Intelligent dynamic compaction system and method based on precise positioning of Beidou

Abstract

The application relates to the field of satellite navigation and positioning technology, in particular to an intelligent dynamic compaction system and method based on Beidou precise positioning, which comprises a Beidou high-precision positioning unit, a multi-source parameter synchronous sensing unit, an intelligent interactive display and control terminal unit, a digital twin cloud platform unit and an intelligent decision and control unit. Through double-redundancy sensing and PTP hardware synchronization, the application realizes high-reliability and millisecond-level fusion collection of construction parameters, provides high-quality data support for accurate decision-making, realizes efficient monitoring and abnormal early warning of multiple roles by virtue of environment self-adaptive interaction and AR visualization technology, improves the efficiency of on-site operation and supervision, realizes self-optimization and precise control of ramming parameters based on digital twin simulation and deep reinforcement learning, and significantly improves the construction quality and efficiency and reduces the dependence on manpower.

Description

Technical Field

[0001] This invention relates to the field of satellite navigation and positioning technology, specifically to an intelligent dynamic compaction system and method based on BeiDou precise positioning. Background Technology

[0002] BeiDou precise positioning refers to a positioning method based on the Chinese BeiDou Navigation Satellite System (BDS), which uses technologies such as multi-frequency signal reception, multi-source data fusion, and anti-interference processing to achieve high-precision, high-reliability, and real-time acquisition of information such as the spatial position, velocity, and attitude of a target. Dynamic compaction, as a key method of foundation treatment, directly affects the safety and stability of subsequent engineering structures. However, existing technologies for dynamic compaction suffer from problems such as low reliability of parameter acquisition, insufficient synchronization accuracy, low efficiency of manual monitoring and decision-making, reliance on experience, difficulty in balancing construction quality and efficiency, and poor process traceability.

[0003] Based on this, the present invention provides an intelligent dynamic compaction system and method based on BeiDou precise positioning to solve the above-mentioned technical problems. Summary of the Invention

[0004] The purpose of this invention is to provide an intelligent dynamic compaction system and method based on BeiDou precise positioning. This invention features dual-channel redundant sensing and PTP hardware synchronization, achieving highly reliable, millisecond-level fusion acquisition of construction parameters, providing high-quality data support for precise decision-making. Furthermore, by leveraging environmental adaptive interaction and AR visualization technology, it enables efficient multi-role monitoring and anomaly early warning, improving on-site operation and supervision efficiency. Based on digital twin simulation and deep reinforcement learning, it achieves autonomous optimization and precise control of compaction parameters, significantly improving construction quality and efficiency while reducing reliance on manual labor.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] This invention provides an intelligent dynamic compaction system based on BeiDou precise positioning, comprising a BeiDou high-precision positioning unit, a multi-source parameter synchronous sensing unit, an intelligent interactive display and control terminal unit, a digital twin cloud platform unit, and an intelligent decision and control unit, wherein:

[0007] The Beidou high-precision positioning unit is used to perform centimeter-level real-time positioning of the dynamic compaction machinery in motion through a smart receiver combining a Beidou / GNSS multi-frequency antenna and an IMU, providing a high-precision spatial reference for hammer trajectory tracking and compaction point layout.

[0008] The multi-source parameter synchronous sensing unit is used to reliably acquire and synchronously fuse multi-source construction parameters such as the number of blows, drop distance, settlement amount and hydraulic energy in dynamic compaction construction through dual-channel redundant sensors, PTP hardware-level time synchronization and MEMS impact-resistant design, and embeds an abnormal data filtering algorithm.

[0009] The intelligent interactive display and control terminal unit is used to intelligently monitor and interactively make decisions on multi-dimensional data of dynamic compaction construction through environmental adaptive touch interaction, AR augmented reality visualization and role-adaptive interface technology.

[0010] The digital twin cloud platform unit: constructs a digital twin construction panorama model based on an edge-cloud collaborative architecture, centrally stores compaction trajectory and quality data, supports multi-role collaborative supervision and construction process traceability, and generates tamper-proof electronic construction logs through blockchain technology;

[0011] The intelligent decision and control unit is based on a hierarchical decision architecture using digital twin local process simulation and multimodal deep reinforcement learning. It autonomously optimizes and precisely controls energy distribution, path planning, and construction termination during the dynamic compaction process.

[0012] The BeiDou high-precision positioning unit includes a multi-frequency antenna module, an IMU inertial measurement module, an anti-interference processing module, and a positioning calculation module, wherein:

[0013] The multi-frequency antenna module is used to receive BeiDou / GNSS satellite signals and perform continuous carrier phase observation and initial positioning in a high-dynamic environment.

[0014] The IMU (Inertial Measurement Unit) module compensates for positioning gaps by providing continuous attitude and acceleration data when satellite signals are blocked or jittered.

[0015] The anti-interference processing module is used to eliminate positioning deviations in metallic environments using a multipath suppression algorithm.

[0016] The positioning and calculation module is used to fuse satellite and IMU data and output centimeter-level real-time spatial coordinates through a Kalman filter algorithm.

[0017] The multi-source parameter synchronous sensing unit includes a redundant sensor acquisition module, a time synchronization fusion module, and a data purification and processing module, wherein:

[0018] The redundant sensing acquisition module is used to integrate a dual-channel laser rangefinder and a hydraulic sensor to reliably acquire construction parameters such as the number of blows, drop distance, settlement amount, and hydraulic energy.

