Intelligent dispatching and remote control method for tower crane

Abstract

The invention provides an intelligent dispatching and remote control method for a tower crane, which is applied to the field of fabricated house building construction and engineering equipment and aims to solve the problems of difficulty in dispatching the tower crane, low construction efficiency, severe operation environment and high safety risk in a clustered fabricated building. The method comprises the following steps: S1, constructing a distributed system architecture comprising a physical layer, a sensing control layer, a data processing decision layer, an application layer and a communication layer, and deploying equipment; s2, collecting environment, tower crane state and BIM model related multi-source data, and transmitting the data to a data fusion processing center for fusion; s3, planning an optimal hoisting path and a task sequence through the intelligent scheduling subsystem based on an ant colony algorithm; s4, a control mode is selected, a hoisting instruction is issued to execute a task, and forced braking is triggered in emergency; and S5, mapping a construction scene through a digital twin system, monitoring and alarming through an intelligent monitoring system, feeding back data to realize closed-loop optimization, and recording a digital file. According to the method, tower crane dispatching and construction efficiency is improved, construction safety is guaranteed, and intelligent construction of the fabricated building is promoted.

Description

Technical Field

[0001] This invention relates to the field of intelligent tower crane scheduling, and in particular to a method for intelligent tower crane scheduling and remote control. Background Technology

[0002] Prefabricated buildings offer advantages such as high construction efficiency, less pollution, and higher construction quality, leading to their increasing adoption worldwide, particularly in Saudi Arabia where large-scale clusters of prefabricated buildings have frequently appeared in recent years. In the construction of prefabricated buildings, tower cranes are the most crucial lifting equipment, responsible for hoisting and installing prefabricated components. However, traditional tower crane scheduling often relies on manual experience, resulting in low construction efficiency and poor scheduling outcomes. This is especially true when a single tower crane is handling multiple buildings simultaneously, making scheduling even more challenging and prone to causing slow progress and resource waste. Furthermore, traditional tower crane operation relies heavily on on-site operators, resulting in harsh working environments and potential safety hazards due to limited visibility.

[0003] Therefore, improving tower crane scheduling efficiency, construction efficiency, and construction safety, enhancing the comfort of tower crane operators, and increasing the level of intelligent construction in prefabricated buildings have become current challenges in clustered prefabricated installation. To address these requirements and problems, this invention proposes an intelligent scheduling and remote control method for tower cranes: First, a tower crane scheduling optimization algorithm is proposed to intelligently plan the hoisting path and task allocation of tower cranes based on construction needs, improving work efficiency and resource utilization. Second, remote control and partially unmanned operation are achieved by configuring various sensors, reducing the labor intensity and safety risks for operators and improving construction efficiency. Finally, a digital twin system and intelligent monitoring system are developed in conjunction with BIM models to synchronize construction progress in real time, providing data support for the automated hoisting and scheduling system of tower cranes.

[0004] This technical solution addresses several issues in traditional tower crane scheduling and operation: In terms of scheduling, intelligent algorithm optimization solves the challenge of scheduling tower cranes for simultaneous hoisting of multiple buildings, improving tower crane utilization. Regarding operation, remote control and partial unmanned operation functions address safety and worker comfort issues during on-site operations. Summary of the Invention

[0005] The main objective of this invention is to provide a method for intelligent scheduling and remote control of tower cranes, which solves the problems of high scheduling difficulty, low construction efficiency, harsh operating environment for operators, and high construction risks in the operation of tower cranes in the aforementioned clustered prefabricated buildings.

[0006] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: a method for intelligent scheduling and remote control of tower cranes, the method comprising: S1. Build a distributed system architecture including the physical layer, perception and control layer, data processing and decision-making layer, application layer and communication layer, and deploy the corresponding devices; S2. Collect multi-source data related to the environment, tower crane status, and BIM model, and transmit them to the data fusion processing center for fusion processing; S3. Based on the ant colony algorithm, the intelligent scheduling subsystem plans the optimal hoisting path and task sequence, and outputs the scheduling results. S4. Select the corresponding control mode, issue hoisting instructions to execute the task, and trigger forced braking in an emergency. S5. The construction scene is mapped through a digital twin system, the intelligent monitoring system monitors the status and issues alarms, and the feedback data achieves closed-loop optimization and records digital archives.

[0007] In the preferred embodiment, in step S1, a smart tower crane with a closed-loop control system, prefabricated houses, and prefabricated components are deployed at the physical layer, with the smart tower crane installed around the prefabricated houses arranged in a triangular pattern. The perception and control layer is equipped with environmental perception sensors, tower crane status perception sensors, and execution control terminals. The environmental perception sensors include high-precision lidar and wide-angle high-definition cameras. The wide-angle high-definition cameras are equipped with AI discrimination programs to identify the status of the hook. The tower crane status perception sensors include a rotary encoder, a trolley position sensor, a lifting height sensor, a torque sensor, an anemometer, a tilt sensor, and a vibration sensor. The execution control terminal includes the tower crane control system and a high-performance edge computing unit. The edge computing unit is installed on the tower crane body and is used for data preprocessing and local obstacle avoidance calculation. The data processing decision-making layer establishes an intelligent scheduling subsystem and a data fusion processing center; The application layer is configured with a remote driver's cab, digital twin, and intelligent monitoring system; The communication layer is equipped with industrial-grade switches and 5G base stations.

[0008] In the preferred embodiment, the data preprocessing and local obstacle avoidance of the edge computing unit include: S101, Data preprocessing step: The steps for the edge computing unit to receive multi-source data collected by sensors in the perception and control layer include: S1011. Data Filtering: The edge computing unit filters out obstacle data from the construction scene collected by high-precision lidar and wide-angle high-definition cameras in the environmental perception sensors. This obstacle data includes building corners, temporary facilities, and other components. Simultaneously, it filters out real-time tower crane operation data collected by the tower crane status perception sensors, including the slewing encoder, trolley position sensor, lifting height sensor, torque sensor, anemometer, tilt sensor, and vibration sensor. This operation data includes slewing angle, trolley position, hook height, torque, wind speed, tower verticality, and vibration parameters. Invalid data is removed; invalid data refers to abnormal values ​​exceeding the normal range caused by sensor malfunctions. S1012. Data spatiotemporal alignment: Based on the time synchronization signal of the 5G base station in the communication layer, a timestamp is added to each type of filtered data. Simultaneously, the 3D coordinate data of obstacles collected by LiDAR and the 2D image obstacle data collected by the camera are uniformly converted into a polar coordinate system with the tower crane's rotation center as the origin. The polar coordinate system is established with the tower crane's rotation center as the origin and the direction pointing towards the prefabricated house A as the positive polar axis. The 3D coordinates (x, y, z) of obstacles collected by LiDAR are converted into polar coordinates. The formula is , , , where r is the polar radius, θ is the polar angle and its range is [0, 2π), and h is the height, to achieve spatial coordinate alignment of multi-source data and obtain standardized data that is consistent in time and space; S1013. Data Denoising and Compression: A moving average filtering algorithm is used to denoise the standardized tower crane vibration parameters and wind speed data. The filtering formula is as follows: ,in Let N be the parameter value after noise reduction at time t, and N be the sliding window length. The sliding window length N is dynamically adjusted according to the sensor sampling frequency. When, N is 10; when When, N is 8; when When N is 5, for Original parameter values ​​at time. The sampling interval for the sensor is defined; the point cloud data of the lidar is compressed using a voxel grid downsampling algorithm to retain the key contour features of obstacles, reduce data transmission and computational load, and obtain the preprocessed effective data. S102, Local obstacle avoidance calculation steps: Based on the preprocessed valid data, the edge computing unit performs the following local obstacle avoidance steps for the tower crane: S1021. Obstacle Modeling: Based on the preprocessed polar coordinate data from high-precision LiDAR, construct a polar coordinate model of static obstacles in the construction scene. Static obstacles include the four corners of buildings and fixed temporary facilities. Each obstacle is represented as a set of regions in polar coordinates. ,in The extreme diameter from the obstacle to the tower crane's slewing center. The polar angle of the obstacle. , The range of the obstacle's extreme diameter. , The range of the polar angle of the obstacle; Combining the real-time location of dynamic obstacles identified by the AI ​​discrimination program of the wide-angle high-definition camera, including components to be hoisted and construction personnel, the AI ​​program identifies the type of dynamic obstacle through a target detection algorithm and calculates the extreme radius r from the dynamic obstacle to the tower crane's slewing center based on a preset distance calibration model within the camera. The camera's intrinsic parameters include focal length f and pixel size d. The calculation formula is as follows: Where H is the actual height of the dynamic obstacle, the component height is known, and the personnel height is set by default as 1.3-1.8 meters. To obtain the pixel height of the dynamic obstacle in the camera image, the polar coordinate information of the dynamic obstacle is added to the obstacle model, thus updating the real-time obstacle set. ; S1022. Tower crane motion trajectory prediction: Based on real-time data preprocessed by the tower crane status sensing sensors, the current slewing angle of the tower crane is obtained. The current position of the trolley is extremely far. Current height of the hook Based on the next control commands issued by the tower crane control system, the control commands include the slewing angular velocity ω, the trolley luffing speed v, and the hook lifting speed. Predicting the future The trajectory of the tower crane hook within a given time period. The value is set between 0.5 and 2 seconds, depending on the tower crane's operating speed. The prediction formula for the trajectory in polar coordinates is as follows: , , ,in , , They are respectively The hook's predicted polar diameter, polar angle, and height are constantly being measured. S1023 Collision detection: Determines the relationship between the predicted trajectory and the real-time obstacle set. Whether there is a collision risk, the collision determination criteria are: if there is , making and Within the obstacle height range, the obstacle height range is obtained from LiDAR data and represented as... That is, satisfying If so, it is determined that there is a risk of collision; S1024. Obstacle avoidance command generation: If a collision risk is determined, the edge computing unit calculates the obstacle avoidance adjustment amount based on the collision detection results and generates a local obstacle avoidance control command. If the collision risk originates from the turning direction, then adjust the turning angular velocity using the following formula: ,in To avoid obstacles, the safe polar angle, the safe polar angle The method for determining is: if Within the range of the polar angle of the obstacle inside, when hour, ;otherwise , To ensure safety and redundancy, the angle is set between 5 and 10 degrees, depending on the complexity of the construction scenario. The sign function ensures that the tower crane rotates in the direction away from the obstacle; If the collision risk originates from the direction of the car's change of direction, then adjust the car's change of speed using the following formula: ,in To avoid obstacles, the safe extreme diameter. Take the nearest polar diameter outside the range of the obstacle's polar diameter; If the collision risk originates from the vertical direction, adjust the hook lifting speed using the following formula: ,in Safe height to avoid obstacles Take the nearest height outside the range of obstacle heights; The adjusted , , As a local obstacle avoidance control command, it is sent to the tower crane control system to realize local obstacle avoidance of the tower crane, and at the same time, the obstacle avoidance process data is uploaded to the data fusion processing center.

