Highway engineering environment safety intelligent monitoring system based on unmanned aerial vehicle and internet of things

Abstract

This invention discloses an intelligent monitoring system for highway engineering environmental safety based on unmanned aerial vehicles (UAVs) and the Internet of Things (IoT), comprising: a data acquisition module: acquiring initial engineering environmental data through multi-view images from UAVs and IoT sensors; a digital optimization module: obtaining standard engineering environmental data through spatiotemporal benchmark optimization based on multi-source fusion sensing data, obtaining three-dimensional geometrically varied voxels through temporal voxel difference based on the standard engineering environmental data, and obtaining a digital model of the engineering entity through Bayesian physical parameter mapping based on the three-dimensional geometrically varied voxels; a risk assessment module: constructing a hazard propagation path graph network by performing optimized physical simulation based on the digital model of the engineering entity, and obtaining a dynamic risk assessment graph by calculating the risk probability in the hazard propagation path graph through Markov random fields; and an intelligent monitoring module: realizing intelligent monitoring of highway engineering environmental safety through threshold analysis based on the dynamic risk assessment graph.

Description

Technical Field

[0001] This invention relates to the field of next-generation information technology, and in particular to an intelligent monitoring system for environmental safety in highway engineering based on unmanned aerial vehicles and the Internet of Things. Background Technology

[0002] Highway engineering environmental safety monitoring technology has evolved from traditional manual inspections to modern automatic monitoring. In the early stages, safety testing mainly relied on regular manual on-site surveys and simple measuring instruments, which had problems such as low efficiency, strong subjectivity, and limited coverage.

[0003] With the development of sensing technology, IoT-based fixed-point sensor monitoring systems have been gradually adopted. By deploying sensor arrays at key locations such as roadbeds and slopes, continuous acquisition of parameters such as stress, displacement, and vibration has been achieved. In recent years, UAV remote sensing technology has been introduced into the field of engineering monitoring, using aerial imagery for surface change identification and 3D modeling, significantly improving the efficiency and scope of data acquisition. Current technological development shows a trend towards multi-source data fusion, and attempts are being made to initially combine UAV remote sensing data with ground-based IoT monitoring data, forming a preliminary integrated air-space-ground monitoring architecture.

[0004] However, existing highway engineering environmental safety monitoring systems mainly remain at the stage of data collection and simple threshold alarms, lacking the ability to have a holistic understanding and in-depth analysis of the engineering entity's condition. Furthermore, traditional methods are unable to establish a correlation model from local parameter anomalies to overall structural risks, making it impossible to predict the propagation path of hidden dangers and conduct dynamic risk assessments. Ultimately, this leads to a disconnect between monitoring and response, a lack of intelligent decision-making linkage between early warning information and disposal measures, and an inability to form a complete closed-loop management system. Therefore, this paper proposes an intelligent monitoring system for highway engineering environmental safety based on unmanned aerial vehicles (UAVs) and the Internet of Things (IoT). Summary of the Invention

[0005] In order to overcome the above-mentioned defects of the prior art and to achieve the above objectives, the present invention proposes the following technical solution:

[0006] A smart monitoring system for environmental safety in highway engineering based on drones and the Internet of Things includes:

[0007] Data acquisition module: Collects initial engineering environment data through multi-view images from drones and IoT sensors;

[0008] Digital optimization module: Based on multi-source fusion sensing data, spatiotemporal benchmark optimization is performed to obtain standard engineering environment data. Based on the standard engineering environment data, three-dimensional reconstruction is performed through temporal voxel difference to obtain three-dimensional geometrically varied voxels. Based on the three-dimensional geometrically varied voxels, a digital model of the engineering entity is obtained through Bayesian physical parameter mapping.

[0009] Risk assessment module: Based on the digital model of the engineering entity, the module determines the probability of propagation by performing optimized physical simulation, constructs a network of hazard propagation path graphs, and obtains a dynamic risk assessment graph by calculating the risk probability in the hazard propagation path graph through Markov random fields.

[0010] Intelligent monitoring module: Based on dynamic risk assessment maps, threshold analysis is used to achieve intelligent monitoring of environmental safety in highway engineering.

