An intelligent transfer trolley for assembled subway station construction and a transfer method thereof

Abstract

This invention discloses an intelligent transfer trolley and its transfer method for prefabricated subway station construction, belonging to the field of prefabricated subway construction. It includes a transfer trolley body, a lighting mechanism fixed to the front end of the transfer trolley body, a walking assembly, a lifting assembly, a multi-source sensing mechanism, and an intelligent decision-making mechanism mounted on the transfer trolley body. The multi-source sensing mechanism is connected to the intelligent decision-making mechanism, which is connected to both the walking assembly and the lifting assembly. The intelligent decision-making mechanism determines the transfer trajectory and lifting height based on the surrounding environment data collected by the multi-source sensing mechanism. Using this intelligent transfer trolley and its transfer method for prefabricated subway station construction, integrated intelligent transfer of prefabricated components—including transportation, positioning, obstacle avoidance, lifting, and installation—is achieved, balancing millimeter-level positioning accuracy, active obstacle avoidance safety, and multi-device collaborative efficiency, thus meeting the requirements of intelligent transfer.

Description

Technical Field

[0001] This invention relates to the field of prefabricated subway construction technology, and in particular to an intelligent transfer trolley for prefabricated subway station construction and its transfer method. Background Technology

[0002] In recent years, rail transit (especially subway) construction has entered a stage of "large-scale and intensive" development. Prefabricated construction technology has become the mainstream technology for subway station structural construction due to its advantages such as less on-site work, shorter construction period, and less construction waste. Prefabricated subway stations rely on large components such as prefabricated segments, prefabricated side walls, and prefabricated columns (each component weighing 5-20 tons). The construction process requires the transfer of components from the ground storage area to the underground construction area and then to the installation station. However, the underground construction space of subways is characterized by its small size, enclosed environment, and complex environment (poor lighting and many obstacles), which places stringent requirements on the accuracy, efficiency, and safety of component transfer. Traditional transfer technologies are no longer adequate to meet this upgraded demand.

[0003] Specifically, the existing component transportation mainly relies on two types of equipment: one is ordinary heavy-duty flatbed trucks, which are diesel-powered and only equipped with basic walking wheel sets, and whose function is limited to horizontal transportation; the other is customized low trailers, which have optimized equipment height for underground spaces, but lack lifting and precise positioning functions, and need to be used with cranes to assist in adjusting the height of components.

[0004] In summary, existing component transportation mainly relies on ordinary heavy-duty flatbed trucks and diesel-powered trailers, which have limited functions and can only achieve transportation. They lack lifting and adjustment and precise positioning functions, and are too large to be suitable for the narrow construction space of subways. Summary of the Invention

[0005] The purpose of this invention is to provide an intelligent transfer trolley and its transfer method for prefabricated subway station construction, thereby solving the above-mentioned technical problems.

[0006] To achieve the above objectives, the present invention provides an intelligent transfer trolley for prefabricated subway station construction, including a transfer trolley body, a lighting mechanism fixed to the front end of the transfer trolley body, a walking assembly, a lifting assembly, a multi-source sensing mechanism, and an intelligent decision-making mechanism. The multi-source sensing mechanism includes a lidar disposed around the transfer trolley body, a vision camera disposed at the front end of the transfer trolley body, a speed sensor disposed inside the transfer trolley body, a laser locator, and a drone that communicates with the intelligent decision-making mechanism and is equipped with a lidar scanning system and a gimbal camera. The multi-source sensing mechanism is connected to the intelligent decision-making mechanism, which is connected to the walking assembly and the lifting assembly respectively. It is used to determine the transfer trajectory and lifting height based on the surrounding environment collected by the multi-source sensing mechanism.

[0007] A method for transporting intelligent transport trolleys used in the construction of prefabricated subway stations includes the following steps: S1. Construction scene preprocessing and calibration of multi-source sensing mechanism: In the prefabricated component stacking area, the prefabricated components are hoisted to the top of the lifting platform of the transfer trolley, and the prefabricated subway station BIM model and multi-source sensing parameters are integrated to construct a BIM semantic map. The joint calibration of laser locator, lidar, vision camera and drone is completed through least squares optimization, and sensor calibration matrix and collaborative calibration parameters under a unified coordinate system are generated. S2, Multi-source fusion environmental perception and state extraction: Based on the calibration parameters and BIM semantic map output by S1, multi-source perception data is collected. Through ICP point cloud registration and weighted fusion algorithm, static obstacle coordinates and dimensions and dynamic obstacle state information are extracted to generate environmental semantic fusion map and real-time state set of obstacle-transfer cart-component. S3. Based on the real-time state set of obstacle-transfer trolley-component output by S2, calculate the heading angle and velocity of dynamic obstacles through historical data, use a linear prediction model to estimate the future time domain position of dynamic obstacles, and construct a risk assessment matrix by combining multi-dimensional indicators of distance, load, space and rise and fall, and output the future state set of dynamic obstacles and risk weight distribution. S4. Based on the BIM semantic map of S1, the environmental semantic fusion map of S2, and the dynamic obstacle future state set and risk matrix of S3, an improved dynamic window that integrates future obstacle constraints is constructed. A multi-objective optimization function including heading, distance, obstacle avoidance, lifting adaptation and risk penalty is designed to solve the optimal trajectory and control sequence of transportation-positioning-obstacle avoidance-lifting-installation integration, and obtain the optimal single-vehicle transfer trajectory from the stacking area to the installation station. S5. Based on the single-vehicle optimal transfer trajectory and multi-vehicle-crane state set output by S4, construct a collaborative interaction graph with equipment as nodes and communication radius as constraints. Train the collaborative strategy through the Deep Graph Q-Learning algorithm and output conflict-free collaborative optimized trajectory and control commands. S6. Based on the collaborative trajectory and control commands of S5, the motion of the trolley is modeled as a linear stochastic system. The uncertainty of the state is constrained by the covariance control law through convex optimization. The trajectory tracking is realized by combining PID control until the trolley reaches the construction area, drives the lifting assembly to lift the components and verifies the installation heading angle and positioning accuracy.