[0019] The time synchronization and fusion module: based on PTP hardware clock synchronization technology, performs millisecond-level spatiotemporal alignment and data fusion of construction parameters;

[0020] The data purification and processing module is used to output standardized and highly reliable construction parameters through outlier filtering algorithms and MEMS shock resistance compensation.

[0021] The time synchronization and fusion module is based on PTP hardware clock synchronization technology, which performs millisecond-level spatiotemporal alignment and data fusion of multi-source parameters. The specific operation is as follows:

[0022] A1: Deploy PTP slave clocks that support hardware timestamps at each sensor node of the dynamic compaction machine;

[0023] A2: Use the PPS pulse signal output by the Beidou high-precision positioning unit as the master clock source to construct a unified time reference for the entire system;

[0024] A3: Each sensor is timed with a hardware timestamp accurate to the microsecond level via an FPGA or dedicated PHY chip at the moment of data acquisition.

[0025] A4: In edge computing nodes, data from laser rangefinders, hydraulic sensors, and IMUs are interpolated, aligned, and fused according to timestamps;

[0026] A5: Outputs a synchronized data stream with a unified time coordinate for subsequent state perception and decision control.

[0027] The intelligent interactive display and control terminal unit includes an environment-adaptive interactive module, an AR visualization module, and a role-playing interface module, wherein:

[0028] The environmental adaptive interaction module is designed to support stable operation and interaction under harsh conditions through an anti-glare touchscreen and temperature and humidity adaptive design.

[0029] The AR visualization module is used to overlay positioning data and construction parameters onto a real-world scene, intuitively displaying information such as the distribution of compaction points and settlement curves.

[0030] The role-based interface module is used to customize the display content according to the different roles of operators and supervisors, and supports parameter correction, over-limit alarms and cloud command interaction.

[0031] The AR visualization module overlays positioning data and construction parameters onto a real-world scene, intuitively displaying information such as the distribution of compaction points and settlement curves. The specific operation is as follows:

[0032] B1: Real-time images of the construction site are captured by the terminal camera, and the location coordinates of the rammer output by the Beidou high-precision positioning unit and the construction parameters of the multi-source parameter synchronous sensing unit are called.

[0033] B2: Perform distortion correction and coordinate system calibration on the real-world image to establish a mapping relationship between pixel coordinates and three-dimensional spatial coordinates of the construction site;

[0034] B3: Based on the AR rendering engine, the distribution of tamping points is superimposed on the corresponding spatial location of the real scene in the form of three-dimensional markers and the settlement curve is superimposed on the form of dynamic line graphs.

[0035] B4: Automatically adjusts the display priority of virtual information for different construction stages, and supports touch zoom and pan operations to adapt to different viewing angles;

[0036] B5: When the deviation of the compaction point exceeds the threshold or the settlement curve is abnormal, information prompts will be enhanced through AR pop-ups and color warnings.

[0037] The digital twin cloud platform unit includes a twin modeling module, a data middleware module, and a collaborative evidence storage module, wherein:

[0038] The twin modeling module: Based on edge-cloud collaborative computing, it constructs a digital twin panoramic model that is mapped in real time to the physical construction scene;

[0039] The data platform module is used to centrally store full information on compaction trajectory and quality data, and supports multi-dimensional query and construction process backtracking.

[0040] The collaborative evidence storage module is used to store construction logs on the blockchain using blockchain technology, and supports multi-role permission management and collaborative supervision.

[0041] The intelligent decision-making and control unit includes a digital twin local simulation module, a reinforcement learning decision-making module, and a precise execution module, wherein:

[0042] The digital twin local simulation module is used to dynamically simulate the local process of dynamic compaction and predict the construction effect under different parameter combinations.

[0043] The reinforcement learning decision module optimizes impact energy allocation, path planning, and hammer stop threshold based on a multimodal deep reinforcement learning algorithm.

[0044] The precision execution module is used to automatically adjust the rammer parameters by linking the decision results with the actuator through the electro-hydraulic proportional valve group.

[0045] The reinforcement learning decision-making module is based on a multimodal deep reinforcement learning algorithm to optimize impact energy allocation, path planning, and hammer stopping threshold. The specific operations are as follows:

[0046] C1: Multimodal State Space Construction: Collecting BeiDou positioning data, multi-source sensing parameters, and digital twin simulation results, a high-dimensional state vector is constructed through feature fusion.

[0047]

[0048] in, Let P(x,y,z) be the foundation bearing capacity predicted by the digital twin at time t, and E be the BeiDou position coordinates. t For the impact energy at time t, St Let F be the settlement at time t. t Let be the hydraulic pressure at time t;

[0049] C2: Bi-objective reward function design: With "maximizing construction efficiency" and "foundation compliance rate" as optimization objectives, the reward function formula is as follows:

[0050]

[0051] Among them, R t Let T be the reward value at time t. t The time required for single-point compaction construction, This is an indicator function; it takes the value 1 if the target is met, and 0 otherwise. t To measure the bearing capacity of the foundation, y target For the target foundation bearing capacity, E opt The energy is the theoretically optimal energy, and α, β, γ are weighting coefficients.

[0052] C3: Deep Reinforcement Learning Training: Employing the DDPG algorithm and iteratively optimizing the strategy through an actor-critic network.