[0011] The process of obtaining standard engineering environmental data includes:

[0012] Based on the initial engineering environment data, and using the satellite time synchronization time of the UAV as the reference axis, the time axis is divided into: The discrete time points of each node are calculated per second, and the engineering environment data after node time alignment is obtained by synchronizing the data through linear interpolation.

[0013] Obtain the physical installation coordinates of each IoT sensor from the initial engineering environment data. and the corresponding visual coordinates in the point cloud Based on physical installation coordinates With corresponding visual coordinates The optimal rigid body transformation rotation matrix is ​​obtained by solving a least-squares problem. Translation vector ;

[0014] Based on the optimal rigid body transformation rotation matrix Translation vector Transform the physical coordinates of all sensors to the visual coordinate system to obtain their coordinates in a unified world coordinate system. This completes the spatial alignment of the data, obtaining standard engineering environment data. .

[0015] The process of obtaining three-dimensional geometrically varied voxels includes:

[0016] Based on standard engineering environmental data Generate 3D dense point cloud ;

[0017] Define a three-dimensional space , three-dimensional space Divided into a uniform voxel grid ;

[0018] 3D dense point cloud at each time point Traverse all voxel meshes The corresponding number of three-dimensional points Set a threshold , Its state at that time Recorded as Otherwise, record as ;

[0019] Compare all 3D points at two consecutive time points and The state, two consecutive states Three-dimensional points that change from 1 to 0 or from 0 to 1 are labeled as variable voxels. All three-dimensional points that change are counted to obtain three-dimensional geometric variable voxels. .

[0020] The process of obtaining a digital model of an engineering entity through Bayesian physical parameter mapping includes:

[0021] Define the core physical parameter λ and anomalous physical events based on prior knowledge. ;

[0022] 3D geometrically varied voxels obtained based on Bayesian algorithm An abnormal physical event occurred. posterior probability Compare the posterior probabilities corresponding to all λ values, use the maximum posterior probability as the confidence level for inference, and iterate through the entire 3D voxel mesh. All voxels, for each voxel Generate a record containing model attributes to obtain a digital model of the engineering entity.

[0023] The process of constructing a network of hazard propagation paths is as follows:

[0024] From the digital twin model of the engineering entity, the spatial locations corresponding to all voxels undergoing geometric changes are extracted as initial risk source nodes in the hazard propagation path network. Key locations of the engineering structure are added as potential risk nodes. Based on the initial risk source nodes and potential risk nodes, an engineering node set is constructed. Project node set Each project node Associate it with its spatial location;

[0025] By optimizing physical simulation, potential hazards can be identified from engineering nodes. propagation to engineering nodes The process will include engineering nodes. propagation to engineering nodes process As an edge in the network of hazard propagation paths, it is also associated with... The corresponding core physical parameters λ and confidence level As an attribute of an edge, it indicates that a directed edge is Traverse the set of project nodes Generate a set of directed edges from all pairs of connected engineering nodes. , ,

[0026] Based on engineering node set With the set of directed edges Constructing a network of hidden danger propagation paths .

[0027] The process of optimizing physical simulations includes:

[0028] The core physical parameter λ includes the cohesion of the slope soil. internal friction angle and the compaction degree of the subgrade soil. ;

[0029] For engineering nodes The physical parameter is the cohesion of the slope soil. Combined with engineering nodes Corresponding internal friction angle Engineering nodes are obtained through the Mohr-Coulomb strength criterion. shear strength ;

[0030] When the project node The associated actual soil stress exceeds the engineering node shear strength Therefore, potential hazards can be identified from engineering nodes. propagation to engineering nodes ;

[0031] For engineering nodes The core physical parameter λ is the internal friction angle of the slope soil. Combined with engineering nodes Corresponding cohesion Calculate engineering nodes using the Mohr-Coulomb strength criterion shear strength ;

[0032] When the project node The associated actual soil stress exceeds the engineering node shear strength Therefore, potential hazards can be identified from engineering nodes. propagation to engineering nodes ;

[0033] For engineering nodes The core physical parameter λ is the compaction degree of the subgrade soil. Combined with engineering nodes Corresponding compaction degree Determine the stress safety threshold at engineering nodes. compaction degree When the stress safety threshold is When the actual stress increment exceeds the stress safety threshold, potential hazards can be identified from engineering nodes. propagation to engineering nodes .