[0008] Therefore, the present invention employs the above-mentioned intelligent transfer trolley and transfer method for prefabricated subway station construction, which has the following beneficial effects: 1. Significantly improved positioning accuracy: Through multi-source fusion calibration of laser locator, UAV and BIM model, the positioning error is ≤5mm and the heading angle attitude error is ≤0.1rad, which meets the high-precision installation requirements of prefabricated components and solves the problem of traditional positioning cumulative error exceeding 5mm; 2. Significantly improved construction efficiency: Integrated path planning reduces equipment scheduling waiting time, the success rate of component installation in place on the first attempt is ≥95%, and the construction cycle is shortened by 30%; multi-vehicle-crane collaborative scheduling eliminates operation conflicts and increases equipment utilization by more than 40%; 3. Comprehensive upgrade in operational safety: Integrating dynamic obstacle future position prediction and multi-dimensional risk assessment, the success rate of active obstacle avoidance is ≥99%; Combined with physical protection such as metal fences and emergency braking, the risk of component slippage is reduced by 60%, and the accident rate during nighttime operations is close to 0. 4. Enhanced environmental adaptability and robustness: Multi-sensor collaborative calibration solves the sensing challenges of poor lighting and numerous obstructions during subway construction; covariance control combined with PID algorithm resists disturbances such as uneven ground and wind, with trajectory tracking error ≤10mm; 5. Green and environmentally friendly with cost optimization: Lithium battery power achieves zero emissions and noise ≤60dB, meeting green construction requirements; it reduces investment in equipment such as cranes and aerial work platforms, lowering overall construction costs by more than 12% and reducing construction waste by 50%; 6. Enhanced intelligence and automation: It eliminates the need for professional manual operation, automatically completing path planning, obstacle avoidance, collaborative scheduling, and precise installation, reducing reliance on skilled technical workers, while enabling data traceability throughout the entire construction process.

[0009] In summary, this invention integrates multi-sensor-BIM-UAV collaborative positioning, improved dynamic path planning, GNN multi-vehicle-equipment collaborative scheduling, and robust trajectory tracking to achieve integrated intelligent transportation of prefabricated components from the stacking area to the installation station. This is particularly relevant for scenarios such as confined spaces, heavy prefabricated components, dense dynamic obstacles, and multi-process collaboration in prefabricated subway stations. Furthermore, by employing dynamic obstacle prediction, multi-objective optimized path planning, multi-vehicle collaborative obstacle avoidance, millimeter-level positioning, and robust execution, it addresses the pain points of traditional transport vehicles, such as low positioning accuracy, limited functionality, insufficient safety, and low level of intelligence. It also adapts to the complex indoor and outdoor environments of subway stations and meets the requirements of green construction.

[0010] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0011] Figure 1 This is a structural schematic diagram of an intelligent transfer trolley for prefabricated subway station construction as described in this invention; Figure 2This is a flowchart illustrating the transfer method of an intelligent transfer trolley used in the construction of prefabricated subway stations according to the present invention.

[0012] Figure Labels 1. Transfer trolley body; 2. Metal guardrail; 3. Lifting platform; 4. Vision camera; 5. Manual driving assistance assembly; 6. Lighting mechanism; 7. LiDAR; 8. Lifting assembly; 9. Walking assembly. Detailed Implementation

[0013] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are merely illustrative of the embodiments of the present invention and are not intended to limit the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of this application. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout.

[0014] It should be noted that the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion, such as a process, method, system, product, or server that includes a series of steps or units, not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such process, method, product, or device.