[0053] ①The Actor network outputs a continuous action space:

[0054] a t =π θ (s t )=[λ t ,ΔP t ,S stop,t ]

[0055] Where, λ t Let ΔP be the energy distribution coefficient at time t. t S is the path correction amount at time t. stop,t Let t be the threshold for stopping the hammer.

[0056] ② The Critic network evaluates the Q-value based on the current state and action, and updates the network parameters θ using gradient descent:

[0057]

[0058] Where η is the learning rate, γ discount Discount factor;

[0059] C4: Online Decision Making and Dynamic Correction: Receives construction feedback in real time, explores and utilizes the Ornstein-Uhlenbeck noise process balance, outputs final decision parameters, and fine-tunes network weights online based on actual feedback reward values.

[0060] This invention also proposes an intelligent dynamic compaction method based on BeiDou precise positioning, comprising the following steps:

[0061] S1: Using BeiDou / GNSS and IMU fusion positioning technology, centimeter-level spatial coordinates and attitude data of dynamic compaction machinery are acquired in real time to establish a spatiotemporal reference.

[0062] S2: Employs dual-channel redundant sensing and PTP hardware clock synchronization to perform millisecond-level fusion acquisition of multi-source construction parameters such as number of blows, drop distance, settlement amount, and hydraulic energy.

[0063] S3: Based on augmented reality technology, construction parameters are overlaid onto the real-world scene, supporting multi-role interactive monitoring and anomaly warning;

[0064] S4: Construct a digital twin model that integrates edge and cloud computing, mapping all construction elements in real time and generating blockchain-based evidence logs;

[0065] S5: Utilize digital twin simulation and deep reinforcement learning to dynamically optimize impact energy distribution and path planning strategies;

[0066] S6: Transform decision commands into electro-hydraulic control signals and continuously revise the decision model based on construction feedback data.

[0067] Compared with the prior art, the beneficial effects of the present invention are:

[0068] This invention achieves highly reliable, millisecond-level fusion acquisition of construction parameters through dual-channel redundant sensing and PTP hardware synchronization, providing high-quality data support for accurate decision-making. With environmental adaptive interaction and AR visualization technology, it enables efficient monitoring and anomaly warning for multiple roles, improving on-site operation and supervision efficiency. Based on digital twin simulation and deep reinforcement learning, it achieves autonomous optimization and precise control of compaction parameters, significantly improving construction quality and efficiency while reducing reliance on manual labor. Attached Figure Description

[0069] Figure 1 This is a system diagram of an intelligent dynamic compaction system based on BeiDou precise positioning according to the present invention.

[0070] Figure 2 This is a flowchart of an intelligent dynamic compaction method based on BeiDou precise positioning according to the present invention.

[0071] Explanation of icon numbers:

[0072] 100. Beidou High-Precision Positioning Unit; 101. Multi-Frequency Antenna Module; 102. IMU Inertial Measurement Module; 103. Anti-Interference Processing Module; 104. Positioning Solution Module; 200. Multi-Source Parameter Synchronous Sensing Unit; 201. Redundant Sensing Acquisition Module; 202. Time Synchronization Fusion Module; 203. Data Cleaning and Processing Module; 300. Intelligent Interactive Display and Control Terminal Unit; 301. Environment Adaptive Interactive Module; 302. AR Visualization Module; 303. Role-Playing Interface Module; 400. Digital Twin Cloud Platform Unit; 401. Twin Modeling Module; 402. Data Platform Module; 403. Collaborative Evidence Storage Module; 500. Intelligent Decision and Control Unit; 501. Digital Twin Local Simulation Module; 502. Reinforcement Learning Decision Module; 503. Precision Execution Module. Detailed Implementation

[0073] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. 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 of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0074] Example 1:

[0075] like Figure 1As shown, this embodiment provides an intelligent dynamic compaction system based on BeiDou precise positioning, including a BeiDou high-precision positioning unit 100, a multi-source parameter synchronous sensing unit 200, an intelligent interactive display and control terminal unit 300, a digital twin cloud platform unit 400, and an intelligent decision and control unit 500. Specifically: the BeiDou high-precision positioning unit 100 is used to perform centimeter-level real-time positioning of the dynamic compaction machinery in motion using an intelligent receiver combining a BeiDou / GNSS multi-frequency antenna and an IMU, providing a high-precision spatial reference for hammer trajectory tracking and compaction point layout; the multi-source parameter synchronous sensing unit 200 is used to perform multi-source construction parameters such as the number of blows, drop distance, compaction settlement, and hydraulic energy during dynamic compaction construction through dual-channel redundant sensors, PTP hardware-level time synchronization, and MEMS impact-resistant design. High-reliability data acquisition and millisecond-level synchronization are integrated, and anomaly data filtering algorithms are embedded. The intelligent interactive display and control terminal unit 300 is used for intelligent monitoring and interactive decision-making of multi-dimensional data from dynamic compaction construction through environment-adaptive touch interaction, AR augmented reality visualization, and role-adaptive interface technology. The digital twin cloud platform unit 400 constructs a digital twin construction panoramic model based on an edge-cloud collaborative architecture, centrally stores compaction trajectory and quality data, supports multi-role collaborative supervision and construction process backtracking, and generates tamper-proof electronic construction logs through blockchain technology. The intelligent decision and control unit 500, based on a hierarchical decision-making architecture of digital twin local process simulation and multimodal deep reinforcement learning, autonomously optimizes and precisely controls energy distribution, path planning, and construction termination during the dynamic compaction process.

