An intelligent monitoring system and method for subgrade compaction degree based on a UAV group

Abstract

The application provides a kind of intelligent monitoring system and method for subgrade compaction degree based on UAV group, relating to the technical field of intelligent monitoring of road engineering.The system is composed of UAV group, multi-source sensing module, cooperative control module, data transmission module, ground control center and calibration module, through the subarea cooperative operation of UAV group, synchronous acquisition of subgrade GNSS positioning data, ground penetrating radar GPR electromagnetic wave data, infrared thermal imaging data and inertial measurement data, after time-space alignment and noise removal, input into the compaction degree inversion model based on deep learning, realize the non-contact, full-coverage, high-precision monitoring of subgrade compaction degree.Practice verification shows that the monitoring coverage efficiency is more than 10 times higher than that of traditional methods, the compaction degree monitoring error is less than or equal to 3%, the weak area of subgrade can be identified in real time, guide the dynamic adjustment of compaction operation, adapt to various subgrade construction scenes such as expressway, railway and municipal road, significantly improve the engineering quality and construction efficiency.

Description

Technical Field

[0001] This invention relates to the field of intelligent monitoring technology for road engineering, specifically to an intelligent monitoring system and method for roadbed compaction based on unmanned aerial vehicle (UAV) swarms. Background Technology

[0002] Subgrade compaction is a core indicator of road engineering quality, directly affecting the road's load-bearing capacity, stability, and service life. Traditional methods for monitoring subgrade compaction mainly include the sand cone method, ring cutter method, and nuclear analyzer method, but they have several significant drawbacks: Extremely low efficiency: The single-point sampling monitoring mode takes 30 to 60 minutes to monitor each point, and large-area roadbed monitoring takes a long time (e.g., 10km of roadbed takes several days), which is difficult to meet the construction progress requirements. Highly destructive: Sand cone method and ring cutter method require excavation of roadbed soil samples, which damages the already compacted roadbed structure and can easily lead to later settlement; Incomplete coverage: Sampling monitoring can only cover 1% to 5% of the total roadbed area, which may easily miss weak areas and lead to potential roadbed quality problems; Data lag: Monitoring data needs to be manually compiled and analyzed, and cannot be fed back to the construction party in real time, making it difficult to dynamically adjust compaction parameters; Poor environmental adaptability: The nuclear instrument method carries radiation risks, and the sand-filling method cannot be operated due to rainy weather or muddy environments.

[0003] Existing drone monitoring technologies mostly focus on monitoring the flatness and settlement deformation of roadbed surfaces, but fail to achieve accurate inversion of compaction degree; single drone monitoring has the problems of limited coverage and insufficient efficiency.

[0004] Therefore, developing a collaborative monitoring system and method based on UAV swarms to achieve efficient, non-contact, full-coverage, and high-precision monitoring of roadbed compaction has become an urgent need in the field of intelligent construction of road engineering. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides an intelligent monitoring system and method for roadbed compaction based on unmanned aerial vehicle (UAV) swarms. Addressing the pain points of traditional roadbed compaction monitoring methods (sand cone method, nuclear analyzer method) such as low efficiency, high destructiveness, incomplete coverage, and data lag, this invention innovatively constructs an integrated technical system of "UAV swarm collaborative perception - multi-source data fusion - AI compaction inversion - real-time closed-loop feedback." This solves the problems of existing UAV monitoring technologies, which often focus on monitoring roadbed surface smoothness and settlement deformation, failing to achieve accurate compaction inversion, and the limited coverage and insufficient efficiency of single-UAV monitoring.