[0015] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0016] like Figure 1 As shown, an intelligent transfer trolley for prefabricated subway station construction includes a transfer trolley body 1, a lighting mechanism 6 fixed to the front end of the transfer trolley body 1, a walking assembly 9, a lifting assembly 8, a multi-source sensing mechanism, and an intelligent decision-making mechanism. The multi-source sensing mechanism includes a lidar 7 set around the transfer trolley body 1, a vision camera 4 set at the front end of the transfer trolley body 1, a speed sensor and a laser locator set inside the transfer trolley body 1, and a drone that communicates with the intelligent decision-making mechanism and is equipped with a lidar 7 scanning system and a gimbal camera. The multi-source sensing mechanism is connected to the intelligent decision-making mechanism, which is connected to the walking assembly 9 and the lifting assembly 8 respectively. It is used to determine the transfer trajectory and lifting height based on the surrounding environment collected by the multi-source sensing mechanism.

[0017] The bottom of the transfer trolley body 1 is provided with a walking assembly 9, the top of the transfer trolley body 1 is provided with a lifting assembly 8, the top output end of the lifting assembly 8 is provided with a lifting platform 3, and the top edge of the lifting platform 3 is fixed with a metal guardrail 2. The lifting platform 3 is used to support prefabricated components for construction and construction personnel.

[0018] The main body 1 of the transfer trolley is also equipped with a manual driving assisted assembly 5, which is connected to the walking assembly 9.

[0019] It should be noted that the structure and working principle of the above-mentioned walking assembly and lifting assembly are common knowledge in this field, so their structure and principle will not be described in detail here.

[0020] like Figure 2 As shown, a method for transporting a smart transport trolley used in the construction of prefabricated subway stations includes the following steps: S1. Construction scene preprocessing and calibration of multi-source sensing mechanism: In the prefabricated component stacking area, the prefabricated components are hoisted to the top of the lifting platform of the transfer trolley, and the prefabricated subway station BIM model and multi-source sensing parameters are integrated to construct a BIM semantic map. The joint calibration of laser locator, lidar, vision camera and drone is completed through least squares optimization, and sensor calibration matrix and collaborative calibration parameters under a unified coordinate system are generated. S2, Multi-source fusion environmental perception and state extraction: Based on the calibration parameters and BIM semantic map output by S1, multi-source perception data is collected. Through ICP point cloud registration and weighted fusion algorithm, static obstacle coordinates and dimensions and dynamic obstacle state information are extracted to generate environmental semantic fusion map and real-time state set of obstacle-transfer cart-component. S3. Based on the real-time state set of obstacle-transfer trolley-component output by S2, calculate the heading angle and velocity of dynamic obstacles through historical data, use a linear prediction model to estimate the future time domain position of dynamic obstacles, and construct a risk assessment matrix by combining multi-dimensional indicators of distance, load, space and rise and fall, and output the future state set of dynamic obstacles and risk weight distribution. S4. Based on the BIM semantic map of S1, the environmental semantic fusion map of S2, and the dynamic obstacle future state set and risk matrix of S3, an improved dynamic window that integrates future obstacle constraints is constructed. A multi-objective optimization function including heading, distance, obstacle avoidance, lifting adaptation and risk penalty is designed to solve the optimal trajectory and control sequence of transportation-positioning-obstacle avoidance-lifting-installation integration, and obtain the optimal single-vehicle transfer trajectory from the stacking area to the installation station. S5. Based on the single-vehicle optimal transfer trajectory and multi-vehicle-crane state set output by S4, construct a collaborative interaction graph with equipment as nodes and communication radius as constraints. Train the collaborative strategy through the Deep Graph Q-Learning algorithm and output conflict-free collaborative optimized trajectory and control commands. S6. Based on the collaborative trajectory and control commands of S5, the motion of the trolley is modeled as a linear stochastic system. The uncertainty of the state is constrained by the covariance control law through convex optimization. The trajectory tracking is realized by combining PID control until the trolley reaches the construction area, drives the lifting assembly to lift the components and verifies the installation heading angle and positioning accuracy.

[0021] Step S1 specifically includes the following steps: S11. Obtain the operating parameters of the transfer trolley: maximum linear speed Minimum linear velocity Maximum angular velocity Minimum angular velocity Maximum acceleration Maximum angular acceleration The driving range threshold of the lithium battery used to power the chassis assembly and the lifting assembly. and power distribution ratio and the maximum height of the lifting platform and minimum height ; S12. Semantic Adaptation: Import the BIM model containing the stacking locations of prefabricated components, installation stations, passageway boundaries, and static obstacles, and unify the coordinate system to the construction area's geodetic coordinate system, labeling the coordinates of static obstacles. and size In addition to the three-dimensional coordinates of prefabricated component stacking, the three-dimensional coordinates of the installation station, and the installation heading angle. Obtain the BIM semantic map This data is then input into the intelligent decision-making body as a global environmental benchmark for path planning; among which, The three-dimensional coordinates of the static obstacle. These represent the length, width, and height of a static obstacle, respectively. S13. Perform joint calibration of the lidar, lidar locator, distance sensor, and vision camera mounted on the transport trolley, and define the calibration matrix for each sensor. , Let these represent the lidar, lidar locator, distance sensor, and vision camera, respectively. Using the lidar locator as the reference, the calibration matrix is ​​optimized using the least squares method. : ; In the formula, Indicates the first Raw data from each sensor; This represents the reference data for the laser positioner; Simultaneously, the global point cloud of the construction area is acquired using the lidar scanning system and gimbal camera of the UAV, and multi-source sensor registration is performed on the transport trolley to solve for the collaborative calibration parameters. : ; In the formula, This indicates the positional offset of the drone relative to the transport vehicle; This indicates the relative heading angle between the drone and the transport vehicle; This indicates the coordinates of the prefabricated components observed by the UAV; Indicates the current location of the transfer cart; This represents the heading angle transition matrix.