[0009] In the preferred embodiment, in step S2, the environmental data includes building construction progress data detected by high-precision lidar, as well as data on the status and location of prefabricated components, the status and location of construction personnel, the position and status of hooks, and the location of lifting points collected by wide-angle high-definition cameras; the tower crane status data includes slewing angle, trolley running position, hook lifting height, tower crane torque, wind speed and direction, tower verticality, and tower crane vibration parameters. BIM model data includes the dimensions, weight, installation location of prefabricated components, and construction progress data for each building; Fusion processing includes spatiotemporal alignment, filtering, and fusion of multi-source data to generate unified and accurate data.

[0010] In the preferred scheme, in step S3, the scheduling objective of the ant colony algorithm is to minimize the total job time, and its objective function model is:

[0011] In the formula, For the task sequence, Starting from, The total number of components is equivalent to the total number of tasks. To the task To the mission The transfer time For the task Fixed working hours, including pickup and installation; The initialization of the ant colony algorithm includes: setting the number of ants to 1-4 times the number of hoisting tasks, and setting the pheromone factor. heuristic factor pheromone evaporation rate pheromone constant Maximum number of iterations Next, create a pheromone matrix of (task number + 1) × (task number + 1), with the initial value set to 0.1; From the task To the mission Transfer time The calculation model is as follows: ; In the formula, For the task The polar angle of the pickup location, For the task Polar angle of the installation location The tower crane's rotational angular velocity (degrees per second). For the task The extreme diameter of the pickup location, For the task Extreme diameter of the installation location, This refers to the tower crane's luffing speed.

[0012] In the preferred scheme, the state transition rule of the ant colony algorithm in step S3 is as follows: ; In the formula, For the first In the nth iteration Only ants from the task Transfer to task The probability, For the first Task during the next iteration To the mission The concentration of pheromones along the path, For the first Task during the next iteration To the mission Heuristic information ( ), For the first The set of tasks that the ant has not yet visited; The process of planning the optimal hoisting path and task sequence includes: Set a virtual starting point, i.e., the tower crane center, for each ant; clear the list of visited tasks and the task sequence; filter out candidate tasks whose predecessor tasks have not been completed from the unvisited tasks; if there are no available tasks, select the task with the fewest missing predecessor tasks and record the number of times; calculate the movement time from the current position to the pick-up point of each candidate task, i.e., take the larger value of the rotation time and the amplitude change time. Based on the above state transition rules, the task selection probability is calculated, the task is selected by roulette wheel and added to the path sequence, the task is marked as visited and the current position is updated to the task installation point. The above process is repeated until all tasks have been visited. Calculate the total path time for each ant. The total path time is the sum of the movement time between tasks and the fixed operation time of each task.

[0013] In the preferred scheme, step S3, the pheromone update mechanism of the ant colony algorithm includes a global update, the specific formula of which is: ; ; ; In the formula, For the first Task during the next iteration To the mission The concentration of pheromones along the path, For path pheromone increment, The number of ants. For the first Only ants on the path The increase in pheromones left on the surface For the first The total path length after an ant completes the entire path, which is the total operation time; Before outputting the scheduling results, the process also includes: comparing the total path time of all ants in this round to determine the optimal path for this round; if the optimal path for this round is better than the historical optimal path, then updating the global optimal path; and performing pheromone evaporation and incremental updates on the pheromone of all paths according to the pheromone update mechanism described above. The planning termination condition is reaching the maximum number of iterations. The second or optimal path shows no improvement after 20-40 consecutive attempts; the scheduling results include the optimal task sequence, total time estimate and performance indicators, including tower crane utilization, time saving rate and constraint satisfaction statistics.

[0014] In the preferred embodiment, in step S4, the control modes include remote manual driving and partial automatic driving; The remote manual operation mode is used for routine lifting tasks, where the operator manually controls the tower crane through multi-channel high-definition video monitoring in the remote cab. The partial autonomous driving mode is used for tasks such as pre-set path hoisting, load swing suppression and restricted area protection, respectively achieving automatic cruising to the target coordinate point, real-time closed-loop control to eliminate swing, and autonomous obstacle avoidance in combination with position sensors; Mode switching requires manual authorization for activation or disabling via a remote control console; Emergency situations include tower crane overload, overspeed, and exceeding limits, which are addressed by using local PLC hard safety logic to achieve forced braking independent of the remote system. Issuing a hoisting instruction means that the intelligent scheduling subsystem sends the instruction to the tower crane control system, which then controls the execution of the hoisting task.

[0015] In the preferred embodiment, in step S5, the digital twin system constructs a virtual environment corresponding to the physical construction site at a 1:1 scale based on a game engine or professional visualization engine. A multi-source data dynamic mapping algorithm is used to achieve real-time synchronization between the physical and virtual scenes. This algorithm establishes a parameter mapping relationship between physical entities and virtual models, and drives the virtual model update based on real-time data. The specific formula is as follows: ; In the formula, for The set of state parameters for the virtual scene model at any time includes tower crane position, hook height, slewing angle, load, and component positioning status; for The set of physical scene entity state parameters at any given time consists of tower crane status and component position data collected by sensors; for A real-time environmental awareness data set, including wind speed and the location of construction workers; The parameter mapping coefficient matrix is ​​pre-calibrated based on the component attributes of the BIM model and the mechanical characteristics of the tower crane. This is a mapping function, through which physical parameters are mapped... Convert the virtual scene into state parameters that can be recognized by the virtual model to achieve a 1:1 dynamic alignment between the virtual scene and the physical scene; The digital twin system dynamically maps information including tower crane position, hook height, slewing angle, current lifting weight, and operating / idle status. It automatically identifies and updates the hoisting and positioning status of prefabricated components through RFID tags, and overlays environmental information such as real-time wind speed. The intelligent monitoring system uses a construction schedule deviation analysis algorithm to calculate and visualize the real-time construction progress of multiple prefabricated houses. It identifies critical path tasks by combining the construction logic relationships in the BIM model and monitors the deviation between the critical path tasks and the plan. The schedule deviation calculation model is as follows: ; ; In the formula, for Time of the first Construction progress deviation rate of the building; for Time of the first The actual number of completed components of a building is counted based on the components in place identified by RFID tags; for Time of the first The planned number of completed components for a building is extracted from the construction plan in the BIM model. for The maximum schedule deviation rate corresponding to the critical path task at any time is determined by comparing the schedule deviation rates of the four buildings to identify the critical path task, thus achieving precise monitoring of the critical path.

[0016] In the preferred embodiment, in step S5, the intelligent monitoring system uses a multi-threshold collaborative alarm algorithm to achieve safety early warning. This algorithm avoids false alarms from a single threshold by setting multi-dimensional alarm thresholds and establishing threshold collaborative judgment logic. The specific alarm triggering model is as follows: ; In the formula, for The alarm trigger result is displayed in real time, with 1 indicating that the alarm is triggered and 0 indicating that it is not triggered. for Real-time wind speed values ​​are collected by an anemometer. The wind speed threshold is set according to the tower crane safety operation specifications. for The actual lifting weight of the tower crane at any given time is calculated by the torque sensor. The overload threshold for the tower crane is set according to the rated lifting capacity of the tower crane. for Time of the first Construction progress deviation rate of the building. The time lag threshold is set backwards from the total project duration. When the alarm threshold is triggered, the system will notify the project's key management personnel via SMS. The closed-loop optimization employs a schedule deviation-driven rescheduling algorithm. When the actual construction progress deviates from the plan by more than a threshold, i.e. This triggers the intelligent scheduling subsystem to re-plan the scheduling. The task priority adjustment model after the re-scheduling is triggered is as follows: ; In the formula, For the first Priority of each pending hoisting task For the first The house number to which each task belongs. This represents the current progress deviation rate for the building. For the first The weight of the component corresponding to each task. The maximum weight of all components to be hoisted. , These are the weighting coefficients. When schedule deviation takes priority ; Based on the adjusted task priority, the intelligent scheduling subsystem re-executes the ant colony algorithm in step S3 to plan the optimal path. At the same time, it pushes the real-time attitude of the tower crane, the surrounding environment, and the location information of the components to be lifted from the digital twin system to the remote cab or the unmanned driving system to assist in control decisions.

[0017] In the preferred embodiment, in step S5, the digital archive records all operating instructions, changes in tower crane operating parameters, and changes in construction status for traceability of the construction process.

[0018] This invention provides an intelligent scheduling and remote control method for tower cranes. This method effectively solves the problems of high scheduling difficulty, low construction efficiency, harsh operating environment for operators, and high construction risks in the operation of tower cranes in clustered prefabricated buildings. In terms of scheduling, by proposing a tower crane scheduling optimization algorithm, the method intelligently plans the hoisting path and task allocation of tower cranes based on construction needs, solving the scheduling problem of tower cranes in simultaneous hoisting operations on multiple buildings, improving the utilization rate of tower cranes and resources, and thus improving construction efficiency. In terms of operation, by configuring various sensors to achieve remote control and partially unmanned operation, operators no longer need to work in harsh on-site environments, reducing their labor intensity and avoiding safety hazards caused by limited visibility during on-site operations, thereby improving operator comfort and construction safety.

[0019] Furthermore, by combining BIM models to develop digital twin systems and intelligent monitoring systems, construction progress can be synchronized in real time, providing data support for the automated hoisting and scheduling systems of tower cranes, promoting the intelligent development of prefabricated building construction, and providing feasible solutions for the construction of tower cranes in similar prefabricated buildings in the future, thus promoting the advancement of prefabricated building and tower crane operation technologies. Attached Figure Description

[0020] The present invention will be further described below with reference to the accompanying drawings and embodiments: Figure 1 This is a construction scene diagram of the present invention; Figure 2 This is a top-down view of the construction site layout of this invention; Figure 3 This is a system architecture diagram of the present invention; Figure 4 This is a diagram of the scheduling system architecture based on the ant colony algorithm of this invention; Figure 5 This is the algorithm flow of the scheduling system of the present invention.