[0006] To achieve the above objectives, the present invention provides the following technical solution: A roadbed compaction intelligent monitoring system based on UAV swarms includes a UAV swarm, a multi-source sensing module, a collaborative control module, a data transmission module, a ground control center, and an on-site calibration module; The drone swarm consists of one master drone and 3 to 8 slave drones. The master drone is equipped with a collaborative control unit, a global positioning module, and a wireless communication module, while the slave drones are equipped with a multi-source sensing module and a local storage unit. The master drone is responsible for path planning, operation coordination, and data aggregation, while the slave drones synchronously collect multi-source data according to the partitioned path. The master and slave drones achieve low-latency communication through 5G + BeiDou. The multi-source sensing module synchronously collects multi-source roadbed data from the UAV along the path, including GNSS positioning data, GPR electromagnetic wave data, infrared thermal imaging data, and IMU attitude data. The collaborative control module, including a path planning algorithm unit, a work coordination unit, and an obstacle avoidance unit, automatically divides the roadbed design drawings into sections ranging from 0.05 to 0.2 km. 2 The monitoring zones are ordered in the order of "critical road sections (such as bridge and culvert connection sections) first, followed by ordinary road sections". The drones operate in parallel paths, and the obstacle avoidance unit avoids obstacles such as construction machinery and trees in real time through lidar. The data transmission module includes a 5G communication module, a BeiDou short message module, and a local cache unit. It prioritizes real-time data transmission via the 5G network, switches to BeiDou short message in areas without 5G signal, and stores data in the local cache unit to prevent data loss. The ground control center includes a data processing server, an AI inversion server, a visualization terminal, and a construction docking interface. It is used to receive multi-source data and complete the fusion processing, run a deep learning inversion model to output a compaction distribution map, visualize the results and push them to the construction terminal, and support data traceability and historical comparison. The on-site calibration module includes standard compacted blocks (with three levels of compaction: 95%, 97%, and 98%), a portable moisture meter, and a soil type analyzer, used before monitoring and every 2km of operation. 2 Then, calibration is performed to correct sensor errors and inversion model parameters, ensuring monitoring accuracy.

[0007] Furthermore, in the drone swarm, the master drone is equipped with a collaborative control unit and a global positioning module, while the slave drones are equipped with a multi-source perception module. The master and slave drones achieve collaborative operation through wireless communication, with a positioning accuracy of ≤5cm (RTK-GNSS), an operating altitude of 5~15m above the roadbed surface, and a flight speed of 3~8m / s.

[0008] Furthermore, the multi-source sensing module includes a ground-penetrating radar, a GNSS positioning unit, an infrared thermal imager, and an IMU inertial measurement unit; Ground-penetrating radar (GPR): Center frequency 500~1500MHz, detection depth 0.3~1.5m, capturing changes in electromagnetic wave reflection caused by differences in soil density in the shallow subgrade layer (0~1.2m); GNSS positioning unit: Supports RTK / PPP-RTK, positioning accuracy ±1cm, records the three-dimensional coordinates of each set of monitoring data to ensure data spatial registration accuracy; Infrared thermal imager: Temperature measurement range -20~80℃, accuracy ±0.5℃, captures the temperature gradient generated by soil friction during compaction. The higher the degree of compaction, the more uniform the temperature gradient. IMU (Inertial Measurement Unit): Sampling frequency 100~200Hz, measures UAV attitude, such as pitch angle, roll angle, and yaw angle, and calibrates sensor data for errors caused by flight attitude fluctuations.

[0009] Furthermore, this intelligent monitoring method for roadbed compaction based on unmanned aerial vehicle (UAV) swarms includes the following steps: Step 1: Monitoring Pretreatment The ground control center imports the roadbed construction design drawings, divides the monitoring zones, configures the drone swarm, and completes sensor calibration through the on-site calibration module; Step 2: Collaborative Monitoring by Unmanned Aerial Vehicle (UAV) Swarm The main UAV plans the operation path for each zone, and the secondary UAVs synchronously collect multi-source data of the roadbed according to the path. The main UAV coordinates the operation progress in real time and avoids monitoring blind spots. Step 3: Multi-source data fusion The data transmission module transmits data to the ground control center in real time, and after spatiotemporal alignment, noise removal, and data registration, a standardized dataset is obtained. Step 4: Compaction Degree AI Inversion The standardized dataset and auxiliary parameters such as soil type, moisture content, and number of compaction passes are input into the deep learning inversion model to output a three-dimensional compaction distribution map of the subgrade. Step 5: Results Feedback and Optimization The ground control center visualizes the compaction results, identifies weak areas, and pushes them to the construction terminal in real time to guide the adjustment of compaction parameters. If the monitoring error exceeds the threshold, a second round of monitoring by a drone swarm will be initiated to form a closed-loop optimization.