[0022] Step S2 specifically includes the following steps: S21. Multi-source sensing data acquisition and fusion; The system utilizes radar sensors to collect real-time data on the distance and speed of surrounding obstacles; laser locators to collect data on the position and heading angle of the transport trolley; distance sensors to collect data on the relative position of prefabricated components and the trolley; and vision cameras to collect environmental image data. Simultaneously, the drone transmits global point cloud data of the construction area every 0.5 seconds, and converts the global point cloud data to the coordinate system of the transport vehicle, then combines it with the calibration matrix. The converted UAV data is fused with the sensor data from the transport vehicle itself and BIM data to obtain fused point cloud data. : ; In the formula, , and Indicates the fusion weights; This represents data from multiple sensors placed on the transport trolley; This indicates the coordinates of the target collected by the UAV in the local coordinate system of the transport vehicle after coordinate transformation; This represents the static obstacle point cloud data of the BIM model; S22. Static semantic annotation: from fused point cloud data Filtering and BIM Semantic Map Point cloud clusters for matching static obstacles in the middle and extracting their coordinates. and size Forming a static obstacle set , Indicates the first A static obstacle feature; Dynamic semantic annotation: Identify moving targets and extract their real-time positions using point cloud difference algorithms. ,speed and heading angle Forming a dynamic obstacle set , Indicates the first One dynamic obstacle feature; among which... Represents the three-dimensional coordinates of the moving target; Represents the three-dimensional velocity components of a moving target; Merged point cloud data with semantic annotation The data is converted into a 3D voxel mesh, and a topological map of traversable areas is constructed based on the 3D voxel mesh to obtain an environmental semantic fusion map. ; S23, Extracting the status of the transfer cart Current location of the transfer cart Heading angle ,speed angular velocity and the remaining power of the lithium battery ; Extracting the state of prefabricated components Actual weight of precast components And the relative positions of precast components and transfer trolleys ; S24, Output Obstacle-Transfer Cart-Component Real-Time Status Set .

[0023] Step S3 specifically includes the following steps: S31. Estimate the future temporal location of dynamic obstacles. ; in, , , , , express The position of obstacles in real time; express The three-dimensional coordinates of the dynamic obstacle in real time; S32. Calculate the comprehensive risk for all candidate path points and generate a risk assessment matrix; where the comprehensive risk... The calculation formula is as follows: ; in, ; ; ; ; In the formula, , , and These represent distance risk, load risk, spatial risk, and rise / fall risk, respectively. , , and All represent weighting coefficients; This represents the planar distance between the transport vehicle and the dynamic obstacle at a future time, and , This represents the planar coordinates of the transfer trolley at the current moment; Indicates a safe distance, and ; and These represent the actual weight of the precast component and the maximum permissible load weight of the transfer trolley, respectively. and These represent the width of the passageway and the width of the transfer cart, respectively. Indicates the target height of the lifting platform; This indicates the height of the installation station.

[0024] Step S4 specifically includes the following steps: S41. Construct an improved dynamic window; S411, Basic Dynamic Window Constraints: Define the initial velocity window based on the vehicle's dynamic parameters. and angular velocity window ; S412, Fusion Prediction of Obstacle Constraints: Incorporating the Future Temporal Location of Dynamic Obstacles By eliminating speed pairs that will collide with future obstacles, an improved dynamic window is constructed: ; ; In the formula, and These represent the improved linear velocity window and the improved angular velocity window, respectively. and These represent the candidate control quantities for the speed and angular velocity of the transfer trolley, respectively. This represents the prediction time step size for dynamic obstacles; Indicates that the car is Minimum distance to predicted obstacles during state movement; Indicates the equivalent radius of the transfer cart; Indicates the equivalent radius of the obstacle; Simultaneously integrates lifting adaptation constraints: based on the installation position height. Set speed constraints for the lifting preparation section. and ; S42. Construct the following multi-objective optimization function. : ; in, ; ; ; ; ; In the formula, , , , and These represent heading, distance, obstacle avoidance, takeoff and landing adaptation, and risk penalty, respectively. , , , and All represent weights; This indicates the deviation in heading angle, and , The planar coordinates of the installation station; This indicates the actual distance between the transport vehicle and the current obstacle; S43. Solving for the optimal velocity pair : ; S44, Based on optimal speed pair Generate the optimal transfer trajectory for a single vehicle : ; in, ; ; ; ; In the formula, Represents the optimal transfer trajectory for a single vehicle. The A trajectory point; Indicates the first The three-dimensional coordinates of the transfer cart corresponding to each time step; Indicates the calculation time step; Indicates the first Planar coordinates at each time step; express The heading angle of the constant-time transfer trolley; Indicates the first The heading angle of the transfer trolley at each time step; Indicates the first The target height of the lifting platform corresponding to each time step; S45, Generate drive control sequence and lifting control sequence ;in, This indicates the total transit time.