[0034] The process of obtaining a dynamic risk assessment map includes:

[0035] Network of hidden danger propagation paths Consider it as a Markov random field The structural basis, and for Each project node Define a random variable , Indicates a high-risk status. Indicates a low-risk status;

[0036] Based on random variables Define a univariate potential function Based on directed edges Define a bivariate potential function Based on univariate potential function and binary potential function Construct joint probability distribution Based on joint probability distribution The comprehensive risk probability of the engineering node is obtained by performing probability distribution vector transfer. ;

[0037] Set of project nodes Risk probability corresponding to all engineering nodes And use spatial interpolation to calculate the final risk probability from the set of spatial coordinates. Based on the final risk probability Obtain dynamic risk assessment map .

[0038] Set of project nodes Risk probability corresponding to all engineering nodes The process of spatial interpolation calculation using a set of spatial coordinates is represented as follows:

[0039]

[0040] in, Represents a directed edge The weight, Represents the set of project nodes Index of the number of all project nodes Indicates project nodes To the project node Euclidean distance, This represents the final risk probability.

[0041] The present invention has the following beneficial effects:

[0042] 1. By employing two core technologies—temporal voxel difference 3D reconstruction and Bayesian physical parameter mapping—a cognitive bridge connecting appearances and underlying causes has been established. This technology can not only accurately detect the location and extent of geometric deformations such as roadbed settlement and slope cracks, but also use Bayesian inference algorithms to infer the most likely physical mechanisms causing these deformations, such as decreased cohesion and increased pore water pressure. This allows the output digital twin model to clearly indicate the problems and their causes, improving the depth and interpretability of condition assessment and providing a direct basis for precise maintenance.

[0043] 2. It can upgrade safety monitoring from passive response to proactive early warning and closed-loop management. By constructing a network of hazard propagation paths through physical simulation and using Markov random calculation to calculate dynamic risk probabilities, it achieves quantitative prediction of the future development trend of hazards. On this basis, the system adaptively determines risk thresholds and automatically classifies them through extreme value statistics methods. At the same time, it directly drives monitoring resources such as drones to conduct targeted verification of high-risk areas, forming a complete intelligent closed loop of risk assessment, early warning issuance, targeted monitoring, and result feedback. This not only improves the probability of discovering high-risk hazards and the timeliness of handling them, but also significantly reduces reliance on manual labor and subjective misjudgment, achieving an overall leap in monitoring efficiency and safety management capabilities. Attached Figure Description

[0044] Figure 1 This is a system block diagram of the intelligent environmental safety monitoring system for highway engineering based on unmanned aerial vehicles and the Internet of Things proposed in this invention. Detailed Implementation

[0045] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0046] Example 1

[0047] like Figure 1 As shown, the intelligent monitoring system for highway engineering environmental safety based on unmanned aerial vehicles and the Internet of Things proposed in this invention includes:

[0048] Data acquisition module: Collects initial engineering environment data through multi-view images from drones and IoT sensors;

[0049] Raw data of the engineering environment can be acquired in parallel using drones and IoT sensors.

[0050] Multi-view image acquisition by drone: A grid-like flight path is used, and the drone's flight altitude is set to... Meters, ensuring image resolution reaches Meters per pixel, meeting engineering monitoring accuracy; overlap between adjacent flight paths set to [value missing]. Drones equipped with Aerial camera with multiple perspectives, vertical overhead shooting, Side shots and horizontal side shots are used to photograph the highway engineering area, including... Mi Roadbed Section Tall The slope section of meters The entire construction area of ​​square meters was photographed. It can capture one frame of image per second, and each frame carries the drone's real-time positioning data and satellite time stamp;

[0051] IoT sensor data acquisition: Deploy sensor arrays according to engineering specifications, with each sensor at a certain interval on the top of the slope. Mi deployment Each displacement sensor is located at a distance of [number] from the edge of the roadbed. Mi deployment Each stress sensor is deployed at the top and bottom of the construction support structure. There are 1 stress sensor, and all sensors continuously collect data at a sampling frequency of [frequency missing]. Each data point carries preset deployment geographic coordinates and timestamps, and is executed once per second.

[0052] Initial engineering environment data carrying timestamps and location information is obtained by acquiring multi-view images from drones and data from IoT sensors.