[0010] Furthermore, the collaborative monitoring strategy in step two is as follows: the master UAV plans its path according to "grid partitioning + priority sorting," while the slave UAVs adopt a "parallel operation + edge supplementary monitoring" mode. The monitoring overlap rate of adjacent UAVs is ≥10%, ensuring no monitoring blind spots, and a single slave UAV can monitor an area of ​​≥5km² per day. 2 .

[0011] Furthermore, the data fusion in step three includes: Spatiotemporal alignment, based on GNSS timestamps and IMU attitude data calibration; Noise removal is achieved using a wavelet transform + Kalman filter algorithm. Data registration associates GPR data with infrared data using GNSS coordinates; The standardized dataset format is ASCII, and the data resolution is ≤0.1m×0.1m.

[0012] Furthermore, the deep learning inversion model is a CNN-LSTM hybrid model. The input features include GPR reflected wave amplitude, phase difference, infrared temperature gradient, and roadbed surface smoothness. Combined with auxiliary parameters such as soil type, moisture content, and number of compaction passes, the model training set contains ≥1000 sets of measured roadbed compaction data, and the prediction error is ≤3%.

[0013] Furthermore, the on-site calibration module includes a standard compacted block, a moisture content meter, and a soil type analyzer, with a calibration frequency of every 2 km monitored. 2 Alternatively, the operation can be performed when the ambient temperature changes by ≥5℃ to ensure sensor accuracy and the reliability of the inversion model.

[0014] Furthermore, the ground control center supports three-dimensional visualization of compaction degree, automatic marking of weak areas, and historical data comparison and analysis functions. The data storage capacity is ≥10TB, and it supports docking with the roadbed construction management platform to realize full-process data traceability from monitoring to construction to acceptance.

[0015] Furthermore, the monitoring method is applicable to the monitoring of roadbed construction for highways, railways, municipal roads, airport runways, etc., and is compatible with different roadbed fillers such as silty soil, clayey soil, sandy soil, and soil-rock mixture. It can operate normally in environments with temperatures ranging from -10 to 45℃ and wind speeds ≤ 6.

[0016] This invention provides an intelligent monitoring system and method for roadbed compaction based on unmanned aerial vehicle (UAV) swarms. It has the following beneficial effects: 1. This invention provides an intelligent monitoring system and method for roadbed compaction based on unmanned aerial vehicle (UAV) swarms. It employs a "master-slave UAV" architecture, operates in parallel by grid partitioning, has an overlap rate of ≥10% with no blind spots, and improves monitoring efficiency by more than 10 times compared to traditional methods, achieving daily coverage of ≥5km. 2 It integrates multiple sensors including GPR, GNSS, infrared thermal imaging, and IMU, and comprehensively captures the physical characteristics of the roadbed surface and shallow layers through spatiotemporal alignment and noise removal technology, providing multi-dimensional data support for compaction inversion.

[0017] 2. This invention provides an intelligent monitoring system and method for roadbed compaction based on UAV swarms. It constructs a CNN-LSTM hybrid deep learning model, integrates auxiliary parameters such as soil type, moisture content, and number of compaction passes, adapts to different roadbed fill materials, and achieves a monitoring error of ≤3%. Compared with a single sensor, the inversion accuracy is improved by 40%. The monitoring data is transmitted to the ground control center in real time, and weak areas are automatically marked and pushed to the construction terminal to guide the compaction machinery to adjust the tonnage and number of passes, realizing dynamic optimization of "monitoring-adjustment-re-monitoring".