[0025] Step S5 specifically includes the following steps: S51. Extract the status of all transfer carts. Obtain the multi-vehicle state set And based on multi-vehicle state sets and crane status Building a collaborative interaction graph ; in, ; ; In the formula, and These represent the nodes and edges of the collaborative interaction graph, respectively. Indicates the first Taiwan transfer cart; Indicates the first One crane; This represents the two nodes corresponding to the edge in the collaborative interaction graph (the nodes are either a transfer cart or a crane). and Representing nodes respectively and nodes Plane coordinates; Indicates the communication radius between nodes; S52. Train the collaborative strategy using the Deep Graph Q-Learning algorithm; S521. Define the cooperative Q function. : ; In the formula, Represents a graph attention network; Represents the feature vector of a node in a collaborative interaction graph; This represents the model parameters of the graph attention network; This represents the weight matrix of the output layer of the graph attention network. Indicates the bias term; Indicates coordinated actions (speed adjustment, trajectory deviation, work sequence adjustment); S522, Constructing a collaborative reward function : ; In the formula, A reward is given for successfully completing the installation collaboratively; Indicates a collision penalty; This indicates a delayed punishment; This indicates a penalty for a conflict between the operating area of ​​the transfer trolley and the crane; S523. Obtain the optimal collaborative strategy through reinforcement learning training, and output the trajectory adjustment amount for each transfer trolley. and control adjustment amount , Indicates the first The trajectory plane offset of the transfer trolley. They represent the first Adjustments to the speed, angular velocity, and height of the lifting platform of the transfer trolley; S53. Generate conflict-free collaborative optimization trajectories. and control commands ; in, ; ; ; ; In the formula, Indicates the first The three-dimensional coordinates of the cooperative trajectory points; Indicates the first The three-dimensional coordinates of the optimal trajectory point for a single vehicle; Indicates the lifting height after coordination; , and These represent the velocity, angular velocity, and platform height after coordination, respectively. , and They represent the first The speed adjustment, angular velocity adjustment, and height adjustment of the transfer trolley.

[0026] Step S6 specifically includes the following steps: S61, Trajectory Tracking; S611. Model the motion of the car as a linear stochastic system, and the state... random disturbance The following system equations are constructed: ; in, ; ; ; In the formula, , and These represent the state transition matrix, control matrix, and disturbance matrix, respectively. The covariance matrix representing the perturbation; S612. Determine the covariance control objective: Make the state covariance... satisfy , Represents the maximum permissible covariance matrix; robust control commands are solved through convex optimization. : ; ; In the formula, Denotes the derivative of the covariance matrix; Indicates the transpose operation; S613, Integrating PID control to achieve trajectory tracking: ; In the formula, Indicates robust driver instructions; , and These represent the proportional, integral, and differential coefficients, respectively. Indicates the tracking error, and , express The three-dimensional coordinates of the collaborative optimization trajectory points at each moment; Indicates tracking error The derivative; S62, Lifting control; S621, Lifting Platform Height Control: Based on Construct the following control law: ; In the formula, and They represent Time and The actual height of the lifting platform at any given time; Indicates the increase / decrease coefficient; S623, Output lifting / lowering execution signal And in the current Determine that the lifting / lowering position is reached; S63. Component installation and status verification; S631. Carriage positioning accuracy verification: When the carriage arrives at the installation station... And positioning error At that time, confirm that the positioning is in place; S632. Component heading angle adjustment: Adjust the heading angle of the transfer trolley using a laser locator and a drone in coordination to ensure the correct heading angle for component installation. , This indicates the target installation heading angle of the prefabricated component corresponding to the installation station; S633, output installation ready signal to notify the crane or construction personnel to finally fix the prefabricated components, and the transportation phase is completed.

[0027] During the transfer process, the distance between the vehicle and obstacles, the heading angle of the vehicle and the ambient light are monitored in real time to divide the state into three levels: safety, warning and emergency. When the warning is issued, the path and speed are dynamically adjusted. When the emergency is triggered, braking and audible and visual alarms are triggered. At the same time, the manual intervention interface is retained. Simultaneously, by collecting positioning accuracy, obstacle avoidance success rate, and collaborative efficiency indicators during the transportation process, the system analyzes obstacle prediction errors and trajectory tracking errors, iteratively optimizes prediction model parameters, path planning objective function weights, and collaborative strategy network parameters, and dynamically updates the BIM semantic map and sensor calibration matrix.