[0021] In the diagram: Remote control room 1; House 2. Detailed Implementation

[0022] Example 1 like Figure 1-4 As shown, a method for intelligent scheduling and remote control of tower cranes includes: S1. Build a distributed system architecture including the physical layer, perception and control layer, data processing and decision-making layer, application layer and communication layer, and deploy the corresponding devices; S2. Collect multi-source data related to the environment, tower crane status, and BIM model, and transmit them to the data fusion processing center for fusion processing; S3. Based on the ant colony algorithm, the intelligent scheduling subsystem plans the optimal hoisting path and task sequence, and outputs the scheduling results. S4. Select the corresponding control mode, issue hoisting instructions to execute the task, and trigger forced braking in an emergency. S5. The construction scene is mapped through a digital twin system, the intelligent monitoring system monitors the status and issues alarms, and the feedback data achieves closed-loop optimization and records digital archives.

[0023] This intelligent scheduling and remote control method for tower cranes achieves intelligent management and control of tower cranes through a five-layer distributed system architecture and five core steps, as follows: Construction scenarios such as Figure 1 As shown, prefabricated houses A, B, C, and D are arranged in a triangular pattern, and tower crane 1 is installed at the symmetrical midpoint of the four houses.

[0024] S1 is designed to build a distributed system architecture and deploy corresponding equipment. The physical layer deploys intelligent tower cranes with closed-loop control systems, prefabricated houses, and precast components. The intelligent tower cranes are installed at the symmetrical midpoints of the prefabricated houses arranged in a triangular pattern. The perception and control layer is configured with environmental perception sensors, tower crane status perception sensors, and execution control terminals. The environmental perception sensors include high-precision LiDAR and wide-angle high-definition cameras. The tower crane status perception sensors include a rotary encoder, trolley position sensor, lifting height sensor, torque sensor, anemometer, tilt sensor, and vibration sensor. The execution control terminal includes the tower crane control system and a high-performance edge computing unit. The high-performance edge computing unit is installed on the tower crane body for data preprocessing and local obstacle avoidance calculations. The data processing and decision-making layer establishes an intelligent scheduling subsystem and a data fusion processing center. The application layer is configured with a remote operator's cab, digital twin, and intelligent monitoring system. The communication layer deploys industrial-grade switches and 5G base stations to ensure real-time data transmission and command issuance.

[0025] S2 collects and fuses multi-source data. Environmental data includes construction progress data detected by high-precision lidar, the status and location of prefabricated components, the status and location of construction personnel, the position and status of hooks, and the location of lifting points, all collected by wide-angle high-definition cameras. The wide-angle high-definition cameras are equipped with AI discrimination programs to identify the hook status. Tower crane status data includes the slewing angle collected by the slewing encoder, the trolley running position collected by the trolley position sensor, the hook lifting height collected by the lifting height sensor, the torque on the tower crane collected by the torque sensor, the wind speed and direction collected by the anemometer, the tower verticality collected by the tilt sensor, and the tower crane vibration parameters collected by the vibration sensor. BIM model-related data includes the dimensions, weight, installation location of prefabricated components, and construction progress data for each building. The collected data is transmitted to the data fusion processing center through the communication layer. After spatiotemporal alignment, filtering, and fusion, unified and accurate data is generated to support subsequent scheduling and control.

[0026] S3 is an ant colony algorithm-based system for planning optimal hoisting schemes. The intelligent scheduling subsystem aims to minimize the total operation time, abstracting the tower crane scheduling problem into a path optimization problem with time windows, multiple task points, and dynamic priorities. During algorithm initialization, the number of ants is set to 1.5 times the number of hoisting tasks, with pheromone factor α=1.0, heuristic factor β=2.0, pheromone evaporation rate ρ=0.5, pheromone constant Q=100, and a maximum iteration count of 200. A pheromone matrix of (number of tasks + 1) × (number of tasks + 1) with an initial value of 0.1 is created. Simultaneously, the transfer time between all task pairs is calculated, and this transfer time is derived by combining the tower crane's rotational angular velocity and luffing speed. During the iteration process, the ants start from the virtual starting point, which is the tower crane center, filter candidate tasks, calculate movement time and selection probability, select tasks using a roulette wheel and update the path. After completing all tasks, the total path time is calculated. After each iteration, the optimal path is compared and updated, pheromones are released proportionally, and the pheromones of the optimal path are enhanced. The process terminates when the maximum number of iterations is reached or the optimal path has not improved for 20 consecutive iterations. The optimal task sequence, time estimate, and performance indicators such as tower crane utilization rate and time saving rate are output.

[0027] S4 is used to select the control mode and execute the lifting task. Control modes include remote manual operation and partial automatic operation. Remote manual operation is used for routine lifting tasks, where the operator manually controls the tower crane through multi-channel high-definition video monitoring in the remote cab, which is equipped with a walkie-talkie system for direct communication with the site. Partial automatic operation is used for pre-set path lifting, load sway suppression, and restricted area protection tasks. It achieves automatic cruising to the target coordinates, real-time closed-loop control to eliminate sway, and autonomous obstacle avoidance using position sensors. Mode switching requires manual authorization. The intelligent scheduling subsystem sends lifting commands to the tower crane control system, and the tower crane executes the lifting task. In case of emergencies such as overload, overspeed, or boundary violations, the local PLC's hard safety logic independently initiates forced braking, ensuring construction safety.

[0028] The S5 achieves closed-loop optimization through digital twins and intelligent monitoring. The digital twin system, based on a game engine or professional visualization engine, constructs a virtual environment corresponding to the physical construction site at a 1:1 scale. It integrates detailed BIM models of four buildings, dynamically mapping tower crane positions, hook heights, slewing angles, lifting weights, operating / idle status, and the placement of prefabricated components. The placement of prefabricated components is updated via RFID tags, while real-time environmental information is overlaid. The intelligent monitoring system automatically calculates and visualizes the building construction progress based on component installation records, identifies critical path tasks and monitors progress deviations, and sets safety thresholds for overload, excessive wind speed, and progress delays. When a threshold is triggered, an alarm is automatically triggered, and management personnel are notified via SMS. Simultaneously, real-time construction progress and tower crane status data are fed back to the intelligent scheduling subsystem. If the actual progress deviates from the plan by more than a threshold, rescheduling is triggered, and all operations and status changes are recorded to form a traceable digital archive.

[0029] 1. System Deployment Phase: First, the physical layer is built by installing intelligent tower cranes with closed-loop control systems at the symmetrical midpoints of the prefabricated houses arranged in a triangular pattern, and placing prefabricated components. Next, in the perception and control layer, high-precision LiDAR and wide-angle HD cameras are installed in the construction site, and state perception sensors such as rotary encoders and trolley position sensors, along with high-performance edge computing units, are installed on the tower crane body to deploy the tower crane control system. Then, the data processing and decision-making layer is built, deploying an intelligent scheduling subsystem and a data fusion processing center. At the application layer, a remote driver's cab with multi-channel HD video surveillance, three driving modes, and a walkie-talkie system is configured to build a digital twin and intelligent monitoring system. Finally, at the communication layer, industrial-grade switches and 5G base stations are deployed to complete the distributed system architecture and equipment deployment.

[0030] 2. Data Acquisition and Fusion Phase: All sensors in the perception and control layer are activated. High-precision lidar detects the construction progress of the building, wide-angle high-definition cameras collect data related to components, personnel, and hooks, and AI identifies the hook status. Tower crane status perception sensors collect parameters such as slewing angle and trolley position. The collected data is transmitted to the data fusion processing center via 5G base stations and industrial-grade switches. Simultaneously, component dimensions, weights, installation locations, and construction progress data from the BIM model are imported. The data fusion processing center performs spatiotemporal alignment, filtering, and fusion of multi-source data to generate unified and accurate data.

[0031] 3. Scheduling and Planning Phase: The intelligent scheduling subsystem calls the ant colony algorithm, inputs the fused BIM component information, tower crane real-time status, and the location of the component to be lifted, initializes the algorithm parameters, including the number of ants, pheromone factor, and evaporation rate, and starts iterative calculation. Candidate tasks are screened by ants, the selection probability is calculated, and the path is constructed. After iteration, the optimal lifting task sequence and path are determined, and the scheduling results, including task order, starting position, target position, time estimate, and performance indicators, are output.

[0032] 4. Lifting Execution Phase: The operator selects the control mode from the remote control cab based on the construction task type. For routine lifting, remote manual operation is selected, allowing manual control of the tower crane via multi-channel high-definition video monitoring. For preset path lifting, load sway suppression, and restricted area protection, partial automatic driving is selected, and the system automatically executes the corresponding operations. Mode switching requires manual authorization via the remote control console. The intelligent scheduling subsystem sends the optimal lifting instructions to the tower crane control system, and the tower crane executes the picking, lifting, and installation operations according to the instructions. During construction, in case of emergencies such as overload, overspeed, or boundary violations, the local PLC's hard safety logic automatically initiates forced braking, stopping the tower crane's operation.

[0033] 5. Monitoring and Optimization Phase: The digital twin system receives fused data in real time, dynamically mapping the tower crane's operating status, component placement, and environmental information within a virtual environment. Operators can monitor the construction scenario in real time through this virtual environment. The intelligent monitoring system monitors construction progress and tower crane operating parameters in real time, displaying the building construction progress and critical path task deviations. When overload, excessive wind speed, or progress delays exceeding thresholds are detected, alarms are automatically triggered and project management personnel are notified via SMS. If the actual construction progress deviates from the plan by more than the threshold, the intelligent scheduling subsystem re-executes the ant colony algorithm based on feedback data, adjusting the hoisting task sequence and path to achieve closed-loop optimization. Simultaneously, the system records all operating instructions, changes in tower crane operating parameters, and changes in construction status throughout the process, forming a digital archive for easy subsequent traceability.