[0053] Digital optimization module: Based on multi-source fusion sensing data, spatiotemporal benchmark optimization is performed to obtain standard engineering environment data. Based on the standard engineering environment data, three-dimensional reconstruction is performed through temporal voxel difference to obtain three-dimensional geometrically varied voxels. Based on the three-dimensional geometrically varied voxels, a digital model of the engineering entity is obtained through Bayesian physical parameter mapping.

[0054] The initial engineering environment data was aligned to a spatiotemporal reference. First, the time axis was divided using the satellite time synchronization of the UAV as the reference axis. Discrete time points in seconds / nodes, for each time node If the time point of a certain data is between and , Data synchronization is achieved using linear interpolation. Time data is , Time data is , With node time The time difference is , and The time difference is Then the project environment data after node time alignment is:

[0055]

[0056] After completing the node time alignment, perform spatial alignment.

[0057] Obtain the physical installation coordinates of each IoT sensor from the initial engineering environment data. The corresponding visual coordinates of the image in the point cloud are obtained through image recognition and intersection with the foreground. Based on physical installation coordinates With corresponding visual coordinates The correspondence is determined by solving the least squares problem to find the optimal rigid body transformation rotation matrix. Translation vector The goal is to minimize the sum of squared errors between the transformed coordinates and the visual coordinates, expressed as:

[0058]

[0059] The optimal rigid body transformation rotation matrix obtained by application Translation vector Transform the physical coordinates of all sensors to the visual coordinate system to obtain their coordinates in a unified world coordinate system. This completes the spatial alignment of the data;

[0060] Finally, after aligning the initial engineering environment data with temporal and spatial references, standard engineering environment data is obtained. ;

[0061] The process of obtaining three-dimensional geometrically transformed voxels through three-dimensional reconstruction is as follows:

[0062] For each monitoring time point Using standard engineering environmental data Drone imagery at that moment and their corresponding poses A three-dimensional dense point cloud at that moment was generated using multi-view stereo vision technology. ;

[0063] Define an axis-aligned three-dimensional cuboid that completely encloses the entire engineering area. The space is divided into a uniform voxel grid. Each voxel is uniquely identified by its index, representing a fixed region in space;

[0064] For the 3D dense point cloud at each time point Traverse all voxel meshes For each 3D point, the number of points falling into the corresponding voxel mesh is counted. Internally, denoted as the voxel in Number of points at time Set a threshold ,if Then it is believed that the voxel is in It is constantly occupied by the surface of the engineering entity, and its state Recorded as Otherwise, record as ;

[0065] For any three-dimensional point, compare its values ​​at two consecutive time points. and The state, if the state changes, that is, the state Depend on Become , or by Become This indicates that its voxels have undergone geometric changes. The 3D point is marked as a variable voxel, and the 3D geometric variable voxels are obtained by counting all 3D points that have changed. ;

[0066] The process of obtaining a digital model of an engineering entity through Bayesian physical parameter mapping is as follows:

[0067] Determine the core physical parameter λ and abnormal physical events of the engineering entity using prior knowledge. ;

[0068] Among them, the core physical parameter λ is set as the cohesion of the slope soil. internal friction angle and the compaction degree of the subgrade soil. And for the set of changing voxels Each voxel in ,set up This indicates that the core physical parameter λ corresponding to the voxel has become abnormal, i.e., an abnormal physical event.

[0069] 3D geometrically varied voxels obtained based on Bayesian algorithm An abnormal physical event occurred. The probability, i.e., the posterior probability. This indicates that the status has been viewed. Depend on Become In this case, the reason is The probability of; ;

[0070] in, Represents core physical parameters The general probability of anomalies is obtained statistically from long-term monitoring data. Indicates the same voxel All They are the same constant, so it doesn't affect the comparison. Indicates the core physical parameters Under abnormal conditions, voxel v undergoes geometric changes. The probability, This indicates that the voxel v undergoes a geometric change. The total probability;

[0071] Compare the posterior probabilities corresponding to all λ values, and use the highest posterior probability as the confidence level for the inference, i.e., for the changing voxel set. Each voxel in Iterate through all predefined core physical parameters Calculate their posterior probabilities respectively. Compare the posterior probability values ​​corresponding to all λ, and denote the physical parameter with the highest posterior probability as the maximum posterior probability value. And as the confidence level for the inference;