[0018] 3. This invention provides an intelligent monitoring system and method for roadbed compaction based on UAV swarms. It does not require contact with the roadbed surface throughout the process, thus avoiding damage to the compacted structure and avoiding the risk of later settlement caused by traditional methods. At the same time, it protects the construction progress, adapts to complex weather and terrain, and has strong environmental adaptability. Attached Figure Description

[0019] Figure 1 This is a diagram illustrating the overall architecture of the intelligent monitoring system for roadbed compaction based on unmanned aerial vehicle (UAV) swarms according to the present invention. Figure 2 This is a flowchart of the intelligent monitoring method for roadbed compaction based on unmanned aerial vehicle (UAV) swarms according to the present invention. Figure 3 This is a schematic diagram of the three-dimensional visualization of roadbed compaction degree according to the present invention. Detailed Implementation

[0020] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. Furthermore, it should be noted that, for ease of description, the accompanying drawings show only the parts relevant to the present invention, and not all of the structures.

[0021] In the description of this invention, unless otherwise explicitly specified and limited, the terms "connected," "linked," and "fixed" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0022] The technical solution of the present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0023] like Figures 1-3 As shown, this embodiment of the invention provides an intelligent monitoring system for roadbed compaction based on a drone swarm, including a drone swarm, a multi-source sensing module, a collaborative control module, a data transmission module, a ground control center, and a field calibration module; Drone swarms: Core components: 1 master drone + 3 to 8 slave drones. The master drone is equipped with a collaborative control unit, a global positioning module, and a wireless communication module; the slave drones are equipped with a multi-source sensing module and a local storage unit. Functionality: The master drone is responsible for path planning, operation coordination, and data aggregation, while the slave drones synchronously collect multi-source data according to the partitioned path. The master and slave drones achieve low-latency communication through 5G + BeiDou.

[0024] Multi-source sensing module: It includes ground-penetrating radar, GNSS positioning unit, infrared thermal imager, and IMU inertial measurement unit; Ground-penetrating radar (GPR): Center frequency 500~1500MHz, detection depth 0.3~1.5m, capturing changes in electromagnetic wave reflection caused by differences in soil density in the shallow subgrade layer (0~1.2m); GNSS positioning unit: Supports RTK / PPP-RTK, positioning accuracy ±1cm, records the three-dimensional coordinates of each set of monitoring data to ensure data spatial registration accuracy; Infrared thermal imager: Temperature measurement range -20~80℃, accuracy ±0.5℃, captures the temperature gradient generated by soil friction during compaction. The higher the degree of compaction, the more uniform the temperature gradient. IMU (Inertial Measurement Unit): Sampling frequency 100~200Hz, measures UAV attitude, such as pitch angle, roll angle, and yaw angle, and calibrates sensor data for errors caused by flight attitude fluctuations.

[0025] Collaborative control module: Core components: path planning algorithm unit, job coordination unit, obstacle avoidance unit; Functionality: After importing roadbed design drawings, automatically divide the roadbed into sections ranging from 0.05 to 0.2 km. 2 The monitoring zones are ordered in order of "critical road sections first (such as bridge and culvert connection sections) and then ordinary road sections". The drones operate in parallel paths, and the obstacle avoidance unit avoids obstacles such as construction machinery and trees in real time through lidar.

[0026] Data transmission module: Core components: 5G communication module, BeiDou short message module, local cache unit; Functionality: Prioritizes real-time data transmission via 5G network; switches to BeiDou short message service in areas without 5G signal; local cache unit stores data to prevent data loss.

[0027] Ground Control Center: Core components: data processing server, AI inversion server, visualization terminal, construction interface; Functionality: Receives multi-source data and performs fusion processing, runs a deep learning inversion model to output a compaction degree distribution map, visualizes the results and pushes them to the construction terminal, and supports data traceability and historical comparison.

[0028] Field calibration module: Core components: standard compacted blocks (95% compaction), portable moisture meter, soil type analyzer; Functionality: Monitoring before and every 2km of operation 2 Then, calibration is performed to correct sensor errors and inversion model parameters, ensuring monitoring accuracy.