[0034] This method effectively solves the problems of high difficulty in tower crane scheduling, low construction efficiency, harsh operating environment for operators, and high construction risks in clustered prefabricated buildings. In terms of scheduling, the ant colony algorithm is used to intelligently plan tower crane hoisting paths and task allocation, eliminating reliance on manual experience and effectively solving the scheduling problem of simultaneous hoisting of multiple buildings, improving tower crane utilization and resource utilization, and accelerating construction progress. In terms of operation, remote control and partial unmanned operation are achieved by configuring various sensors, eliminating the need for operators to work in harsh on-site environments, reducing labor intensity, and avoiding safety hazards caused by limited visibility during on-site operations, thus improving operator comfort and construction safety. In terms of intelligence, the digital twin system and intelligent monitoring system developed in conjunction with the BIM model can synchronize construction progress in real time, providing accurate data support for the automated tower crane hoisting and scheduling system, promoting the intelligentization of prefabricated building construction, and providing feasible technical ideas for subsequent tower crane construction in similar prefabricated buildings, thus promoting the development of prefabricated building and tower crane operating technology.

[0035] Example 2 To further illustrate with reference to Example 1, in step S1, a smart tower crane with a closed-loop control system, prefabricated houses and prefabricated components are deployed at the physical layer. The smart tower crane is installed around the prefabricated houses arranged in a triangular pattern. The perception and control layer is equipped with environmental perception sensors, tower crane status perception sensors, and execution control terminals. The environmental perception sensors include high-precision lidar and wide-angle high-definition cameras. The wide-angle high-definition cameras are equipped with AI discrimination programs to identify the status of the hook. The tower crane status perception sensors include a rotary encoder, a trolley position sensor, a lifting height sensor, a torque sensor, an anemometer, a tilt sensor, and a vibration sensor. The execution control terminal includes the tower crane control system and a high-performance edge computing unit. The edge computing unit is installed on the tower crane body and is used for data preprocessing and local obstacle avoidance calculation. The data processing decision-making layer establishes an intelligent scheduling subsystem and a data fusion processing center; The application layer is configured with a remote driver's cab, digital twin, and intelligent monitoring system; The communication layer is equipped with industrial-grade switches and 5G base stations.

[0036] In the preferred embodiment, the data preprocessing and local obstacle avoidance of the edge computing unit include: S101, Data preprocessing step: The steps for the edge computing unit to receive multi-source data collected by sensors in the perception and control layer include: S1011. Data Filtering: The edge computing unit filters out obstacle data from the construction scene collected by high-precision lidar and wide-angle high-definition cameras in the environmental perception sensors. This obstacle data includes building corners, temporary facilities, and other components. Simultaneously, it filters out real-time tower crane operation data collected by the tower crane status perception sensors, including the slewing encoder, trolley position sensor, lifting height sensor, torque sensor, anemometer, tilt sensor, and vibration sensor. This operation data includes slewing angle, trolley position, hook height, torque, wind speed, tower verticality, and vibration parameters. Invalid data is removed; invalid data refers to abnormal values ​​exceeding the normal range caused by sensor malfunctions. S1012. Data spatiotemporal alignment: Based on the time synchronization signal of the 5G base station in the communication layer, a timestamp is added to each type of filtered data. Simultaneously, the 3D coordinate data of obstacles collected by LiDAR and the 2D image obstacle data collected by the camera are uniformly converted into a polar coordinate system with the tower crane's rotation center as the origin. The polar coordinate system is established with the tower crane's rotation center as the origin and the direction pointing towards the prefabricated house A as the positive polar axis. The 3D coordinates (x, y, z) of obstacles collected by LiDAR are converted into polar coordinates. The formula is , , , where r is the polar radius, θ is the polar angle and its range is [0, 2π), and h is the height, to achieve spatial coordinate alignment of multi-source data and obtain standardized data that is consistent in time and space; S1013. Data Denoising and Compression: A moving average filtering algorithm is used to denoise the standardized tower crane vibration parameters and wind speed data. The filtering formula is as follows: ,in Let N be the parameter value after noise reduction at time t, and N be the sliding window length. The sliding window length N is dynamically adjusted according to the sensor sampling frequency. When, N is 10; when When, N is 8; when When N is 5, for Original parameter values ​​at time. The sampling interval for the sensor is defined; the point cloud data of the lidar is compressed using a voxel grid downsampling algorithm to retain the key contour features of obstacles, reduce data transmission and computational load, and obtain the preprocessed effective data. S102, Local obstacle avoidance calculation steps: Based on the preprocessed valid data, the edge computing unit performs the following local obstacle avoidance steps for the tower crane: S1021. Obstacle Modeling: Based on the preprocessed polar coordinate data from high-precision LiDAR, construct a polar coordinate model of static obstacles in the construction scene. Static obstacles include the four corners of buildings and fixed temporary facilities. Each obstacle is represented as a set of regions in polar coordinates. ,in The extreme diameter from the obstacle to the tower crane's slewing center. The polar angle of the obstacle. , The range of the obstacle's extreme diameter. , The range of the polar angle of the obstacle; Combining the real-time location of dynamic obstacles identified by the AI ​​discrimination program of the wide-angle high-definition camera, including components to be hoisted and construction personnel, the AI ​​program identifies the type of dynamic obstacle through a target detection algorithm and calculates the extreme radius r from the dynamic obstacle to the tower crane's slewing center based on a preset distance calibration model within the camera. The camera's intrinsic parameters include focal length f and pixel size d. The calculation formula is as follows: Where H is the actual height of the dynamic obstacle, the component height is known, and the personnel height is set by default as 1.3-1.8 meters. To obtain the pixel height of the dynamic obstacle in the camera image, the polar coordinate information of the dynamic obstacle is added to the obstacle model, thus updating the real-time obstacle set. ; S1022. Tower crane motion trajectory prediction: Based on real-time data preprocessed by the tower crane status sensing sensors, the current slewing angle of the tower crane is obtained. The current position of the trolley is extremely far. Current height of the hook Based on the next control commands issued by the tower crane control system, the control commands include the slewing angular velocity ω, the trolley luffing speed v, and the hook lifting speed. Predicting the future The trajectory of the tower crane hook within a given time period. The value is set between 0.5 and 2 seconds, depending on the tower crane's operating speed. The prediction formula for the trajectory in polar coordinates is as follows: , , ,in , , They are respectively The hook's predicted polar diameter, polar angle, and height are constantly being measured. S1023 Collision detection: Determines the relationship between the predicted trajectory and the real-time obstacle set. Whether there is a collision risk, the collision determination criteria are: if there is , making and Within the obstacle height range, the obstacle height range is obtained from LiDAR data and represented as... That is, satisfying If so, it is determined that there is a risk of collision; S1024. Obstacle avoidance command generation: If a collision risk is determined, the edge computing unit calculates the obstacle avoidance adjustment amount based on the collision detection results and generates a local obstacle avoidance control command. If the collision risk originates from the turning direction, then adjust the turning angular velocity using the following formula: ,in To avoid obstacles, the safe polar angle, the safe polar angle The method for determining is: if Within the range of the polar angle of the obstacle inside, when hour, ;otherwise , To ensure safety and redundancy, the angle is set between 5 and 10 degrees, depending on the complexity of the construction scenario. The sign function ensures that the tower crane rotates in the direction away from the obstacle; If the collision risk originates from the direction of the car's change of direction, then adjust the car's change of speed using the following formula: ,in To avoid obstacles, the safe extreme diameter. Take the nearest polar diameter outside the range of the obstacle's polar diameter; If the collision risk originates from the vertical direction, adjust the hook lifting speed using the following formula: ,in Safe height to avoid obstacles Take the nearest height outside the range of obstacle heights; The adjusted , , As a local obstacle avoidance control command, it is sent to the tower crane control system to realize local obstacle avoidance of the tower crane, and at the same time, the obstacle avoidance process data is uploaded to the data fusion processing center.

[0037] Step S1 of this intelligent scheduling and remote control method for tower cranes revolves around the construction of a distributed system architecture and equipment deployment. It also includes optimized solutions for edge computing unit data preprocessing and local obstacle avoidance, as detailed below: In the S1 system architecture and equipment deployment, the physical layer deploys intelligent tower cranes with closed-loop control systems, prefabricated houses, and precast components. The intelligent tower cranes are installed around the prefabricated houses arranged in a triangular pattern. The perception and control layer is configured with environmental perception sensors, tower crane status perception sensors, and execution control terminals. The environmental perception sensors include high-precision LiDAR and wide-angle high-definition cameras. The wide-angle high-definition cameras are equipped with AI discrimination programs to identify the hook status. The tower crane status perception sensors include a rotary encoder, trolley position sensor, lifting height sensor, torque sensor, anemometer, tilt sensor, and vibration sensor. The execution control terminal includes the tower crane control system and a high-performance edge computing unit. The high-performance edge computing unit is installed on the tower crane body for data preprocessing and local obstacle avoidance calculations. The data processing and decision-making layer builds an intelligent scheduling subsystem and a data fusion processing center. The application layer is configured with a remote operator's cab, digital twin, and intelligent monitoring system. The communication layer deploys industrial-grade switches and 5G base stations.

[0038] In the preferred scheme, the data preprocessing and local obstacle avoidance process of the edge computing unit is divided into S101 data preprocessing step and S102 local obstacle avoidance calculation step.

[0039] In the S101 data preprocessing step, the edge computing unit receives multi-source data collected by sensors in the perception and control layer. First, it performs S1011 data filtering. The edge computing unit filters out obstacle data of the construction scene collected by high-precision lidar and wide-angle high-definition camera in the environmental perception sensor. This obstacle data includes building corners, temporary facilities, and other components. At the same time, it filters out the real-time operation data of the tower crane collected by the tower crane status perception sensor, including the slewing encoder, trolley position sensor, lifting height sensor, torque sensor, anemometer, tilt sensor, and vibration sensor. This operation data includes slewing angle, trolley position, hook height, torque, wind speed, tower verticality, and vibration parameters. Invalid data is removed. Invalid data is abnormal values ​​that exceed the normal range caused by sensor malfunctions. Next, S1012 data spatiotemporal alignment is performed. Based on the time synchronization signal of the 5G base station in the communication layer, a timestamp is added to each type of filtered data. At the same time, the obstacle 3D coordinate data collected by the lidar and the obstacle image data collected by the camera are uniformly converted into a polar coordinate system with the tower crane's rotation center as the origin. The polar coordinate system is established with the tower crane's rotation center as the origin and the direction pointing to the prefabricated house A as the positive polar axis. The obstacle 3D coordinates (x, y, z) collected by the lidar are converted into polar coordinates. The formula is , , Where r represents the polar radius, θ represents the polar angle (ranging from [0, 2π)), and h represents the height, this transformation aligns the spatial coordinates of multi-source data, resulting in standardized data with spatiotemporal consistency. Then, S1013 data denoising and compression are performed. A moving average filtering algorithm is used to denoise the standardized tower crane vibration parameters and wind speed data. The filtering formula is: ,in Represents the parameter value after noise reduction at time t, and N represents the sliding window length. The sliding window length N is dynamically adjusted according to the sensor sampling frequency. When, N is 10; when When, N is 8; when When N is 5, The original parameter values ​​at time t - kΔt are represented, and Δt represents the sensor sampling interval. The lidar point cloud data is compressed using a voxel grid downsampling algorithm to retain the key contour features of obstacles, reduce data transmission and computational load, and obtain the preprocessed effective data.