[0076] It should be noted that the Beidou high-precision positioning unit 100 provides a centimeter-level spatial reference, the multi-source parameter synchronous sensing unit 200 realizes millisecond-level synchronous acquisition of construction parameters, the intelligent interactive display and control terminal unit 300 performs AR visualization monitoring, the digital twin cloud platform unit 400 constructs a full-element construction model and stores it on the blockchain, and finally the intelligent decision and control unit 500 completes autonomous decision-making and control based on digital twin simulation and deep reinforcement learning.

[0077] In this embodiment, it should also be noted that the BeiDou high-precision positioning unit 100 includes a multi-frequency antenna module 101, an IMU inertial measurement module 102, an anti-interference processing module 103, and a positioning calculation module 104, wherein: the multi-frequency antenna module 101 is used to receive BeiDou / GNSS satellite signals and perform continuous carrier phase observation and initial positioning in a high-dynamic environment; the IMU inertial measurement module 102 provides continuous attitude and acceleration data to compensate for positioning gaps when satellite signals are blocked or jittered; the anti-interference processing module 103 is used to eliminate positioning deviations in metallic environments using a multipath suppression algorithm; and the positioning calculation module 104 is used to fuse satellite and IMU data and output centimeter-level real-time spatial coordinates using a Kalman filter algorithm.

[0078] It should be noted that the multi-frequency antenna module 101 receives satellite signals, and the IMU inertial measurement module 102 provides continuous attitude data when the signal is interrupted. The anti-interference processing module 103 eliminates multipath errors, and finally the positioning calculation module 104 fuses the multi-source data and uses the Kalman filter algorithm to output centimeter-level real-time coordinates.

[0079] Furthermore, it should be noted that the multi-frequency antenna module 101 uses a combination of BeiDou B1I / B2I and GPS L1 / L2 frequencies for reception, supporting carrier phase differential. When the dynamic compaction machine is moving at high dynamic speeds (≤30km / h), the initial positioning error is ≤0.5m achieved through a phase smoothing pseudorange algorithm, providing a foundation for subsequent centimeter-level calculations. The multipath suppression algorithm employs a combined strategy of "wavelet threshold denoising + Kalman filter residual analysis": For multipath errors caused by metal reflections (such as the steel structure of the compaction frame), wavelet transform is used to decompose high-frequency noise in the signal (threshold is 3σ, where σ is the historical positioning standard deviation), and residual checks are used to eliminate outliers with deviations >0.3m, ultimately reducing the multipath error to ≤5cm.

[0080] In this embodiment, it should also be noted that the multi-source parameter synchronous sensing unit 200 includes a redundant sensor acquisition module 201, a time synchronization fusion module 202, and a data purification and processing module 203. The redundant sensor acquisition module 201 integrates a dual-path laser rangefinder and a hydraulic sensor to reliably acquire construction parameters such as the number of impacts, drop distance, settlement, and hydraulic energy. The time synchronization fusion module 202, based on PTP hardware clock synchronization technology, performs millisecond-level spatiotemporal alignment and data fusion of the construction parameters. The specific operation is as follows: A1: [The text abruptly ends here, likely due to an incomplete sentence or a formatting error.] The node deployment supports PTP slave clocks with hardware timestamps; A2: The PPS pulse signal output by the Beidou high-precision positioning unit 100 is used as the master clock source to construct a unified time reference for the entire system; A3: Each sensor adds a hardware timestamp accurate to the microsecond level through FPGA or dedicated PHY chip at the time of data acquisition; A4: In the edge computing nodes, data from the laser rangefinder, hydraulic sensor, and IMU are interpolated, aligned, and fused according to the timestamps; A5: A synchronous data stream with a unified time coordinate is output for subsequent state perception and decision control. Data purification and processing module 203: Used to output standardized and highly reliable construction parameters through outlier filtering algorithms and MEMS shock resistance compensation.

[0081] It should be noted that the redundant sensor acquisition module 201 achieves highly reliable acquisition of construction parameters. The time synchronization fusion module 202 completes millisecond-level spatiotemporal alignment of multi-source data based on PTP hardware clock synchronization technology. Then, the data purification and processing module 203 performs outlier filtering and shock compensation, and finally outputs standardized and highly reliable synchronized construction parameters.

[0082] Furthermore, it should be noted that the dual-channel laser rangefinder uses 650nm visible red light, with a range of 0.5-50m, an accuracy of ±1mm@10m, and a sampling rate of 100Hz. It is installed symmetrically on both sides of the tamping frame (2m apart), and a redundancy switching mechanism is used to ensure the reliability of the drop distance acquisition. The hydraulic sensor is a strain gauge type, with a range of 0-30MPa, an accuracy of 0.1% FS, and a sampling rate of 500Hz. It is installed at the oil inlet of the main cylinder to collect the hydraulic energy during tamping in real time (energy = pressure × cylinder displacement × efficiency coefficient 0.9). The interpolation alignment in A4 uses a "linear interpolation + sliding window" algorithm: for the asynchronous data from the laser rangefinder and the hydraulic sensor, an isochronous series is generated by linear interpolation with 100Hz as the reference; when the time series is fused, a sliding window of 5 data points is used to calculate the mean, and instantaneous jumps caused by vibration are eliminated (if the deviation of a single sample value from the window mean is >5%, it is judged as abnormal and replaced by the previous mean).