[0029] The intelligent monitoring method for roadbed compaction based on UAV swarms includes the following steps: Step 1: Monitoring Pretreatment Data import: The ground control center imports the roadbed construction design drawings, including the route alignment, width, and fill material type, sets monitoring thresholds, and sets a compaction degree qualification standard of ≥93%; Zoning planning: Automatically divides monitoring zones, 0.1km 2 / area, mark key road sections (such as bridge and culvert connection sections, road sections with embankment height > 5m) as priority monitoring areas; Equipment configuration: Deploy 1 master drone + 5 slave drones, and install and debug multi-source sensing modules; On-site calibration: Place the standard compacted block on the roadbed surface, start the calibration program, correct the accuracy of GPR and infrared sensors, and enter parameters such as soil type silty soil and design moisture content of 12%.

[0030] Step 2: Collaborative Monitoring by Unmanned Aerial Vehicle (UAV) Swarm Path distribution: The master drone generates the partitioned operation path with a grid spacing of 0.5m and distributes it to each slave drone; Synchronous operation: The UAV takes off along the path, maintains a flight altitude of 5-15m and a flight speed of 5m / s, and simultaneously collects GPR, GNSS, infrared, and IMU data. The main UAV monitors the operation progress in real time and coordinates with adjacent UAVs to avoid collisions. Data upload: Collected data is transmitted to the ground control center in real time via 5G network and cached locally.

[0031] Step 3: Multi-source data fusion Spatiotemporal alignment: Based on GNSS timestamps and IMU attitude data, the acquisition time and spatial location of different sensors are calibrated to ensure that the data correspond one-to-one; Noise Removal: Wavelet transform is used to remove electromagnetic interference from GPR data, and Kalman filtering is used to correct temperature noise in infrared data; Data registration: Based on GNSS three-dimensional coordinates, GPR reflected wave data, infrared temperature data, and surface flatness data are correlated to generate a standardized dataset with a resolution of 0.1m×0.1m.

[0032] Step 4: Compaction Degree AI Inversion Feature extraction: Four core features were extracted from the standardized dataset: GPR reflection amplitude, phase difference, infrared temperature gradient, and surface smoothness. Model input: Input the core features and auxiliary parameters (such as soil type, measured moisture content of 11.8%, and number of compaction passes of 6) into the CNN-LSTM inversion model; Output results: The model outputs a three-dimensional compaction distribution map of the roadbed, with the compaction values ​​at each point marked.

[0033] Step 5: Results Feedback and Optimization Visualization: The ground control center uses a color heat map to show the compaction distribution, with weak areas marked in red with a compaction degree of <93%; Construction feedback: Coordinates of vulnerable areas and suggested parameter adjustments are pushed to the construction machinery terminals in real time; Secondary supplementary testing: After construction adjustments, the drone swarm initiates secondary supplementary testing on weak areas to verify whether the compaction degree meets the standard of ≥93%, thus forming a closed loop.

[0034] Implementation Case 1: Monitoring of Subgrade Construction for a Highway ① Configuration parameters Project parameters: Roadbed length 10km, width 28m, fill material is silty soil, design compaction degree ≥93%, construction machinery is 20t vibratory roller; System configuration: 1 master drone (DJI M300RTK) + 5 slave drones (DJI M210RTK), with the slave drones equipped with GPR (500MHz), GNSS (RTK precision), infrared thermal imager, and IMU; Monitoring parameters: flight altitude 8m, flight speed 5m / s, monitoring zone 0.1km 2 / zone, adjacent drones overlap rate 15%.