[0040] In the S102 local obstacle avoidance calculation step, the edge computing unit realizes local obstacle avoidance of the tower crane based on the preprocessed effective data. First, obstacle modeling is performed in S1021. Based on the polar coordinate data preprocessed by high-precision lidar, a polar coordinate model of static obstacles in the construction scene is constructed. Static obstacles include the four corners of the building and fixed temporary facilities. Each obstacle is represented as a set of regions in polar coordinates. ,in The extreme diameter representing the distance from the obstacle to the center of rotation of the tower crane. Represents the polar angle of the obstacle. , Represents the range of the obstacle's polar diameter. , This represents the range of the obstacle's polar angle. Combined with the real-time location of dynamic obstacles identified by the AI ​​discrimination program of the wide-angle high-definition camera, these dynamic obstacles include components to be hoisted and construction personnel. The AI ​​program identifies the type of dynamic obstacle through a target detection algorithm and calculates the polar radius *r* from the dynamic obstacle to the tower crane's slewing center based on a preset distance calibration model within the camera. The camera's intrinsic parameters include focal length *f* and pixel size *d*. The calculation formula is as follows: Where H represents the actual height of the dynamic obstacle, the component height is known, and the personnel height is set to the default range of 1.3-1.8 meters. Representing the pixel height of dynamic obstacles in the camera image, the polar coordinate information of the dynamic obstacles is added to the obstacle model to update the real-time obstacle set. Next, the motion trajectory prediction of the S1022 tower crane is performed. Based on the real-time data preprocessed by the tower crane's status perception sensor, the current slewing angle of the tower crane is obtained. The current position of the trolley is extremely far. Current height of the hook Based on the next control commands issued by the tower crane control system, the control commands include the slewing angular velocity ω, the trolley luffing speed v, and the hook lifting speed. Predicting the future The trajectory of the tower crane hook within a given time period. The value is set between 0.5 and 2 seconds, depending on the tower crane's operating speed. The prediction formula for the trajectory in polar coordinates is as follows: , , ,in , , These represent the predicted polar diameter, predicted polar angle, and predicted height of the hook at time t + Δt, respectively. Then, collision detection (S1023) is performed to determine the predicted trajectory's alignment with the real-time obstacle set. Whether there is a collision risk, the collision determination criteria are: if there is , making and Within the obstacle height range, the obstacle height range is obtained from LiDAR data and represented as... That is, satisfying If so, it is determined that there is a risk of collision.

[0041] Finally, the S1024 obstacle avoidance command is generated. If a collision risk is determined, the edge computing unit calculates the obstacle avoidance adjustment amount based on the collision detection results and generates a local obstacle avoidance control command: if the collision risk originates from the turning direction, the turning angular velocity is adjusted, and the adjustment formula is as follows: ,in Represents the safe polar angle for avoiding obstacles. The method for determining is: if Within the range of the polar angle of the obstacle inside, when hour, ;otherwise , This represents a safety redundancy angle and ranges from 5 to 10 degrees, depending on the complexity of the construction scenario. The sign function ensures the tower crane rotates away from obstacles; if the collision risk originates from the trolley's luffing direction, the trolley's luffing speed is adjusted using the following formula: ,in Represents the safe radius for avoiding obstacles. Take the nearest extreme diameter outside the range of the obstacle's extreme diameter; if the collision risk originates in the height direction, adjust the hook lifting speed using the following formula: ,in This represents the safe height for avoiding obstacles. Take the nearest height outside the obstacle's height range; adjust the... , , As a local obstacle avoidance control command, it is sent to the tower crane control system to realize local obstacle avoidance of the tower crane, and at the same time, the obstacle avoidance process data is uploaded to the data fusion processing center.

[0042] In actual deployment, the first step is hardware deployment. First, determine the installation locations of the intelligent tower cranes around the prefabricated houses arranged in a triangular pattern. Install the intelligent tower cranes with closed-loop control systems and place prefabricated components. Install high-precision LiDAR and wide-angle HD cameras in key areas of the construction site. Install rotary encoders, trolley position sensors, lifting height sensors, torque sensors, anemometers, tilt sensors, and vibration sensors at corresponding locations on the tower crane body. Install high-performance edge computing units on the tower crane body and connect them to the various sensors to deploy the tower crane control system. In the control center, build a data processing and decision-making layer, deploying intelligent scheduling subsystem servers and data fusion processing center servers. At the application layer, configure a remote operator's cab equipped with control devices, display devices, and walkie-talkies to build a hardware platform for a digital twin and intelligent monitoring system. At the communication layer, deploy industrial-grade switches and cover the construction area with 5G base stations, completing the connection and debugging of all hardware devices.

[0043] The second step involves software and module deployment. The edge computing unit runs on a Linux operating system and deploys a data preprocessing module and a local obstacle avoidance calculation module written in C++. The data preprocessing module integrates data filtering, spatiotemporal alignment, noise reduction, and compression functions. The spatiotemporal alignment function implements the conversion from three-dimensional coordinates to polar coordinates through programming, and the conversion process calls mathematical libraries for calculations. , , The noise reduction function is implemented by writing code for a moving average filtering algorithm. The computation and compression function integrates a voxel grid downsampling algorithm library; the local obstacle avoidance calculation module integrates obstacle modeling, trajectory prediction, collision detection, and obstacle avoidance command generation functions, with trajectory prediction implemented through code. , , The calculation and obstacle avoidance command generation function is implemented through programming. , , Calculation and safety polar angle The data fusion processing center deploys a data streaming module based on Apache Kafka and a data fusion processing module based on Spark Streaming to achieve the reception, spatiotemporal alignment, filtering, and fusion of multi-source data. The intelligent scheduling subsystem is developed using Python, employs the Scikit-learn library to implement core algorithms, is deployed on a server, and establishes a data interaction interface with the data fusion processing center. The digital twin and intelligent monitoring system uses Unity or Unreal Engine to build a virtual environment, integrates a BIM model import module, and develops a dynamic mapping module to achieve synchronization with the physical scene; the intelligent monitoring system uses the Qt framework to develop a visual interface and integrates an alarm logic module. The remote driver's cab deploys control software developed in C# to achieve driving mode switching, command sending, and video monitoring functions, and is equipped with walkie-talkie communication software.

[0044] The third step involves system integration and trial operation. This includes starting all hardware and software modules, testing the edge computing unit's data preprocessing function, and checking the accuracy of data filtering, spatiotemporal alignment, and noise reduction and compression effects. It also involves testing the local obstacle avoidance calculation function, simulating obstacle scenarios, and verifying the accuracy of obstacle modeling, the rationality of trajectory prediction, collision detection sensitivity, and the effectiveness of obstacle avoidance commands. Furthermore, it involves testing the data reception and fusion capabilities of the data fusion processing center to ensure real-time and accurate data transmission. The step also includes testing the data interaction between the intelligent scheduling subsystem and other modules to verify the generation and issuance of scheduling commands. Additionally, it involves testing the dynamic mapping effect of the digital twin system and the monitoring and alarm functions of the intelligent monitoring system. Finally, it involves testing the control functions and communication effects of the remote control cab. After successful integration, trial operation is conducted, continuously monitoring the system's operating status, recording operational data, and optimizing and adjusting any problems encountered to ensure stable system operation.

[0045] In the preferred embodiment, in step S2, the environmental data includes building construction progress data detected by high-precision lidar, as well as data on the status and location of prefabricated components, the status and location of construction personnel, the position and status of hooks, and the location of lifting points collected by wide-angle high-definition cameras; the tower crane status data includes slewing angle, trolley running position, hook lifting height, tower crane torque, wind speed and direction, tower verticality, and tower crane vibration parameters. BIM model data includes the dimensions, weight, installation location of prefabricated components, and construction progress data for each building; Fusion processing includes spatiotemporal alignment, filtering, and fusion of multi-source data to generate unified and accurate data.

[0046] First, multi-source data is collected, categorized into three types: environmental data, tower crane status data, and BIM model-related data. Environmental data originates from environmental sensing sensors in the perception and control layer. High-precision LiDAR is responsible for detecting construction progress data, while wide-angle high-definition cameras collect data on the status and location of prefabricated components, construction personnel, hook positions and status, and lifting point positions. Furthermore, the wide-angle high-definition cameras are equipped with AI-powered discrimination programs that can automatically identify the hook's hooking status.

[0047] The tower crane status data comes from tower crane status sensing sensors, specifically including the slewing angle collected by the slewing encoder, the trolley running position collected by the trolley position sensor, the hook lifting height collected by the lifting height sensor, the torque on the tower crane collected by the torque sensor, the wind speed and direction collected by the anemometer, the tower verticality collected by the tilt sensor, and the tower crane vibration parameters collected by the vibration sensor. BIM model-related data is extracted from the BIM model of the prefabricated building, covering the dimensions, weight, and installation location of prefabricated components, as well as the construction progress data for each building.

[0048] After data collection is completed, all collected multi-source data are transmitted to the data fusion processing center of the data processing decision layer through industrial-grade switches and 5G base stations at the communication layer.