[0083] In this embodiment, it should also be noted that the intelligent interactive display and control terminal unit 300 includes an environment-adaptive interaction module 301, an AR visualization module 302, and a role-playing interface module 303. Specifically: the environment-adaptive interaction module 301 is used to support stable operation and interaction under harsh working conditions through an anti-glare touchscreen and temperature / humidity adaptive design; the AR visualization module 302 is used to overlay positioning data and construction parameters onto a real-world scene, intuitively displaying information such as the distribution of compaction points and settlement curves. The specific operation is as follows: B1: Real-time scene images of the construction site are collected through the terminal camera, and the compaction machine position output by the Beidou high-precision positioning unit 100 is simultaneously invoked. The system includes: B1: Setting coordinates and multi-source parameter synchronous sensing unit 200 to sense construction parameters; B2: Performing distortion correction and coordinate system calibration on the real-world image to establish a mapping relationship between pixel coordinates and three-dimensional spatial coordinates of the construction site; B3: Based on the AR rendering engine, overlaying the compaction point distribution as three-dimensional markers and the settlement curve as a dynamic line graph onto the corresponding spatial positions of the real-world image on a proportional basis; B4: Automatically adjusting the display priority of virtual information for different construction stages and supporting touch zoom and pan operations to adapt to different viewing angles; B5: When the compaction point position deviation exceeds the threshold or the settlement curve shows abnormalities, strengthening information prompts through AR pop-ups and color warnings. Role-based interface module 303: Used to customize display content according to different roles of operators and supervisors, supporting parameter correction, over-limit alarms, and cloud command interaction.

[0084] It should be noted that the environment adaptive interaction module 301 ensures stable operation under harsh working conditions, the AR visualization module 302 realizes the intelligent overlay of construction data and real-world images, and the role-based interface module 303 provides personalized interactive functions.

[0085] Furthermore, it should be noted that distortion correction in B2 adopts Zhang's calibration method, acquiring 20 sets of images using a 9×9 checkerboard grid (grid spacing 50mm), calculating the intrinsic parameter matrix (focal length fx=1200±50, fy=1200±50) and distortion coefficients (k1=-0.3±0.05, k2=0.1±0.03); coordinate system calibration uses the rotation center of the ramming machine positioned by Beidou as the origin, establishing a "Northeast Sky" coordinate system, ensuring that the mapping error between pixel coordinates and three-dimensional coordinates is ≤3 pixels (corresponding to actual distance ≤10cm). In B5, threshold settings: the ramming point position deviation threshold is 30cm (based on the ramming point coordinates in the design drawings), and the settlement curve anomaly is judged as "the difference in settlement amount after 3 consecutive ramming blows >5cm or the settlement amount after a single blow >20cm". At this time, the AR pop-up window displays "deviation value + adjustment suggestion", and the color warning uses three dynamic labels: red (overlimit), yellow (warning, deviation 20-30cm), and green (normal). Adjustment suggestions: When the settlement difference is >5cm for 3 consecutive times, increase the energy by 10%-15%; when the settlement amount is >20cm for a single time, shift the position 30-50cm and recompact.

[0086] In this embodiment, it should also be noted that the digital twin cloud platform unit 400 includes a twin modeling module 401, a data middle platform module 402, and a collaborative evidence storage module 403, wherein: the twin modeling module 401: constructs a digital twin panoramic model that is mapped in real time to the physical construction scene based on edge-cloud collaborative computing; the data middle platform module 402: is used to centrally store the full information of compaction trajectory and quality data, and supports multi-dimensional query and construction process backtracking; the collaborative evidence storage module 403: is used to store construction logs on the blockchain for evidence storage through blockchain technology, and supports multi-role permission management and collaborative supervision.

[0087] It should be noted that the twin modeling module 401 constructs a real-time mapped digital twin model of the construction scene, the data middle platform module 402 realizes centralized storage and intelligent analysis of all construction data, and the collaborative evidence storage module 403 completes blockchain on-chain evidence storage and multi-role collaborative supervision.

[0088] Furthermore, it should be noted that a three-tiered storage architecture of "hot data - warm data - cold data" is adopted: hot data (real-time construction parameters for the past 24 hours) is stored in an in-memory database (such as Redis), supporting millisecond-level queries for AR visualization and reinforcement learning decision-making; warm data (construction trajectory and quality inspection data for the past 30 days) is stored in a relational database (such as MySQL), indexed by "equipment ID-date-compacting point number", supporting multi-dimensional combined queries (such as "querying all compaction point settlements of machine number H01 on May 10, 2024"); cold data (historical data and completion archives for more than 30 days) is stored in a distributed file system (such as HDFS), with a compression ratio of ≥5:1, meeting long-term archiving requirements (retention period ≥5 years), and supporting construction process retrospective and post-audit.