[0028] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. An intelligent transfer trolley for prefabricated subway station construction, characterized in that: It includes a transport vehicle body, a lighting mechanism fixed to the front end of the transport vehicle body, a walking assembly, a lifting assembly, a multi-source sensing mechanism, and an intelligent decision-making mechanism. The multi-source sensing mechanism includes a lidar set around the transport vehicle body, a visual camera set at the front end of the transport vehicle body, a speed sensor set inside the transport vehicle body, a laser locator, and a drone that communicates with the intelligent decision-making mechanism and is equipped with a lidar scanning system and a gimbal camera. The multi-source sensing mechanism is connected to the intelligent decision-making mechanism, which is connected to the walking assembly and the lifting assembly respectively. It is used to determine the transfer trajectory and lifting height based on the surrounding environment collected by the multi-source sensing mechanism.

2. The intelligent transfer trolley for prefabricated subway station construction according to claim 1, characterized in that: The bottom of the transfer trolley is equipped with a walking assembly, the top of the transfer trolley is equipped with a lifting assembly, the top output end of the lifting assembly is equipped with a lifting platform, and the top edge of the lifting platform is fixed with a metal guardrail. The lifting platform is used to support prefabricated components for construction and construction personnel.

3. The intelligent transfer trolley for prefabricated subway station construction according to claim 2, characterized in that: The transport vehicle is also equipped with a manual driving assisted assembly, which is connected to the walking assembly.

4. The method for transporting an intelligent transport trolley used in the construction of prefabricated subway stations as described in claim 3, characterized in that: Includes the following steps: S1. Construction scene preprocessing and calibration of multi-source sensing mechanism: In the prefabricated component stacking area, the prefabricated components are hoisted to the top of the lifting platform of the transfer trolley, and the prefabricated subway station BIM model and multi-source sensing parameters are integrated to construct a BIM semantic map. The joint calibration of laser locator, lidar, vision camera and drone is completed through least squares optimization, and sensor calibration matrix and collaborative calibration parameters under a unified coordinate system are generated. S2, Multi-source fusion environmental perception and state extraction: Based on the calibration parameters and BIM semantic map output by S1, multi-source perception data is collected. Through ICP point cloud registration and weighted fusion algorithm, static obstacle coordinates and dimensions and dynamic obstacle state information are extracted to generate environmental semantic fusion map and real-time state set of obstacle-transfer cart-component. S3. Based on the real-time state set of obstacle-transfer trolley-component output by S2, calculate the heading angle and velocity of dynamic obstacles through historical data, use a linear prediction model to estimate the future time domain position of dynamic obstacles, and construct a risk assessment matrix by combining multi-dimensional indicators of distance, load, space and rise and fall, and output the future state set of dynamic obstacles and risk weight distribution. S4. Based on the BIM semantic map of S1, the environmental semantic fusion map of S2, and the dynamic obstacle future state set and risk matrix of S3, an improved dynamic window that integrates future obstacle constraints is constructed. A multi-objective optimization function including heading, distance, obstacle avoidance, lifting adaptation and risk penalty is designed to solve the optimal trajectory and control sequence of transportation-positioning-obstacle avoidance-lifting-installation integration, and obtain the optimal single-vehicle transfer trajectory from the stacking area to the installation station. S5. Based on the single-vehicle optimal transfer trajectory and multi-vehicle-crane state set output by S4, construct a collaborative interaction graph with equipment as nodes and communication radius as constraints. Train the collaborative strategy through the Deep Graph Q-Learning algorithm and output conflict-free collaborative optimized trajectory and control commands. S6. Based on the collaborative trajectory and control commands of S5, the motion of the trolley is modeled as a linear stochastic system. The uncertainty of the state is constrained by the covariance control law through convex optimization. The trajectory tracking is realized by combining PID control until the trolley reaches the construction area, drives the lifting assembly to lift the components and verifies the installation heading angle and positioning accuracy.

5. The method for transporting an intelligent transport trolley used in the construction of prefabricated subway stations according to claim 4, characterized in that: Step S1 specifically includes the following steps: S11. Obtain the operating parameters of the transfer trolley: maximum linear speed Minimum linear velocity Maximum angular velocity Minimum angular velocity Maximum acceleration Maximum angular acceleration The driving range threshold of the lithium battery used to power the chassis assembly and the lifting assembly. and power distribution ratio and the maximum height of the lifting platform and minimum height ; S12. Semantic Adaptation: Import the BIM model containing the stacking locations of prefabricated components, installation stations, passageway boundaries, and static obstacles, and unify the coordinate system to the construction area's geodetic coordinate system, labeling the coordinates of static obstacles. and size In addition to the three-dimensional coordinates of prefabricated component stacking, the three-dimensional coordinates of the installation station, and the installation heading angle. Obtain the BIM semantic map This data is then input into the intelligent decision-making body as a global environmental benchmark for path planning; among which, The three-dimensional coordinates of the static obstacle. These represent the length, width, and height of a static obstacle, respectively. S13. Perform joint calibration of the lidar, lidar locator, distance sensor, and vision camera mounted on the transport trolley, and define the calibration matrix for each sensor. , Let these represent the lidar, lidar locator, distance sensor, and vision camera, respectively. Using the lidar locator as the reference, the calibration matrix is ​​optimized using the least squares method. : ； In the formula, Indicates the first Raw data from each sensor; This represents the reference data for the laser positioner; Simultaneously, the global point cloud of the construction area is acquired using the lidar scanning system and gimbal camera of the UAV, and multi-source sensor registration is performed on the transport trolley to solve for the collaborative calibration parameters. : ； In the formula, This indicates the positional offset of the drone relative to the transport vehicle; This indicates the relative heading angle between the drone and the transport vehicle; This indicates the coordinates of the prefabricated components as observed by the UAV; Indicates the current location of the transfer cart; This represents the heading angle transition matrix.