[0049] Subsequently, the data fusion processing center performs fusion processing on the multi-source data. The processing includes three key stages: spatiotemporal alignment, filtering, and fusion. The spatiotemporal alignment stage uses the time synchronization signal provided by the 5G base station at the communication layer to add a unified timestamp to each type of data. At the same time, it converts data from different spatial coordinate systems, such as the three-dimensional coordinate data collected by LiDAR and the two-dimensional image data collected by cameras, into a polar coordinate system with the tower crane's slewing center as the origin, achieving consistency of data in both time and space. The filtering stage removes outliers caused by sensor malfunctions, such as slewing angle data that exceeds the normal operating range of the tower crane or component position data that deviates significantly from the actual position, ensuring data validity. The fusion stage uses data fusion algorithms to integrate the spatiotemporally aligned and filtered environmental data, tower crane status data, and BIM model-related data, eliminating data redundancy and conflicts, and ultimately generating unified and accurate data. This provides reliable data support for subsequent intelligent scheduling subsystem path planning, remote cab control decisions, and dynamic mapping of the digital twin system.

[0050] Data acquisition preparation: First, check whether the environmental perception sensors and tower crane status perception sensors in the perception control layer are operating normally, and ensure that all sensors such as high-precision lidar, wide-angle high-definition camera, rotary encoder, and trolley position sensor are firmly installed and properly connected; at the same time, import the BIM model of the prefabricated building in advance in the data fusion processing center, complete the interface adaptation between the BIM model and the data fusion system, and ensure that the relevant data of the BIM model can be successfully extracted.

[0051] Data acquisition is initiated: All sensors in the perception and control layer are activated through the system control platform. The high-precision lidar begins to continuously scan the construction area, detecting the progress of building construction in real time and recording data. The wide-angle high-definition camera starts real-time shooting, simultaneously collecting the status and position data of prefabricated components, construction personnel, hooks, and lifting points. The AI ​​discrimination program runs automatically to identify and mark the hook status. The tower crane status perception sensor collects various parameters such as the slewing angle, trolley position, and hook height during the tower crane's operation in real time. All data collected by the sensors is transmitted to the data fusion processing center in real time through 5G base stations and industrial-grade switches.

[0052] Data Fusion Processing: After receiving multi-source data, the data fusion processing center first performs spatiotemporal alignment, adding a timestamp to each data point based on the time synchronization signal of the 5G base station. Then, it converts the 3D coordinates of the LiDAR and the 2D image coordinates of the camera into polar coordinates with the tower crane's slewing center as the origin, achieving uniformity of different data in the spatiotemporal dimension. Next, data filtering is performed, filtering out abnormal data caused by sensor malfunctions and removing them by using preset normal range thresholds. For example, angle data exceeding the tower crane's rated slewing angle range and extreme wind speed data exceeding construction safety standards collected by the anemometer are identified as outliers and removed. Finally, data fusion is performed, using a data fusion algorithm to integrate the filtered environmental data, tower crane status data, and BIM model-related data. For example, the installation position data of prefabricated components in the BIM model is compared and calibrated with the actual position data of the components collected by the camera to eliminate data deviations and generate a unified and accurate dataset.

[0053] Data Application and Feedback: After generating unified and accurate data, the data fusion processing center pushes it to the intelligent scheduling subsystem, the remote operator's cab, and the digital twin system. The intelligent scheduling subsystem plans the optimal hoisting path and task sequence based on this data. The remote operator's cab uses the data to monitor the construction scene and tower crane status in real time to execute operations. The digital twin system uses the data to dynamically map the physical scene. Simultaneously, during subsequent construction, the above data acquisition and fusion process is continuously repeated to ensure real-time data updates and provide continuous dynamic data support for each system. When significant fluctuations or anomalies occur in the data, the system triggers self-checks to investigate sensor or transmission link problems, ensuring the stability of data acquisition and fusion.

[0054] The multi-source data acquisition and fusion processing in step S2 provides the core data foundation for intelligent scheduling and remote control of tower cranes, and its beneficial effects are mainly reflected in three aspects: First, it ensures the comprehensiveness and accuracy of data. By collecting environmental, tower crane status, and BIM model-related data in a categorized manner, it covers all dimensions of information such as construction scenarios, equipment operation, and component information, avoiding decision-making biases caused by missing data. At the same time, through spatiotemporal alignment, filtering, and fusion processing, it eliminates data redundancy and outliers, ensuring the uniformity and accuracy of the output data. This provides a reliable basis for subsequent intelligent scheduling, remote control, and other processes, reducing hoisting path planning errors or control risks caused by data errors.

[0055] Secondly, it improves construction efficiency and collaboration. Real-time data is transmitted at high speed through 5G base stations, and combined with the rapid response of integrated processing, it ensures that the data can support the operation of various systems in a timely manner. For example, the intelligent scheduling subsystem can dynamically adjust the hoisting sequence based on real-time updated component positions and tower crane status data to avoid tower crane waiting due to data lag. At the same time, unified and accurate data enables data sharing between the intelligent scheduling subsystem, remote cab, and digital twin system, breaking down data barriers between various links and improving the overall collaborative operation efficiency of the system.

[0056] Third, it enhances construction safety. By collecting data on the status and location of construction personnel through wide-angle high-definition cameras and combining it with parameters such as torque and wind speed from the tower crane status data, it is possible to determine in real time whether construction personnel are in the dangerous area of ​​tower crane operation and whether the tower crane is in a safe operating state. For example, when the wind speed data exceeds the safety threshold, the data fusion processing center can quickly feed this information back to the monitoring system, trigger an alarm and prompt the suspension of work, thereby reducing construction safety hazards. At the same time, accurate hook position and status data also provide data support to avoid risks such as hook misoperation and component falling.

[0057] Example 3 In step S3, the scheduling objective of the ant colony algorithm is to minimize the total job time, and its objective function model is:

[0058] In the formula, For the task sequence, Starting from, The total number of components is equivalent to the total number of tasks. To the task To the mission The transfer time For the task Fixed working hours, including pickup and installation; The initialization of the ant colony algorithm includes: setting the number of ants to 1-4 times the number of hoisting tasks, and setting the pheromone factor. heuristic factor pheromone evaporation rate pheromone constant Maximum number of iterations Next, create a pheromone matrix of (task number + 1) × (task number + 1), with the initial value set to 0.1; From the task To the mission Transfer time The calculation model is as follows: ; In the formula, For the task The polar angle of the pickup location, For the task Polar angle of the installation location The tower crane's rotational angular velocity (degrees per second). For the task The extreme diameter of the pickup location, For the task Extreme diameter of the installation location, This refers to the tower crane's luffing speed.

[0059] In the preferred scheme, the state transition rule of the ant colony algorithm in step S3 is as follows: ; In the formula, For the first In the nth iteration Only ants from the task Transfer to task The probability, For the first Task during the next iteration To the mission The concentration of pheromones along the path, For the first Task during the next iteration To the mission Heuristic information ( ), For the first The set of tasks that the ant has not yet visited; The process of planning the optimal hoisting path and task sequence includes: Set a virtual starting point, i.e., the tower crane center, for each ant; clear the list of visited tasks and the task sequence; filter out candidate tasks whose predecessor tasks have not been completed from the unvisited tasks; if there are no available tasks, select the task with the fewest missing predecessor tasks and record the number of times; calculate the movement time from the current position to the pick-up point of each candidate task, i.e., take the larger value of the rotation time and the amplitude change time. Based on the above state transition rules, the task selection probability is calculated, the task is selected by roulette wheel and added to the path sequence, the task is marked as visited and the current position is updated to the task installation point. The above process is repeated until all tasks have been visited. Calculate the total path time for each ant. The total path time is the sum of the movement time between tasks and the fixed operation time of each task.

[0060] In the preferred scheme, step S3, the pheromone update mechanism of the ant colony algorithm includes a global update, the specific formula of which is: ; ; ; In the formula, For the first Task during the next iteration To the mission The concentration of pheromones along the path, For path pheromone increment, The number of ants. For the first Only ants on the path The increase in pheromones left on the surface For the first The total path length after an ant completes the entire path, which is the total operation time; Before outputting the scheduling results, the process also includes: comparing the total path time of all ants in this round to determine the optimal path for this round; if the optimal path for this round is better than the historical optimal path, then updating the global optimal path; and performing pheromone evaporation and incremental updates on the pheromone of all paths according to the pheromone update mechanism described above. The planning termination condition is reaching the maximum number of iterations. The second or optimal path shows no improvement after 20-40 consecutive attempts; the scheduling results include the optimal task sequence, total time estimate and performance indicators, including tower crane utilization, time saving rate and constraint satisfaction statistics.

[0061] In step S3 of the tower crane intelligent scheduling and remote control method, the core is to plan the optimal lifting path and task sequence of the tower crane based on the ant colony algorithm. The whole process revolves around setting the scheduling target, algorithm initialization, key model construction, path planning execution, and result output, as detailed below: The scheduling objective of the ant colony algorithm is to minimize the total job time, and its objective function model is: ,in This represents the total operation time. Represents a task sequence. Represents the starting point. This represents the total number of components, which is also the total number of tasks. Representative from the task To the mission The transfer time Representative task The fixed working time includes the time for both picking up the parts and installing them.

[0062] The initialization phase of the ant colony algorithm requires parameter setting and pheromone matrix creation: the number of ants is set to 1-4 times the number of hoisting tasks, and the pheromone factor... Set to 1.0, pheromone factor Heuristic factors are used to influence the weight of path accumulation on the path selection of ants. Set to 2.0, heuristic factor. The pheromone evaporation rate is used to influence the weight of prior knowledge on the ant's path selection. Set to 0.5, pheromone evaporation rate Pheromones are used to prevent premature convergence of algorithms. Set to 100, pheromone constant Used to control the intensity of pheromone updates and the maximum number of iterations. Set to 200 times; simultaneously create a (task number + 1) × (task number + 1) pheromone matrix, with an initial value of 0.1. The pheromone matrix is ​​used to record the pheromone concentration of the paths between each task.

[0063] From the task To the mission Transfer time Calculated using a specific model, the formula is: ,in Representative task The polar angle of the pickup location, Representative task Polar angle of the installation location This represents the tower crane's angular velocity, measured in degrees per second. Representative task The extreme diameter of the pickup location, Representative task Extreme diameter of the installation location, Representing the tower crane's luffing speed, this formula compares the tower crane's luffing speed from the task... Installation point to task The larger of the rotation time and amplitude change time at the pick-up point is taken as the transfer time. .