[0089] In this embodiment, it should also be noted that the intelligent decision and control unit 500 includes a digital twin local simulation module 501, a reinforcement learning decision module 502, and a precision execution module 503, wherein: the digital twin local simulation module 501 is used to dynamically simulate the local process of dynamic compaction and predict the construction effect under different parameter combinations; the reinforcement learning decision module 502 is based on a multimodal deep reinforcement learning algorithm to optimize the compaction energy distribution, path planning, and hammer stopping threshold; the specific operation is as follows: C1: Multimodal state space construction: collect Beidou positioning data, multi-source sensing parameters, and digital twin simulation results, and construct a high-dimensional state vector through feature fusion:

[0090]

[0091] in, Let P(x,y,z) be the foundation bearing capacity predicted by the digital twin at time t, and E be the BeiDou position coordinates. t For the impact energy at time t, S t Let F be the settlement at time t. t Let C1 be the hydraulic pressure at time t; C2: Bi-objective reward function design: with "maximizing construction efficiency" and "foundation compliance rate" as optimization objectives, the reward function formula is:

[0092]

[0093] Among them, R t Let T be the reward value at time t. t The time required for single-point compaction construction, This is an indicator function; it takes the value 1 if the target is met, and 0 otherwise. t To measure the bearing capacity of the foundation, y target For the target foundation bearing capacity, E optTheoretically optimal energy, α, β, γ are weight coefficients; C3: Deep reinforcement learning training: using the DDPG algorithm, with iterative optimization strategy through an actor-critic network: ① The actor network outputs a continuous action space:

[0094] a t =π θ (s t )=[λ t ,ΔP t ,S stop,t ]

[0095] Where, λ t Let ΔP be the energy distribution coefficient at time t. t S is the path correction amount at time t. stop,t ① The stopping threshold at time t; ② The Critic network evaluates the Q-value based on the current state and action, and updates the network parameters θ through gradient descent:

[0096]

[0097] Where η is the learning rate, γ discount C4: Online decision-making and dynamic correction: Receives construction feedback in real time, explores and utilizes the Ornstein-Uhlenbeck noise process balance, outputs final decision parameters, and fine-tunes network weights online based on actual feedback reward values. Precision execution module 503: Used to automatically adjust rammer parameters by linking the decision results with the actuator via an electro-hydraulic proportional valve group.

[0098] It should be noted that the digital twin local simulation module 501 predicts the construction effect, the reinforcement learning decision module 502 optimizes the construction parameter decision based on the multimodal deep reinforcement learning algorithm, and then the precision execution module 503 converts the decision command into an electro-hydraulic control signal.

[0099] Furthermore, it should be noted that the electro-hydraulic proportional valve assembly uses PWM control with a control frequency of 100Hz, based on the energy distribution coefficient λ output by reinforcement learning. t Adjusting the cylinder flow rate (λ) t =1 corresponds to a maximum flow rate of 100 L / min), path correction ΔP t Converted to walking motor speed (ΔP) t =1m corresponds to a rotational speed of 1000r / min), hammer stop threshold S stop,t When triggered, a power failure signal is output to the lifting system, with a response delay of ≤0.5s.

[0100] Example 2:

[0101] like Figure 2As shown in this embodiment, an intelligent dynamic compaction method based on BeiDou precise positioning specifically includes the following steps:

[0102] S1. Establishment of centimeter-level positioning benchmarks

[0103] ① Receive satellite signals through the BeiDou / GNSS multi-frequency antenna (B1I / B2I+L1 / L2) and combine them with IMU inertial measurement data;

[0104] ② Wavelet threshold denoising (3σ) and Kalman residual test (>0.3m rejection) are used to suppress multipath error;

[0105] ③ Output real-time positioning results with a plane ±1cm and an elevation ±2cm;

[0106] S2. Synchronous acquisition of multi-source parameters

[0107] ① Dual-channel laser rangefinder (650nm, ±1mm@10m) and hydraulic sensor (0.1%FS); synchronous data acquisition:

[0108] I. Drop distance and number of impacts: Data from laser rangefinder; redundancy switching is triggered when the difference between the two channels is >3mm.

[0109] II. Hydraulic Energy: Pressure × Cylinder Displacement × 0.9 Efficiency Coefficient;

[0110] ②PTP hardware synchronization (Xilinx Zynq-7030 FPGA timestamping, error ≤100μs);

[0111] S3: AR Visual Monitoring

[0112] ① Real-scene integration:

[0113] I. Calibrate the camera using a 9×9 checkerboard pattern (reprojection error ≤ 2 pixels);

[0114] II. Map the coordinates of the tamping point (northeast celestial coordinate system) onto the AR screen (error ≤ 10cm);

[0115] ② Abnormal warning:

[0116] I. Abnormal Settlement: A red alert is issued when the difference is >5cm for 3 consecutive times, and it is recommended to "increase the energy by 10-15%";

[0117] II. Positional deviation: Yellow alert when >30cm, "re-measure coordinates" recommended;

[0118] S4. Digital Twin Modeling and Evidence Preservation

[0119] ① Model update: Finite element mesh reconstruction is triggered after every 50 impacts;

[0120] ② Data layering:

[0121]

[0122]

[0123] ③ Blockchain notarization: A block (including the Merkle root hash) is generated every 10 ramming operations, using PBFT consensus;

[0124] S5: Intelligent Decision-Making and Control

[0125] ① DDPG algorithm execution:

[0126] I. State Vector:

[0127]

[0128] in, Let P(x,y,z) be the foundation bearing capacity predicted by the digital twin at time t, and E be the BeiDou position coordinates. t For the impact energy at time t, S t Let F be the settlement at time t. t Let be the hydraulic pressure at time t;

[0129] II. OU noise exploration: θ = 0.15, σ = 0.2, attenuation coefficient k = 0.01;

[0130] III. Network Update: Critic updates every 10 hits, with a learning rate η = 0.001;

[0131] ② Electro-hydraulic control:

[0132] I. Energy Distribution: PWM 100Hz regulates cylinder flow (linearly corresponding to 0-100L / min);

[0133] II. Emergency Stop Response: Stop hammer command delay ≤ 0.5s;

[0134] S6. Closed-loop feedback optimization:

[0135] ① Feed back the actual compaction volume and foundation bearing capacity data to the digital twin model;

[0136] ② Dynamically adjust the weights of the DDPG reward function (α, β, γ are optimized every 24 hours).