[0072] Then, digital twin instantiation is performed, traversing the entire 3D voxel mesh. All voxels, for each voxel Generate a record containing the model data. The model attributes include:

[0073] Spatial location: the three-dimensional coordinates of the voxel center;

[0074] Geometric state: , Indicates idle / no entity. It indicates that it is occupied by the engineering entity;

[0075] Geometric transformation: This indicates that the status has changed during the monitoring period;

[0076] Abnormal physical events: In voxel mesh Core physical parameters occurred in The abnormality;

[0077] Inference confidence level: , range between;

[0078] By integrating the records of all voxels, a structured model is formed, which is the digital model of the engineering entity. It not only records the voxels of geometric changes, but also explains the inferred physical parameter anomalies and their confidence levels.

[0079] Specifically, the digital model of the engineering entity is created by traversing the entire voxel mesh. For each voxel Create a data record containing its spatial coordinates and geometric state. Change indicators, inferred anomalous physical events, and confidence levels By organizing the records of all these voxels according to their spatial relationships, the final digital model of the engineering entity is formed.

[0080] Risk assessment module: Based on the digital model of the engineering entity, the module determines the probability of propagation by performing optimized physical simulation, constructs a network of hazard propagation path graphs, and obtains a dynamic risk assessment graph by calculating the risk probability in the hazard propagation path graph through Markov random fields.

[0081] First, construct a network diagram showing the propagation path of potential hazards:

[0082] Extract all voxels marked as having undergone geometric changes from the digital twin model of the engineering entity. The spatial locations corresponding to these geometrically changed voxels are used as the initial risk source nodes in the hazard propagation path network.

[0083] At the same time, key locations of the engineering structure, such as the slope toe, roadbed joints, and anchorage points of the support structure, are added as potential risk nodes.

[0084] All initial risk source nodes and potential risk nodes together constitute the set of engineering nodes in the hazard propagation path graph network. Each project node Associate its spatial location with the corresponding physical parameter type λ of the anomaly and its corresponding confidence level. ;

[0085] The process of performing optimized physical simulations to determine the probability of propagation includes:

[0086] For the set of engineering nodes Any two spatially adjacent engineering nodes and Based on physical simulation, potential hazards are identified from engineering nodes. spread to The propagation capability depends on the engineering nodes. The associated core physical parameter λ type is used to perform the corresponding physical simulation to obtain the edge set of the hidden danger propagation path graph network;

[0087] For adjacent engineering nodes and Based on engineering nodes The associated core physical parameter λ type is used to perform the corresponding optimized physical simulation:

[0088] If the project node The physical parameter is the cohesion of the slope soil. Combined with engineering nodes Corresponding internal friction angle Calculate engineering nodes using the Mohr-Coulomb strength criterion shear strength :

[0089]

[0090] in, Indicates project nodes Corresponding cohesion , Indicates project nodes Corresponding internal friction angle , Indicates stress;

[0091] If the project node The associated actual soil stress exceeds the engineering node shear strength Therefore, potential hazards can be identified from engineering nodes. propagation to engineering nodes ;

[0092] If the project node The core physical parameter λ is the internal friction angle of the slope soil. Combined with engineering nodes Corresponding cohesion Similarly, the engineering nodes are calculated using the Mohr-Coulomb strength criterion. shear strength :

[0093]

[0094] in, Indicates project nodes The corresponding cohesion, Indicates project nodes The corresponding internal friction angle, Indicates stress;

[0095] If the project node The associated actual shear stress corresponds to soil stress exceeding the engineering node. shear strength Therefore, potential hazards can be identified from engineering nodes. propagation to engineering nodes ;

[0096] If the project node The core physical parameter λ is the compaction degree of the subgrade soil. Then combine the engineering nodes Corresponding compaction degree Determine the stress safety threshold at engineering nodes. compaction degree When the stress safety threshold is If the actual stress increment exceeds the stress safety threshold, the potential hazard can be identified from the engineering nodes. propagation to engineering nodes ;

[0097] By optimizing physical simulation, potential hazards can be identified from engineering nodes. propagation to engineering nodes The process will include engineering nodes. propagation to engineering nodes process As an edge in the network of hazard propagation paths, it is also associated with... The corresponding core physical parameters λ and confidence level As an attribute of an edge, it indicates that a directed edge is ;