[0064] In the preferred scheme, the state transition rule of the ant colony algorithm is used to determine the probability of an ant choosing a task, and the formula is: ; ,in Representing the In the nth iteration Only ants from the task Transfer to task The probability, Representing the Task during the next iteration To the mission The concentration of pheromones along the path, Representing the Task during the next iteration To the mission The inspirational information, and , Representing the The set of tasks that the ant has not yet visited, when belong When, calculate the probability according to the formula, when Not belonging to When the probability is 0, the probability is 0.

[0065] The process of planning the optimal hoisting path and task sequence must be executed according to a fixed procedure: Set a virtual starting point for each ant (the tower crane's center), and simultaneously clear the ant's visited task list and task sequence; filter candidate tasks from unvisited tasks, excluding tasks with incomplete prerequisite tasks; if no selectable task exists, select the task with the fewest missing prerequisite tasks and record the number of times this occurs; calculate the travel time from the current position to the pick-up point of each candidate task, taking the larger value of the rotation time and the amplitude change time; calculate the selection probability of each candidate task based on the above state transition rules, select the task using a roulette wheel method and add it to the path sequence, mark the task as visited, and update the ant's current position to the task's installation point; repeat the above process until all tasks have been visited; calculate the total path time for each ant, which is the sum of the travel time between all tasks in the path and the fixed operation time of each task.

[0066] In the preferred scheme, the pheromone update mechanism of the ant colony algorithm includes a global update, which is implemented through three formulas: the first formula ,in Representing the Task during the next iteration To the mission The concentration of pheromones along the path, Representative path Pheromones increment; the second formula ,in Represents the number of ants. Representing the Only ants on the path The pheromone increment left on the surface; the third formula ; ,in Representing the The total path length after an ant completes the entire path, which is also the total operation time, when the ant... Only one ant passed through the path At that time, according to Calculate the pheromone increment; if it has not been passed, the increment is 0.

[0067] Before outputting the scheduling results, optimal path updates and pheromone adjustments must be completed: Compare the total path time of all ants in this round to determine the optimal path for this round. If the total time of the optimal path in this round is better than the total time of the historical optimal path, then update the global optimal path. According to the above pheromone update mechanism, the pheromone of all paths is first evaporated proportionally $\rho$, and then the path pheromone is incrementally updated. The planning termination condition is reaching the maximum number of iterations or the optimal path has no continuous improvement. The maximum number of iterations ranges from 200 to 500 times, and the number of times the optimal path has no continuous improvement ranges from 20 to 40 times. The scheduling results include three parts: the optimal task sequence, the total time estimate, and performance indicators. The performance indicators include tower crane utilization rate, time saving rate, and constraint satisfaction statistics.

[0068] In actual deployment, the first step is to prepare algorithm parameters and task data. The basic parameters of the ant colony algorithm are preset in the intelligent scheduling subsystem, including the ratio of ant numbers to hoisting tasks (range 1-4 times) and pheromone factors. Heuristic factors pheromone volatilization rate pheromone constant Simultaneously, based on the actual construction needs of the project, the maximum number of iterations (200-500 times) and the number of times the optimal path continuously fails to improve (20-40 times) are set; task data is obtained from the data fusion processing center, including the polar coordinates (polar angle) of the pick-up location for each task. Polar diameter ), Installation location polar coordinates (polar angle) Polar diameter Fixed working time and tower crane rotational angular velocity Amplitude variation speed These parameters are then compiled into a task data list.

[0069] The second step involves software and module deployment. The intelligent scheduling subsystem is developed using Python, with PyCharm as the development tool. It integrates the Scikit-learn library for algorithm model building and utilizes the NumPy library for matrix operations and formula calculations, such as the initialization and updating of the pheromone matrix and the calculation of transition time $c_{ij}$. A task data import module is developed to support reading task data from the database (such as MySQL) of the data fusion processing center, automatically parsing and converting it into a format recognizable by the algorithm. A path planning core module is developed, encapsulating the initialization, state transition, pheromone update, and path selection processes of the ant colony algorithm into code logic. The state transition rules and pheromone update formulas are implemented by calling a mathematical calculation library. Finally, a results output module is developed, capable of displaying the optimal task sequence, total time estimate, and performance indicators in tabular form and generating exportable report files.

[0070] The third step involves algorithm testing and debugging. A simulated construction scenario is built, and simulated task data (10-20 hoisting tasks, including different component retrieval and installation locations and fixed operation times) is input. The intelligent scheduling subsystem is then started to run the ant colony algorithm. The changes in pheromone concentration and the fluctuation of total path time during algorithm iteration are observed to verify whether the state transition rule can correctly calculate the task selection probability and whether the pheromone update mechanism can effectively enhance the pheromone concentration of the optimal path. The algorithm results are compared under different numbers of ants and the maximum number of iterations. For example, the path planning efficiency and total time are tested when the number of ants is 1, 2, and 4 times the number of tasks, respectively, to determine the optimal parameter combination suitable for the project scenario. For path planning anomalies encountered during testing (such as repeated task selection or abnormally high total time), the code logic is investigated, and formula calculation errors or parameter setting problems are corrected.

[0071] The fourth step involves actual operation and application of results. The debugged intelligent scheduling subsystem is integrated with the data fusion processing center to establish real-time data interaction, ensuring dynamic acquisition of the latest task data and tower crane parameters. Before construction begins, the system is activated to run the ant colony algorithm. The current lifting task data is input, and the algorithm completes initialization, iterative path search, and pheromone update according to the process, outputting the optimal task sequence, total time estimate, and performance indicators. The optimal task sequence and lifting instructions are then sent to the tower crane control system, and the scheduling results are simultaneously displayed in the remote operator's cab and the digital twin system to guide on-site lifting operations. During construction, if task data changes (such as adding components or adjusting task priorities), the algorithm is restarted for path planning, updating the scheduling results to ensure that the lifting operation is always executed according to the optimal plan.

[0072] Example 4 In step S4, the control modes include remote manual driving and partial automatic driving; The remote manual operation mode is used for routine lifting tasks, where the operator manually controls the tower crane through multi-channel high-definition video monitoring in the remote cab. The partial autonomous driving mode is used for tasks such as pre-set path hoisting, load swing suppression and restricted area protection, respectively achieving automatic cruising to the target coordinate point, real-time closed-loop control to eliminate swing, and autonomous obstacle avoidance in combination with position sensors; Mode switching requires manual authorization for activation or disabling via a remote control console; Emergency situations include tower crane overload, overspeed, and exceeding limits, which are addressed by using local PLC hard safety logic to achieve forced braking independent of the remote system. Issuing a hoisting instruction means that the intelligent scheduling subsystem sends the instruction to the tower crane control system, which then controls the execution of the hoisting task.

[0073] In step S4 of the intelligent scheduling and remote control method for tower cranes, the core focuses on the selection of tower crane control mode, execution of lifting commands, and handling of emergencies. The specific process and rules are as follows: Step S4 comprises four key components: control mode division, mode switching rules, emergency braking mechanism, and lifting command issuance logic. Control modes are divided into two categories: remote manual operation and partial automatic operation. The remote manual operation mode is specifically for routine lifting tasks. In this mode, the operator, located in a remote cab, can view the construction scene, tower crane operating status, and hook and component positions in real time through multi-channel high-definition video monitoring, and manually control the tower crane to complete a series of operations such as component retrieval, lifting, and installation. The partial automatic operation mode targets three specific tasks: preset path lifting tasks, load sway suppression tasks, and restricted area protection tasks. In preset path lifting tasks, this mode enables the tower crane to automatically cruise to the target coordinate point without continuous manual control. In load sway suppression tasks, real-time closed-loop control technology eliminates the sway of the hook and components, ensuring lifting stability. In restricted area protection tasks, combined with tower crane position data collected by position sensors, the system automatically identifies construction restricted areas and autonomously avoids obstacles, preventing the tower crane from entering dangerous areas.

[0074] Mode switching must adhere to strict manual authorization rules. Operators must operate via a remote control console, manually activating or disabling the corresponding driving mode to ensure the safety and controllability of mode switching and prevent risks caused by misoperation. Emergency situations are defined as three scenarios: tower crane overload, tower crane overspeed, and tower crane exceeding boundaries. When any of these emergency situations occur, the local PLC's hard safety logic is activated. This logic operates independently of the remote system and is not affected by remote control signals. It can directly trigger the tower crane's forced braking, quickly stopping all tower crane actions and ensuring equipment and construction safety.

[0075] The hoisting instruction issuance process is clearly defined as follows: the intelligent scheduling subsystem first generates hoisting instructions, which include information such as task sequence, starting position, target position, and path planning. Then, the instructions are sent to the tower crane control system, which finally receives the instructions and controls the tower crane's slewing, trolley luffing, hook lifting and lowering and other actuators to complete the specific hoisting task.

[0076] How to use: 1. Control Mode Selection Preparation: First, check the connection status between the remote cab and the tower crane control system to ensure that the multi-channel high-definition video monitoring signals are clear and the control equipment (such as control handles and buttons) respond normally. At the same time, confirm that the equipment used for partial automatic driving and emergency detection, such as position sensors, torque sensors, and speed sensors, is operating stably. The operator retrieves the list of hoisting tasks to be executed at the remote control console and clarifies whether the task type is a regular hoisting task, a preset path hoisting, load swing suppression, or restricted area protection task, so as to provide a basis for selecting the corresponding control mode.

[0077] 2. Control Mode Activation and Operation: If the task to be performed is a routine hoisting task, the operator selects and activates the remote manual driving mode through the remote control console. Then, the operator observes the construction scene through multiple high-definition video monitors, focusing on the position of precast components, the status of the hook, and the movement of construction personnel. The operator manually controls the control equipment in the remote cab to control the tower crane's rotation, trolley movement, and hook lifting and lowering, completing operations such as component retrieval, transfer to the installation position, and installation. During the process, the operator adjusts the control actions in real time to ensure hoisting accuracy. If the task to be performed is a pre-planned path hoisting, load sway suppression, or restricted area protection task, the operator first manually activates the partial automatic driving mode through the remote control console. For pre-planned path hoisting tasks, the target coordinates are imported into the system in advance. After the mode is activated, the tower crane automatically cruises to the target coordinates along the planned path. For load sway suppression tasks, after the mode is activated, the system automatically starts a real-time closed-loop control program to monitor and eliminate load sway. For restricted area protection tasks, after the mode is activated, the system combines position sensor data to identify restricted areas in real time and control the tower crane to autonomously avoid obstacles.