[0137] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0138] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. An intelligent dynamic compaction system based on BeiDou precise positioning, characterized in that, It includes a BeiDou high-precision positioning unit (100), a multi-source parameter synchronous sensing unit (200), an intelligent interactive display and control terminal unit (300), a digital twin cloud platform unit (400), and an intelligent decision and control unit (500), wherein: The Beidou high-precision positioning unit (100) is used to perform centimeter-level real-time positioning of the dynamic compaction machinery in motion through a smart receiver combining a Beidou / GNSS multi-frequency antenna and an IMU, providing a high-precision spatial reference for hammer trajectory tracking and compaction point layout. The multi-source parameter synchronous sensing unit (200) is used to reliably collect and synchronously fuse multi-source construction parameters such as the number of blows, drop distance, settlement amount and hydraulic energy in dynamic compaction construction through dual-channel redundant sensors, PTP hardware-level time synchronization and MEMS impact-resistant design, and embeds an abnormal data filtering algorithm. The intelligent interactive display and control terminal unit (300) is used to intelligently monitor and interactively make decisions on multi-dimensional data of dynamic compaction construction through environmental adaptive touch interaction, AR augmented reality visualization and role-adaptive interface technology. The digital twin cloud platform unit (400) constructs a digital twin construction panorama model based on an edge-cloud collaborative architecture, centrally stores compaction trajectory and quality data, supports multi-role collaborative supervision and construction process traceability, and generates tamper-proof electronic construction logs through blockchain technology. The intelligent decision and control unit (500) is based on a hierarchical decision architecture of digital twin local process simulation and multimodal deep reinforcement learning, which autonomously optimizes and precisely controls the energy distribution, path planning and construction termination of the dynamic compaction process. The intelligent decision and control unit (500) includes a digital twin local simulation module (501), a reinforcement learning decision module (502), and a precise execution module (503), wherein: The digital twin local simulation module (501) is used to dynamically simulate the local process of dynamic compaction and predict the construction effect under different parameter combinations. The reinforcement learning decision module (502) optimizes the impact energy allocation, path planning, and hammer stop threshold based on a multimodal deep reinforcement learning algorithm. The precision execution module (503) is used to automatically adjust the ramming machine parameters by linking the electro-hydraulic proportional valve group with the actuator. The reinforcement learning decision module (502) is based on a multimodal deep reinforcement learning algorithm to optimize the impact energy allocation, path planning, and hammer stopping threshold. The specific operations are as follows: C1: Multimodal State Space Construction: Collecting BeiDou positioning data, multi-source sensing parameters, and digital twin simulation results, a high-dimensional state vector is constructed through feature fusion. ； in, The foundation bearing capacity predicted by the digital twin at time t. The coordinates of the BeiDou position. For the energy of impact at time t, Let be the settlement at time t. Let be the hydraulic pressure at time t; C2: Bi-objective reward function design: With "maximizing construction efficiency" and "foundation compliance rate" as optimization objectives, the reward function formula is as follows: ； in, Let be the reward value at time t. The time required for single-point compaction construction, This is an indicator function; it takes the value 1 if the target is met, and 0 otherwise. To measure the bearing capacity of the foundation, For the target foundation bearing capacity, For the theoretically optimal energy, These are the weighting coefficients; C3: Deep Reinforcement Learning Training: Employing the DDPG algorithm and iteratively optimizing the strategy through an actor-critic network. ①The Actor network outputs a continuous action space: ； in, Let be the energy distribution coefficient at time t. Let be the path correction amount at time t. Let t be the threshold for stopping the hammer. ② The Critic network evaluates the Q-value based on the current state and action, and updates the network parameters through gradient descent. : ； in, For learning rate, Discount factor; C4: Online Decision Making and Dynamic Correction: Receives construction feedback in real time, explores and utilizes the Ornstein-Uhlenbeck noise process balance, outputs final decision parameters, and fine-tunes network weights online based on actual feedback reward values.

2. The intelligent dynamic compaction system based on BeiDou precise positioning according to claim 1, characterized in that, The BeiDou high-precision positioning unit (100) includes a multi-frequency antenna module (101), an IMU inertial measurement module (102), an anti-interference processing module (103), and a positioning calculation module (104), wherein: The multi-frequency antenna module (101) is used to receive BeiDou / GNSS satellite signals and perform continuous carrier phase observation and initial positioning in a high dynamic environment; The IMU inertial measurement module (102) compensates for positioning gaps by providing continuous attitude and acceleration data when satellite signals are blocked or jittered. The anti-interference processing module (103) is used to eliminate positioning deviations in a metallic environment by employing a multipath suppression algorithm; The positioning calculation module (104) is used to fuse satellite and IMU data and output centimeter-level real-time spatial coordinates through the Kalman filter algorithm.