[0098] Traverse the set of project nodes For all possible pairs of engineering nodes that can form a connection, generate a set of directed edges using the physical simulation rules described above. , Ultimately, a network of potential hazard propagation paths was obtained. ;

[0099] The process of obtaining a dynamic risk assessment map includes:

[0100] Network of hidden danger propagation paths Consider it as a Markov random field The structural basis of graph networks The structure directly serves as The topology, for Each project node Define a random variable Its value is defined as:

[0101] Indicates a high-risk status;

[0102] Indicates a low-risk status;

[0103] Based on random variables Define a univariate potential function based on directed edges. Define the bivariate potential function as follows:

[0104] Univariate potential function : Reflects the risk tendency and risk confidence level of the project node itself. Directly related When, the univariate potential function equal , When, the univariate potential function equal ;

[0105] Binary potential function Defined when there is a directed edge. The propagation correlation of risk on adjacent engineering nodes is defined as follows: ,in It is an indicator function; it is true when the two states are the same. Different as β is the potential function parameter. Directed edges defined by prior knowledge The weights;

[0106] Furthermore, the joint probability distribution of the entire Markov random field is composed of the product of all potential functions, and the goal is to calculate that the marginal probability of each engineering node being in a high-risk state is equal to 1.

[0107] Construct joint probability distribution ,entire The joint probability distribution is the product of all univariate and bivariate potential functions, then normalized by a normalization constant. The adjustment is made to ensure that the sum of the probabilities of all possible state combinations is 1, and its mathematical expression is:

[0108]

[0109] in, This represents the marginal probability that each project node is in a high-risk state;

[0110] Specifically, joint probability distribution Includes the probability distribution vector of each engineering node. , And each from the engineering node propagation to engineering nodes The probability distribution vector, whose content integrates the engineering nodes Its own state tendency (represented by its potential function) and engineering nodes From the engineering nodes The probability distribution vector received by all other neighboring project nodes from the previous round;

[0111] Based on joint probability distribution The belief propagation algorithm is used to obtain the comprehensive risk probability of engineering nodes by passing probability distribution vectors along the network's connection edges. The process is as follows:

[0112] Obtain the corresponding probability distribution vector on all connected edges until the values ​​of all probability distribution vectors in the network no longer change significantly, which means that the algorithm has converged to a stable state.

[0113] After the probability distribution vector converges, for each engineering node in the network Calculate the overall risk probability of it being in a high-risk state, denoted as . The process is as follows:

[0114] Input the network diagram of the hidden danger transmission path. Network of hidden danger propagation paths Each directed edge in probability distribution vector on Assign initial values, ;

[0115] For each directed edge in the graph Based on the probability distribution vectors transmitted from other neighboring nodes to the project node at the current moment, and the project node's own potential function... Get from project nodes propagation to engineering nodes The new probability distribution vector , ;

[0116] Repeat this operation for all edges to complete one round of probability distribution vector update. Repeat this iteration multiple times until the change in all probability distribution vectors is less than the preset threshold, at which point convergence is determined.

[0117] After the probability distribution vector converges, for each engineering node Collect the new probability distribution vectors sent by all its neighboring nodes. Project nodes Its own potential function The belief value is obtained by multiplying it by all the received probability distribution vectors. For all belief values Perform normalization processing to ensure Obtain project nodes Overall risk probability , For the random variables corresponding to the project nodes;

[0118] Based on the comprehensive risk probability of each project node The set of engineering nodes and their corresponding known three-dimensional spatial coordinates. Risk probability corresponding to all engineering nodes And spatial interpolation calculations are performed using a set of spatial coordinates:

[0119]

[0120] in, Represents a directed edge The weight, Represents the set of project nodes Index of the number of all project nodes Indicates project nodes To the project node Euclidean distance, This represents the final risk probability.

[0121] Specifically, due to project milestones The risk is only distributed in key locations of the engineering structure. In order to obtain a risk distribution that covers the entire continuous engineering area, it is necessary to estimate the uncovered areas.

[0122] Finally, based on the interpolation results, a three-dimensional regular grid data field covering the entire target area is generated.

