperature and vibration of pipelines; The data signal-to-noise ratio is improved through deep learning noise reduction algorithms, and the strain-deformation conversion model of multiple physical fields is integrated to accurately restore the deformation state of the pipeline; S2. Aerial monitoring: Use multi-sensor platforms carried by drones, including high-resolution lidar and hyperspectral imagers, to conduct aerial inspections; Collect 3D terrain data, spectral data and visible light images of landslide risk areas to generate ultra-high precision digital elevation models; Through multi-source data fusion and deep learning algorithms, the detailed features of geological disasters such as landslide boundary, sliding surface location, crack depth and direction are identified; Real-time monitoring of surface deformation and vegetation changes, and prediction of landslide development trends; S3. Air-ground collaborative analysis: Fuse aerial monitoring data with ground monitoring data to build a high-precision pipe-soil interaction simulation model; Use advanced hybrid attention convolutional neural networks to perform real-time intelligent analysis on fused data to identify potential disaster risk areas in pipelines; Based on the risk assessment results, early warning classification is carried out and corresponding emergency plans are formulated.

2. The method for collaborative monitoring and early warning of underground gas pipelines and open spaces in landslide hazard areas according to claim 1 is characterized in that: Improve the data signal-to-noise ratio through deep learning noise reduction algorithms, including: Preprocess the multi-physical quantity data of strain, temperature, and vibration transmitted by optical fiber to remove the noise caused by changes in the external environment; Use convolutional neural networks to automatically learn the characteristics of signal noise, and combine them with long short-term memory networks to perform time series modeling to identify and extract noise components in the signal; The noise filtering strategy is optimized through deep learning algorithms to gradually improve the signal-to-noise ratio. The processed signals will be used for subsequent pipeline deformation status analysis.

3. The method for collaborative monitoring and early warning of underground gas pipelines and open spaces in landslide hazard areas according to claim 1 is characterized in that: The multi-physics field strain-deformation conversion model includes: The de-noised data is input into a multi-physics field strain-deformation conversion model, which is used to calculate the strain change of the pipeline after being stressed through finite element analysis based on the physical coupling relationship between the pipeline and the surrounding soil; Combining real-time strain data with environmental factors, the model estimates the deformation state of the pipeline under different working conditions based on the strain gradient, accurately restores the deformation characteristics of the pipeline under the geological disaster of landslide, and then evaluates the health status of the pipeline; The pipeline deformation data output by the model is compared with the actual monitoring data to correct and optimize the accuracy of the strain-deformation conversion model.

4. The method for collaborative monitoring and early warning of underground gas pipelines and open spaces in landslide hazard areas according to claim 1 is characterized in that: Generate ultra-high-precision digital elevation models, including: The multi-beam laser radar carried by the drone sends laser pulses and measures the time of reflected echoes to obtain high-precision point cloud data on the ground. After noise removal and point cloud registration preprocessing, preliminary ground data is obtained. The hyperspectral imager carried by the drone obtains spectral data of different bands on the ground to distinguish different ground covers; the spectral data is processed by registration with the geographic location and combined with the point cloud data for elevation correction; Combine LiDAR and hyperspectral data, apply Kalman filtering or deep learning data fusion methods to optimize the elevation model, reduce inconsistencies between data, and generate a more accurate digital elevation model; After surface smoothing and noise filtering, the digital elevation model is generated into an ultra-high precision digital elevation model.

5. The method for collaborative monitoring and early warning of underground gas pipelines and open spaces in landslide hazard areas according to claim 1 is characterized in that: Through multi-source data fusion and deep learning algorithms, the geological disaster details of landslide boundary, sliding surface location, crack depth and direction are identified, including: Use image registration technology to align visible light images and DEM data to ensure that all data are processed in the same coordinate system; Use the YOLOv8 model to extract features from visible light images, including the boundaries and cracks of landslides; The visual features extracted by YOLOv8 are fused with the spatial information in the laser point cloud and DEM data, and the elevation information and terrain features provided by the laser point cloud and DEM data are used to assist in identifying the boundaries and crack locations of the landslide; The YOLOv8 model is trained to identify key geological features in the landslide area, including landslide body boundaries, sliding surfaces, crack depths and their orientations; output accurate landslide area feature information; By conducting spatiotemporal sequence analysis on the identified landslide area characteristics, the development trend of the disaster can be predicted to provide decision support for disaster prevention and mitigation.

6. The method for collaborative monitoring and early warning of underground gas pipelines and open spaces in landslide hazard areas according to claim 1 is characterized in that: The advanced hybrid attention convolutional neural network includes: Convolutional layers, used to extract low-level features of pipeline monitoring data; The spatial attention module is used to assign different weights to each input feature map to improve the ability to identify local disaster risk areas; Temporal attention module, which is used to capture important moments in time series data; The fusion layer is used to fuse the outputs of the spatial attention module and the temporal attention module to further enhance the perception ability of the model; The fully connected layer classifies and predicts the fused features to quickly and accurately locate the areas with pipeline disaster risks.

7. The system for monitoring and early warning of underground gas pipelines and open spaces in landslide-prone areas is characterized by: include: Monitoring equipment layer, including densely distributed fiber demodulation equipment and multi-functional drone perception platform; Data processing and analysis layer, including high-precision pipe-soil interaction simulation model and geological hazard identification and assessment system; Early warning release and management level, including intelligent cloud platform early warning system and emergency response command and decision support system.

8. The open-ground coordinated monitoring and early warning system for buried gas pipelines in landslide-prone areas according to claim 7 is characterized in that: Also includes: The edge computing terminal uses a distributed parallel computing architecture to process data from optical distributed fiber demodulation equipment and multi-functional drone perception platforms at high speed.

9. The open-ground coordinated monitoring and early warning system for buried gas pipelines in landslide hazard areas according to claim 7 is characterized in that: Multifunctional UAV perception platform, including: The UAV automatic take-off and landing and autonomous battery replacement ecosystem enables centimeter-level precision landing and rapid battery replacement of UAVs; The real-time data transmission protocol based on 5G communication ensures that the monitoring data is transmitted to the ground control center in a high-speed and stable manner.

10. The open-ground coordinated monitoring and early warning system for buried gas pipelines in landslide hazard areas according to claim 9 is characterized in that: The drone automatic take-off and landing and autonomous battery replacement ecosystem includes: High-precision positioning system, using GPS or GLONASS, ensures centimeter-level precision landing of the drone; Automatic control system to realize automatic take-off, landing and inspection of drones; Efficient energy management system, intelligent battery management system based on municipal facilities and fast battery replacement platform ensure long-term operation of drones.

Images