6. The method for transporting an intelligent transport trolley used in the construction of prefabricated subway stations according to claim 5, characterized in that: Step S2 specifically includes the following steps: S21. Multi-source sensing data acquisition and fusion; The system utilizes radar sensors to collect real-time data on the distance and speed of surrounding obstacles; laser locators to collect data on the position and heading angle of the transport trolley; distance sensors to collect data on the relative position of prefabricated components and the trolley; and vision cameras to collect environmental image data. Simultaneously, the drone transmits global point cloud data of the construction area every 0.5 seconds, and converts the global point cloud data to the coordinate system of the transport vehicle, then combines it with the calibration matrix. The converted UAV data is fused with the sensor data from the transport vehicle itself and BIM data to obtain fused point cloud data. : ； In the formula, , and Indicates the fusion weights; This represents data from multiple sensors placed on the transport trolley; This indicates the coordinates of the target collected by the UAV in the local coordinate system of the transport vehicle after coordinate transformation; This represents the static obstacle point cloud data of the BIM model; S22. Static semantic annotation: from fused point cloud data Filtering and BIM Semantic Map Point cloud clusters for matching static obstacles in the middle and extracting their coordinates. and size Forming a static obstacle set , Indicates the first A static obstacle feature; Dynamic semantic annotation: Identify moving targets and extract their real-time positions using point cloud difference algorithms. ,speed and heading angle Forming a dynamic obstacle set , Indicates the first One dynamic obstacle feature; among which... Represents the three-dimensional coordinates of the moving target; Represents the three-dimensional velocity components of a moving target; Merged point cloud data with semantic annotation The data is converted into a 3D voxel mesh, and a topological map of traversable areas is constructed based on the 3D voxel mesh to obtain an environmental semantic fusion map. ; S23, Extracting the status of the transfer cart Current location of the transfer cart Heading angle ,speed angular velocity and remaining lithium battery power ; Extracting the state of prefabricated components Actual weight of precast components And the relative positions of precast components and transfer trolleys ; S24, Output Obstacle-Transfer Cart-Component Real-Time Status Set .

7. The method for transporting an intelligent transport trolley used in the construction of prefabricated subway stations according to claim 6, characterized in that: Step S3 specifically includes the following steps: S31. Estimate the future temporal location of dynamic obstacles. ; in, , , , , express The position of obstacles in real time; express The three-dimensional coordinates of the dynamic obstacle in real time; S32. Calculate the comprehensive risk for all candidate path points and generate a risk assessment matrix; where the comprehensive risk... The calculation formula is as follows: ； in, ； ； ； ； In the formula, , , and These represent distance risk, load risk, spatial risk, and rise / fall risk, respectively. , , and All represent weighting coefficients; This represents the planar distance between the transport vehicle and the dynamic obstacle at a future time, and , This represents the planar coordinates of the transfer trolley at the current moment; Indicates a safe distance, and ; and These represent the actual weight of the precast component and the maximum permissible load weight of the transfer trolley, respectively. and These represent the width of the passageway and the width of the transfer cart, respectively. Indicates the target height of the lifting platform; This indicates the height of the installation station.