[0123] In this 3D regular grid data field, each grid cell contains a spatial coordinate and an interpolated risk probability value. This three-dimensional regular grid risk probability field is the dynamic risk assessment diagram. ;

[0124] Specifically, generate dynamic risk assessment maps. It can use visualization technology to map risk probability values ​​to colors, such as blue to red to represent low to high, and render dynamic risk assessment maps using methods such as volume drawing or isosurfaces. It can interactively display the spatial distribution of risks and can be overlaid with the original engineering model to achieve precise risk positioning.

[0125] Intelligent monitoring module: Based on dynamic risk assessment maps, threshold analysis is used to achieve intelligent monitoring of environmental safety in highway engineering.

[0126] First, integrate the dynamic risk assessment graph. It automatically identifies the coordinates of high-risk areas marked on the map and their risk levels, and immediately sends alarm information containing the specific location, risk type and risk value to the monitoring center and mobile terminals according to the preset red warning threshold.

[0127] Meanwhile, the system automatically generates drone inspection mission instructions, plans key flight routes for high-risk areas, sets high-resolution image acquisition parameters, and after receiving the command, the drone performs multi-angle fine scanning of the target area and transmits the collected real-time data back. Monitoring personnel judge the risk situation based on the on-site images and data and issue disposal instructions through the system.

[0128] All early warning, inspection, and handling records are automatically archived, forming a complete monitoring and handling closed loop. The system continuously compares historical risk assessment maps with the latest monitoring results to dynamically optimize early warning thresholds and inspection strategies, thereby achieving intelligent adaptive cyclic monitoring of highway engineering environmental safety.

[0129] In the application, several formulas are calculated by removing dimensions and taking their numerical values. The formulas are established by collecting a large amount of data and simulating the most recent real situation. Some coefficients or weights in the formulas are set by those skilled in the art according to the actual situation, so they will not be elaborated here.

[0130] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0131] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A smart monitoring system for environmental safety in highway engineering based on unmanned aerial vehicles (UAVs) and the Internet of Things (IoT), characterized in that: include: Data acquisition module: Collects initial engineering environment data through multi-view images from drones and IoT sensors; Digital optimization module: Based on multi-source fusion sensing data, spatiotemporal benchmark optimization is performed to obtain standard engineering environment data. Based on the standard engineering environment data, three-dimensional reconstruction is performed through temporal voxel difference to obtain three-dimensional geometrically varied voxels. Based on the three-dimensional geometrically varied voxels, a digital model of the engineering entity is obtained through Bayesian physical parameter mapping. The process of obtaining a digital model of an engineering entity through Bayesian physical parameter mapping includes: Define the core physical parameter λ and anomalous physical events based on prior knowledge. ; 3D geometrically varied voxels obtained based on Bayesian algorithm An abnormal physical event occurred. posterior probability Compare the posterior probabilities corresponding to all λ values, use the maximum posterior probability as the confidence level for inference, and iterate through the entire 3D voxel mesh. All voxels, for each voxel Generate a record containing model attributes to obtain a digital model of the engineering entity; Risk assessment module: Based on the digital model of the engineering entity, the module determines the probability of propagation by performing optimized physical simulation, constructs a network of hazard propagation path graphs, and obtains a dynamic risk assessment graph by calculating the risk probability in the hazard propagation path graph through Markov random fields. The process of constructing a network of hazard propagation paths is as follows: From the digital twin model of the engineering entity, the spatial locations corresponding to all voxels undergoing geometric changes are extracted as initial risk source nodes in the hazard propagation path network. Key locations of the engineering structure are added as potential risk nodes. Based on the initial risk source nodes and potential risk nodes, an engineering node set is constructed. Project node set Each project node Associate it with its spatial location; By optimizing physical simulation, potential hazards can be identified from engineering nodes. propagation to engineering nodes The process will include engineering nodes. propagation to engineering nodes process As an edge in the network of hazard propagation paths, it is also associated with... The corresponding core physical parameters λ and confidence level As an attribute of an edge, it indicates that a directed edge is Traverse the set of project nodes Generate a set of directed edges from all pairs of connected engineering nodes. , , Based on engineering node set With the set of directed edges Constructing a network of hidden danger propagation paths ; Intelligent monitoring module: Based on dynamic risk assessment maps, threshold analysis is used to achieve intelligent monitoring of environmental safety in highway engineering.