[0035] ② Implementation Results Monitoring efficiency: Five drones operated in parallel, completing 8km of roadbed monitoring in a single day in 6 hours, a 96% reduction compared to traditional methods; Monitoring accuracy: 20 points were randomly selected for verification using the sand cone method. The compaction monitoring error was ≤2.8%, which meets the engineering requirements. Weak Area Identification: Three weak areas at bridge-culvert connection sections were successfully identified, with compaction degrees ranging from 88% to 92%. After adjustment suggestions were sent, secondary monitoring showed that the compaction degree was ≥93% in all three areas. Project benefits: Reduced rework costs by approximately 800,000 yuan, shortened the construction period by 15 days, and increased the roadbed quality qualification rate from 92% using traditional methods to 98%.

[0036] The following points should be noted in this article: 1. The accompanying drawings of the embodiments disclosed herein only relate to the structures involved in the embodiments disclosed herein; other structures can be referred to in a general design.

[0037] 2. Where there is no conflict, the embodiments of this disclosure and the features in the embodiments can be combined with each other to obtain new embodiments.

[0038] Although embodiments of the present invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions, and variations 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. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

Claims

1. A roadbed compaction intelligent monitoring system based on unmanned aerial vehicle (UAV) swarms, characterized in that, It includes a drone swarm, a multi-source sensing module, a collaborative control module, a data transmission module, a ground control center, and an on-site calibration module; The drone swarm consists of one master drone and 3 to 8 slave drones. The master drone is equipped with a collaborative control unit, a global positioning module, and a wireless communication module, while the slave drones are equipped with a multi-source sensing module and a local storage unit. The master drone is responsible for path planning, operation coordination, and data aggregation, while the slave drones synchronously collect multi-source data according to the partitioned path. The master and slave drones achieve low-latency communication through 5G + BeiDou. The multi-source sensing module synchronously collects multi-source roadbed data from the UAV along the path, including GNSS positioning data, GPR electromagnetic wave data, infrared thermal imaging data, and IMU attitude data; The collaborative control module, including a path planning algorithm unit, a work coordination unit, and an obstacle avoidance unit, automatically divides the roadbed design drawings into sections ranging from 0.05 to 0.2 km. 2 The monitoring zones are sorted according to "critical road sections first, then ordinary road sections". The drones operate in parallel paths, and the obstacle avoidance unit avoids obstacles in real time through lidar. The data transmission module includes a 5G communication module, a BeiDou short message module, and a local cache unit. It prioritizes real-time data transmission via the 5G network, switches to BeiDou short message in areas without 5G signal, and stores data in the local cache unit to prevent data loss. The ground control center includes a data processing server, an AI inversion server, a visualization terminal, and a construction docking interface. It is used to receive multi-source data and complete the fusion processing, run a deep learning inversion model to output a compaction distribution map, visualize the results and push them to the construction terminal, and support data traceability and historical comparison. The on-site calibration module, including standard compacted blocks, a portable moisture meter, and a soil type analyzer, is used before monitoring and every 2km of operation. 2 Then, calibration is performed to correct sensor errors and inversion model parameters, ensuring monitoring accuracy.

2. The intelligent monitoring system for roadbed compaction based on unmanned aerial vehicle (UAV) swarms according to claim 1, characterized in that, The swarm of drones consists of a master drone equipped with a collaborative control unit and a global positioning module, and slave drones equipped with a multi-source sensing module. The master and slave drones achieve collaborative operation through wireless communication. The positioning accuracy is ≤5cm, the operating altitude is 5~15m above the roadbed surface, and the flight speed is 3~8m / s.

3. The intelligent monitoring system for roadbed compaction based on unmanned aerial vehicle (UAV) swarms according to claim 1, characterized in that, The multi-source sensing module includes a ground-penetrating radar, a GNSS positioning unit, an infrared thermal imager, and an IMU inertial measurement unit. Ground Penetrating Radar (GPR): Center frequency 500~1500MHz, detection depth 0.3~1.5m, capturing changes in electromagnetic wave reflection caused by differences in soil density in the shallow subgrade layer (0~1.2m); GNSS positioning unit: Supports RTK / PPP-RTK, positioning accuracy ±1cm, records the three-dimensional coordinates of each set of monitoring data to ensure data spatial registration accuracy; Infrared thermal imager: Temperature measurement range -20~80℃, accuracy ±0.5℃, captures the temperature gradient generated by soil friction during compaction. The higher the degree of compaction, the more uniform the temperature gradient. IMU (Inertial Measurement Unit): Sampling frequency 100~200Hz, measures UAV attitude, and calibrates sensor data for errors caused by flight attitude fluctuations.