8. The method for transporting an intelligent transport trolley used in the construction of prefabricated subway stations according to claim 7, characterized in that: Step S4 specifically includes the following steps: S41. Construct an improved dynamic window; S411, Basic Dynamic Window Constraints: Define the initial velocity window based on the vehicle's dynamic parameters. and angular velocity window ; S412, Fusion Prediction of Obstacle Constraints: Incorporating the Future Temporal Location of Dynamic Obstacles By eliminating speed pairs that will collide with future obstacles, an improved dynamic window is constructed: ； ； In the formula, and These represent the improved linear velocity window and the improved angular velocity window, respectively. and These represent the candidate control quantities for the speed and angular velocity of the transfer trolley, respectively. This represents the prediction time step size for dynamic obstacles; Indicates that the car is Minimum distance to predicted obstacles during state movement; Indicates the equivalent radius of the transfer cart; Indicates the equivalent radius of the obstacle; Simultaneously integrates lifting adaptation constraints: based on the installation position height. Set speed constraints for the lifting preparation section. and ; S42. Construct the following multi-objective optimization function. : ； in, ； ； ； ； ； In the formula, , , , and These represent heading, distance, obstacle avoidance, takeoff and landing adaptation, and risk penalty, respectively. , , , and All represent weights; This indicates the deviation in heading angle, and , The planar coordinates of the installation station; This indicates the actual distance between the transport vehicle and the current obstacle; S43. Solving for the optimal velocity pair : ; S44, Based on optimal speed pair Generate the optimal transfer trajectory for a single vehicle : ； in, ； ； ； ； In the formula, Represents the optimal transfer trajectory for a single vehicle. The A trajectory point; Indicates the first The three-dimensional coordinates of the transfer cart corresponding to each time step; Indicates the calculation time step; Indicates the first Planar coordinates at each time step; express The heading angle of the constant-time transfer trolley; Indicates the first The heading angle of the transfer trolley at each time step; Indicates the first The target height of the lifting platform corresponding to each time step; S45, Generate drive control sequence and lifting control sequence ;in, This indicates the total transit time.

9. A method for transporting an intelligent transport trolley used in the construction of prefabricated subway stations according to claim 8, characterized in that: Step S5 specifically includes the following steps: S51. Extract the status of all transfer carts. Obtain the multi-vehicle state set And based on multi-vehicle state sets and crane status Building a collaborative interaction graph ; in, ； ； In the formula, and These represent the nodes and edges of the collaborative interaction graph, respectively. Indicates the first Taiwan transfer cart; Indicates the first One crane; This represents the two nodes corresponding to an edge in the collaborative interaction graph; and Representing nodes respectively and nodes Plane coordinates; Indicates the communication radius between nodes; S52. Train the collaborative strategy using the Deep Graph Q-Learning algorithm; S521. Define the cooperative Q function. : ； In the formula, Represents a graph attention network; Represents the feature vector of a node in a collaborative interaction graph; This represents the model parameters of the graph attention network; This represents the weight matrix of the output layer of the graph attention network. Indicates the bias term; Indicates coordinated action; S522, Constructing a collaborative reward function : ； In the formula, A reward is given for successfully completing the installation collaboratively; Indicates a collision penalty; This indicates a delayed punishment; This indicates a penalty for a conflict between the operating area of ​​the transfer trolley and the crane; S523. Obtain the optimal collaborative strategy through reinforcement learning training, and output the trajectory adjustment amount for each transfer trolley. and control adjustment amount , Indicates the first The trajectory plane offset of the transfer trolley. They represent the first Adjustments to the speed, angular velocity, and height of the lifting platform of the transfer trolley; S53. Generate conflict-free collaborative optimization trajectories. and control commands ; in, ； ； ； ； In the formula, Indicates the first The three-dimensional coordinates of the cooperative trajectory points; Indicates the first The three-dimensional coordinates of the optimal trajectory point for a single vehicle; Indicates the lifting height after coordination; , and These represent the velocity, angular velocity, and platform height after coordination, respectively. , and They represent the first The speed adjustment, angular velocity adjustment, and height adjustment of the transfer trolley.

10. A method for transporting an intelligent transport trolley used in the construction of prefabricated subway stations according to claim 9, characterized in that: Step S6 specifically includes the following steps: S61, Trajectory Tracking; S611. Model the motion of the car as a linear stochastic system, and the state... random disturbance The following system equations are constructed: ； in, ； ； ； In the formula, , and These represent the state transition matrix, control matrix, and disturbance matrix, respectively. The covariance matrix representing the perturbation; S612. Determine the covariance control objective: Make the state covariance... satisfy , Represents the maximum permissible covariance matrix; robust control commands are solved through convex optimization. : ； ； In the formula, Denotes the derivative of the covariance matrix; Indicates the transpose operation; S613, Integrating PID control to achieve trajectory tracking: ； In the formula, Indicates robust driver instructions; , and These represent the proportional, integral, and differential coefficients, respectively. Indicates the tracking error, and , express The three-dimensional coordinates of the collaborative optimization trajectory points at each moment; Indicates tracking error The derivative; S62, Lifting control; S621, Lifting Platform Height Control: Based on Construct the following control law: ； In the formula, and They represent Time and The actual height of the lifting platform at any given time; Indicates the increase / decrease coefficient; S623, Output lifting / lowering execution signal And in the current Determine that the lifting / lowering position is reached; S63. Component installation and status verification; S631. Carriage positioning accuracy verification: When the carriage arrives at the installation station... And positioning error At that time, confirm that the positioning is in place; S632. Component heading angle adjustment: Adjust the heading angle of the transfer trolley using a laser locator and a drone in coordination to ensure the correct heading angle for component installation. , This indicates the target installation heading angle of the prefabricated component corresponding to the installation station; S633, output installation ready signal to notify the crane or construction personnel to finally fix the prefabricated components, and the transportation phase is completed.

Images