2. The intelligent monitoring system for highway engineering environmental safety based on unmanned aerial vehicles and the Internet of Things as described in claim 1, characterized in that, The process of obtaining standard engineering environmental data includes: Based on the initial engineering environment data, and using the satellite time synchronization time of the UAV as the reference axis, the time axis is divided into: The discrete time points of each node are calculated per second, and the engineering environment data after node time alignment is obtained by synchronizing the data through linear interpolation. Obtain the physical installation coordinates of each IoT sensor from the initial engineering environment data. and the corresponding visual coordinates in the point cloud Based on physical installation coordinates With corresponding visual coordinates The optimal rigid body transformation rotation matrix is ​​obtained by solving a least-squares problem. Translation vector ; Based on the optimal rigid body transformation rotation matrix Translation vector Transform the physical coordinates of all sensors to the visual coordinate system to obtain their coordinates in a unified world coordinate system. This completes the spatial alignment of the data, obtaining standard engineering environment data. .

3. The intelligent monitoring system for highway engineering environmental safety based on unmanned aerial vehicles and the Internet of Things as described in claim 2, characterized in that, The process of obtaining three-dimensional geometrically varied voxels includes: Based on standard engineering environmental data Generating 3D dense point clouds from point clouds ; Define a three-dimensional space , three-dimensional space Divided into a uniform voxel grid ; 3D dense point cloud at each time point Traverse all voxel meshes The corresponding number of three-dimensional points Set a threshold , Its state at that time Recorded as Otherwise, record as ; Compare all 3D points at two consecutive time points and The state, two consecutive states Three-dimensional points that change from 1 to 0 or from 0 to 1 are labeled as variable voxels. All three-dimensional points that change are counted to obtain three-dimensional geometric variable voxels. .

4. The intelligent monitoring system for highway engineering environmental safety based on unmanned aerial vehicles and the Internet of Things as described in claim 1, characterized in that, The process of optimizing physical simulations includes: The core physical parameter λ includes the cohesion of the slope soil. internal friction angle and the compaction degree of the subgrade soil. ; For engineering nodes The physical parameter is the cohesion of the slope soil. Combined with engineering nodes Corresponding internal friction angle Engineering nodes are obtained through the Mohr-Coulomb strength criterion. shear strength ; When the project node The associated actual soil stress exceeds the engineering node shear strength Therefore, potential hazards can be identified from engineering nodes. propagation to engineering nodes ; For engineering nodes The core physical parameter λ is the internal friction angle of the slope soil. Combined with engineering nodes Corresponding cohesion Calculate engineering nodes using the Mohr-Coulomb strength criterion shear strength ; When the project node The associated actual soil stress exceeds the engineering node shear strength Therefore, potential hazards can be identified from engineering nodes. propagation to engineering nodes ; For engineering nodes The core physical parameter λ is the compaction degree of the subgrade soil. Combined with engineering nodes Corresponding compaction degree Determine the stress safety threshold at engineering nodes. compaction degree When the stress safety threshold is When the actual stress increment exceeds the stress safety threshold, potential hazards can be identified from engineering nodes. propagation to engineering nodes .

5. The intelligent monitoring system for highway engineering environmental safety based on unmanned aerial vehicles and the Internet of Things as described in claim 1, characterized in that, The process of obtaining a dynamic risk assessment map includes: Network of hidden danger propagation paths Consider it as a Markov random field The structural basis, and for Each project node Define a random variable , Indicates a high-risk status. Indicates a low-risk status; Based on random variables Define a univariate potential function Based on directed edges Define a bivariate potential function Based on univariate potential function and binary potential function Construct joint probability distribution Based on joint probability distribution The comprehensive risk probability of the engineering node is obtained by performing probability distribution vector transfer. ; Set of project nodes Risk probability corresponding to all engineering nodes And use spatial interpolation to calculate the final risk probability from the set of spatial coordinates. Based on the final risk probability Obtain dynamic risk assessment map .

6. The intelligent monitoring system for highway engineering environmental safety based on unmanned aerial vehicles and the Internet of Things as described in claim 5, characterized in that, Set of project nodes Risk probability corresponding to all engineering nodes The process of spatial interpolation calculation using a set of spatial coordinates is represented as follows: in, Represents a directed edge The weight, Represents the set of project nodes Index of the number of all project nodes Indicates project nodes To the project node Euclidean distance, This represents the final risk probability.

Images