3. The intelligent dynamic compaction system based on BeiDou precise positioning according to claim 1, characterized in that, The multi-source parameter synchronous sensing unit (200) includes a redundant sensing acquisition module (201), a time synchronization fusion module (202), and a data purification and processing module (203), wherein: The redundant sensor acquisition module (201) is used to integrate a dual-channel laser rangefinder and a hydraulic sensor to reliably acquire construction parameters such as the number of impacts, drop distance, settlement amount, and hydraulic energy. The time synchronization fusion module (202) is based on PTP hardware clock synchronization technology to perform millisecond-level spatiotemporal alignment and data fusion of construction parameters; The data purification and processing module (203) is used to output standardized and highly reliable construction parameters through outlier filtering algorithms and MEMS shock resistance compensation.

4. The intelligent dynamic compaction system based on BeiDou precise positioning according to claim 3, characterized in that, The time synchronization fusion module (202) is based on PTP hardware clock synchronization technology and performs millisecond-level spatiotemporal alignment and data fusion of multi-source parameters. The specific operation is as follows: A1: Deploy PTP slave clocks that support hardware timestamps at each sensor node of the dynamic compaction machine; A2: Using the PPS pulse signal output by the Beidou high-precision positioning unit (100) as the master clock source, a unified time reference for the entire system is constructed; A3: Each sensor is timed with a hardware timestamp accurate to the microsecond level via an FPGA or dedicated PHY chip at the moment of data acquisition. A4: In edge computing nodes, data from laser rangefinders, hydraulic sensors, and IMUs are interpolated, aligned, and fused according to timestamps; A5: Outputs a synchronized data stream with a unified time coordinate for subsequent state perception and decision control.

5. The intelligent dynamic compaction system based on BeiDou precise positioning according to claim 1, characterized in that, The intelligent interactive display and control terminal unit (300) includes an environment-adaptive interaction module (301), an AR visualization module (302), and a role-playing interface module (303), wherein: The environment adaptive interaction module (301) is used to support stable operation and interaction under harsh working conditions through the anti-glare touch screen and temperature and humidity adaptation design. The AR visualization module (302) is used to overlay positioning data and construction parameters onto the real-world scene to intuitively display information on the distribution of compaction points and settlement curves. The role-based interface module (303) is used to customize the display content according to the different roles of the operator and supervisor, and supports parameter correction, over-limit alarm and cloud command interaction.

6. The intelligent dynamic compaction system based on BeiDou precise positioning according to claim 5, characterized in that, The AR visualization module (302) overlays positioning data and construction parameters onto the real-world image, intuitively displaying information such as the distribution of compaction points and settlement curves. The specific operation is as follows: B1: Real-time scene images of the construction site are collected through the terminal camera, and the location coordinates of the rammer output by the Beidou high-precision positioning unit (100) and the construction parameters of the multi-source parameter synchronous sensing unit (200) are called simultaneously. B2: Perform distortion correction and coordinate system calibration on the real-world image to establish a mapping relationship between pixel coordinates and three-dimensional spatial coordinates of the construction site; B3: Based on the AR rendering engine, the distribution of tamping points is superimposed on the corresponding spatial location of the real scene in the form of three-dimensional markers and the settlement curve is superimposed on the form of dynamic line graphs. B4: Automatically adjusts the display priority of virtual information for different construction stages, and supports touch zoom and pan operations to adapt to different viewing angles; B5: When the deviation of the compaction point exceeds the threshold or the settlement curve is abnormal, information prompts will be enhanced through AR pop-ups and color warnings.

7. The intelligent dynamic compaction system based on BeiDou precise positioning according to claim 1, characterized in that, The digital twin cloud platform unit (400) includes a twin modeling module (401), a data middleware module (402), and a collaborative evidence storage module (403), wherein: The twin modeling module (401) constructs a digital twin panoramic model that is mapped in real time to the physical construction scene based on edge-cloud collaborative computing; The data platform module (402) is used to centrally store the full information of compaction trajectory and quality data, and supports multi-dimensional query and construction process backtracking. The collaborative evidence storage module (403) is used to store construction logs on the blockchain using blockchain technology, and supports multi-role permission management and collaborative supervision.

8. A method for intelligent dynamic compaction based on BeiDou precise positioning, and an intelligent dynamic compaction system based on BeiDou precise positioning as described in any one of claims 1-7, characterized in that, Includes the following steps: S1: Using BeiDou / GNSS and IMU fusion positioning technology, centimeter-level spatial coordinates and attitude data of dynamic compaction machinery are acquired in real time to establish a spatiotemporal reference. S2: Employs dual-channel redundant sensing and PTP hardware clock synchronization to perform millisecond-level fusion acquisition of multi-source construction parameters such as number of blows, drop distance, settlement amount, and hydraulic energy. S3: Based on augmented reality technology, construction parameters are overlaid onto the real-world scene, supporting multi-role interactive monitoring and anomaly warning; S4: Construct a digital twin model that integrates edge and cloud computing, mapping all construction elements in real time and generating blockchain-based evidence logs; S5: Utilize digital twin simulation and deep reinforcement learning to dynamically optimize impact energy distribution and path planning strategies; S6: Transform decision commands into electro-hydraulic control signals and continuously revise the decision model based on construction feedback data.

Images