4. A method for intelligent monitoring of roadbed compaction based on unmanned aerial vehicle (UAV) swarms, characterized in that, The method includes the following steps: Step 1: Monitoring Pretreatment The ground control center imports the roadbed construction design drawings, divides the monitoring zones, configures the drone swarm, and completes sensor calibration through the on-site calibration module; Step 2: Collaborative Monitoring by Unmanned Aerial Vehicle (UAV) Swarm The main UAV plans the operation path for each zone, and the secondary UAVs synchronously collect multi-source data of the roadbed according to the path. The main UAV coordinates the operation progress in real time and avoids monitoring blind spots. Step 3: Multi-source data fusion The data transmission module transmits data to the ground control center in real time, and after spatiotemporal alignment, noise removal, and data registration, a standardized dataset is obtained. Step 4: Compaction Degree AI Inversion The standardized dataset and auxiliary parameters such as soil type, moisture content, and number of compaction passes are input into the deep learning inversion model to output a three-dimensional compaction distribution map of the subgrade. Step 5: Results Feedback and Optimization The ground control center visualizes the compaction results, identifies weak areas, and pushes them to the construction terminal in real time to guide the adjustment of compaction parameters. If the monitoring error exceeds the threshold, a second round of monitoring by a drone swarm will be initiated to form a closed-loop optimization.

5. The intelligent monitoring method for roadbed compaction based on unmanned aerial vehicle (UAV) swarms according to claim 4, characterized in that, The collaborative monitoring strategy in step two is as follows: the master UAV plans its path according to "grid partitioning + priority sorting", the slave UAVs adopt "parallel operation + edge supplementary measurement" mode, the monitoring overlap rate of adjacent UAVs is ≥10% to ensure no monitoring blind spots, and the daily monitoring area of ​​a single slave UAV is ≥5km². 2 .

6. The intelligent monitoring method for roadbed compaction based on unmanned aerial vehicle (UAV) swarms according to claim 4, characterized in that, The data fusion in step three includes: Spatiotemporal alignment, based on GNSS timestamps and IMU attitude data calibration; Noise removal is achieved using a wavelet transform + Kalman filter algorithm. Data registration associates GPR data with infrared data using GNSS coordinates; The standardized dataset format is ASCII, and the data resolution is ≤0.1m×0.1m.

7. The intelligent monitoring method for roadbed compaction based on unmanned aerial vehicle (UAV) swarms according to claim 4, characterized in that, The deep learning inversion model is a CNN-LSTM hybrid model. The input features include GPR reflected wave amplitude, phase difference, infrared temperature gradient, and roadbed surface smoothness. Combined with auxiliary parameters such as soil type, moisture content, and number of compaction passes, the model training set contains ≥1000 sets of measured roadbed compaction data, and the prediction error is ≤3%.

8. The intelligent monitoring method for roadbed compaction based on unmanned aerial vehicle (UAV) swarms according to claim 4, characterized in that, The on-site calibration module includes a standard compacted block, a moisture content meter, and a soil type analyzer, with a calibration frequency of every 2 km monitored. 2 Alternatively, the operation can be performed when the ambient temperature changes by ≥5℃ to ensure sensor accuracy and the reliability of the inversion model.

9. The intelligent monitoring method for roadbed compaction based on unmanned aerial vehicle (UAV) swarms according to claim 4, characterized in that, The ground control center supports three-dimensional visualization of compaction degree, automatic marking of weak areas, and historical data comparison and analysis. It has a data storage capacity of ≥10TB and supports connection with the roadbed construction management platform to achieve full-process data traceability from monitoring to construction to acceptance.

Images