[0078] 3. Mode Switching Operation: When it is necessary to switch from remote manual driving mode to partial automatic driving mode, or from partial automatic driving mode to remote manual driving mode, the operator must first pause the current hoisting task, ensure that the tower crane is in a stable state, and then find the mode switching function module through the remote control console, manually complete the authorization operation, activate the target mode and disable the current mode. After the mode switch is completed, a simple test of the tower crane's control function is performed. After confirming that the equipment responds normally in the new mode, the hoisting task can be continued.

[0079] 4. Emergency Handling and Command Execution: During the lifting operation, the system monitors the tower crane's operating parameters in real time. When an emergency such as overload, overspeed, or exceeding limits is detected, the local PLC's hard safety logic automatically activates, triggering forced braking. The tower crane immediately stops all operations. After the operator detects the emergency braking, they investigate the cause of the emergency and release the brake after handling it. In the lifting command execution phase, the lifting commands generated by the intelligent scheduling subsystem are automatically sent to the tower crane control system. After receiving the commands, the tower crane control system analyzes the task requirements and controls the slewing mechanism, luffing mechanism, and hoisting mechanism to work together to execute tasks such as picking up parts, lifting, and installation. The operator monitors the command execution in real time from the remote control cab. If any deviation is detected, adjustments are made promptly through the corresponding control mode.

[0080] The control mode design and execution mechanism in step S4 provides a key guarantee for the safe and efficient operation of tower cranes, and its beneficial effects are mainly reflected in three aspects: First, it improves the adaptability and efficiency of hoisting tasks. By dividing the operation into two modes, remote manual driving and partial automatic driving, it matches regular and specific hoisting tasks respectively. Regular tasks rely on operator experience to ensure flexible response to complex scenarios, while specific tasks reduce human intervention through automated operation. For example, hoisting along a preset path does not require continuous manual operation, and load sway suppression improves stability through closed-loop control. This effectively reduces the labor intensity of operators while improving hoisting efficiency and accuracy.

[0081] Secondly, it enhances construction safety. The manual authorization rules for mode switching avoid mode confusion caused by misoperation and prevent safety hazards caused by unauthorized mode switching. The local PLC hard safety logic is independent of the remote system. In emergency situations such as tower crane overload, overspeed, or crossing boundaries, it can quickly and forcibly brake without being affected by remote signals, greatly reducing the risk of equipment damage and personnel injury. At the same time, the restricted area protection function of partial automatic driving further reduces the possibility of tower crane entering dangerous areas through autonomous obstacle avoidance.

[0082] Third, it ensures the reliability and controllability of command execution. The hoisting command is issued from the intelligent scheduling subsystem to the tower crane control system, forming a clear command transmission link to avoid command transmission deviation. At the same time, the operator can monitor the execution of the command in real time in the remote cab and make timely adjustments in combination with the corresponding control mode to ensure that the hoisting task is executed according to the plan. This achieves automated command execution while retaining the space for manual supervision and intervention, thus balancing automation and controllability.

[0083] Example 5 In step S5, the digital twin system constructs a virtual environment corresponding to the physical construction site at a 1:1 scale based on a game engine or professional visualization engine. A multi-source data dynamic mapping algorithm is used to achieve real-time synchronization between the physical and virtual scenes. This algorithm establishes a parameter mapping relationship between physical entities and the virtual model, and drives the virtual model update based on real-time data. The specific formula is as follows: ; In the formula, for The set of state parameters for the virtual scene model at any time includes tower crane position, hook height, slewing angle, load, and component positioning status; for The set of physical scene entity state parameters at any given time consists of tower crane status and component position data collected by sensors; for A real-time environmental awareness data set, including wind speed and the location of construction workers; The parameter mapping coefficient matrix is ​​pre-calibrated based on the component attributes of the BIM model and the mechanical characteristics of the tower crane. This is a mapping function, through which physical parameters are mapped... Convert the virtual scene into state parameters that can be recognized by the virtual model to achieve a 1:1 dynamic alignment between the virtual scene and the physical scene; The digital twin system dynamically maps information including tower crane position, hook height, slewing angle, current lifting weight, and operating / idle status. It automatically identifies and updates the hoisting and positioning status of prefabricated components through RFID tags, and overlays environmental information such as real-time wind speed. The intelligent monitoring system uses a construction schedule deviation analysis algorithm to calculate and visualize the real-time construction progress of multiple prefabricated houses. It identifies critical path tasks by combining the construction logic relationships in the BIM model and monitors the deviation between the critical path tasks and the plan. The schedule deviation calculation model is as follows: ; ; In the formula, for Time of the first Construction progress deviation rate of the building; for Time of the first The actual number of completed components of a building is counted based on the components in place identified by RFID tags; for Time of the first The planned number of completed components for a building is extracted from the construction plan in the BIM model. for The maximum schedule deviation rate corresponding to the critical path task at any time is determined by comparing the schedule deviation rates of the four buildings to identify the critical path task, thus achieving precise monitoring of the critical path.

[0084] In the preferred embodiment, in step S5, the intelligent monitoring system uses a multi-threshold collaborative alarm algorithm to achieve safety early warning. This algorithm avoids false alarms from a single threshold by setting multi-dimensional alarm thresholds and establishing threshold collaborative judgment logic. The specific alarm triggering model is as follows: ; In the formula, for The alarm trigger result is displayed in real time, with 1 indicating that the alarm is triggered and 0 indicating that it is not triggered. for Real-time wind speed values ​​are collected by an anemometer. The wind speed threshold is set according to the tower crane safety operation specifications. for The actual lifting weight of the tower crane at any given time is calculated by the torque sensor. The overload threshold for the tower crane is set according to the rated lifting capacity of the tower crane. for Time of the first Construction progress deviation rate of the building. The time lag threshold is set backwards from the total project duration. When the alarm threshold is triggered, the system will notify the project's key management personnel via SMS. The closed-loop optimization employs a schedule deviation-driven rescheduling algorithm. When the actual construction progress deviates from the plan by more than a threshold, i.e. This triggers the intelligent scheduling subsystem to re-plan the scheduling. The task priority adjustment model after the re-scheduling is triggered is as follows: ; In the formula, For the first Priority of each pending hoisting task For the first The house number to which each task belongs. This represents the current progress deviation rate for the building. For the first The weight of the component corresponding to each task. The maximum weight of all components to be hoisted. , These are the weighting coefficients. When schedule deviation takes priority ; Based on the adjusted task priority, the intelligent scheduling subsystem re-executes the ant colony algorithm in step S3 to plan the optimal path. At the same time, it pushes the real-time attitude of the tower crane, the surrounding environment, and the location information of the components to be lifted from the digital twin system to the remote cab or the unmanned driving system to assist in control decisions.

[0085] In the preferred embodiment, in step S5, the digital archive records all operating instructions, changes in tower crane operating parameters, and changes in construction status for traceability of the construction process.

[0086] In step S5 of the tower crane intelligent scheduling and remote control method, the core revolves around the dynamic mapping of the digital twin system, the operation of the intelligent monitoring system, and closed-loop optimization. It also includes digital archive recording functions. The specific process and model are as follows: Step S5 first constructs a digital twin system. This system builds a virtual environment that corresponds 1:1 to the physical construction site, based on a game engine or professional visualization engine. A multi-source data dynamic mapping algorithm is used to achieve real-time synchronization between the physical and virtual scenes. The multi-source data dynamic mapping algorithm establishes a parameter mapping relationship between physical entities and the virtual model, and drives the virtual model update based on real-time data. The specific formula is as follows: .in The virtual scene model state parameter set represents the set of parameters for time t. The set of parameters for the virtual scene model includes the tower crane position, hook height, slewing angle, load, and component positioning status. This represents the set of physical scene entity state parameters at time t. The set of physical scene entity state parameters consists of tower crane status and component position data collected by sensors. This represents the set of environmental sensing data at time t, which includes wind speed and the location of construction workers. The parameter mapping coefficient matrix is ​​pre-calibrated based on the component attributes and tower crane mechanical characteristics of the BIM model. Represents a mapping function, which maps physical parameters according to... The system converts state parameters into those recognizable by the virtual model, achieving a 1:1 dynamic alignment between the virtual and physical scenes. The digital twin system dynamically maps information including tower crane position, hook height, slewing angle, current load, and operating / idle status. It automatically identifies and updates the hoisting and positioning status of prefabricated components via RFID tags, while simultaneously overlaying and displaying real-time environmental information such as wind speed in the virtual environment.

[0087] The intelligent monitoring system uses a construction schedule deviation analysis algorithm to calculate and visualize the real-time construction progress of multiple prefabricated houses. It identifies critical path tasks by combining the construction logic relationships in the BIM model and monitors the deviations between these critical path tasks and the planned schedule. The construction schedule deviation analysis algorithm comprises two calculation models; the first model is... ,in The deviation rate of the construction progress of the i-th building at time t; The actual number of completed components for the i-th building at time t is represented by the statistics of the in-place components identified by RFID tags. This represents the planned number of completed components for the i-th building at time t, extracted from the BIM model's construction schedule. The second model is... ,in The maximum schedule deviation rate corresponding to the critical path task at time t is represented by the schedule deviation rate of the four houses. The critical path task is determined by comparing the schedule deviation rates of the four houses, so as to achieve accurate monitoring of the critical path.

[0088] In the preferred embodiment, the intelligent monitoring system employs a multi-threshold collaborative alarm algorithm to achieve security early warning. This algorithm avoids false alarms based on a single threshold by setting multi-dimensional alarm thresholds and establishing threshold collaborative judgment logic. The specific alarm triggering model is as follows: .in This represents the alarm trigger result at time t, where 1 indicates that the alarm has been triggered and 0 indicates that it has not been triggered. This represents the real-time wind speed value at time t, which is collected by an anemometer. This represents the wind speed exceeding the threshold, which is set according to the tower crane safety operation specifications. This represents the actual lifting weight of the tower crane at time t, which is calculated from the torque sensor. This represents the tower crane overload threshold, which is set according to the tower crane's rated lifting capacity. The deviation rate of the construction progress of the i-th building at time t; This represents the project lag threshold, which is set backwards from the total project duration. When the alarm threshold is triggered, the system notifies key project management personnel via SMS.

[0089] The closed-loop optimization employs a schedule deviation-driven rescheduling algorithm. When the actual construction progress deviates from the plan by more than a threshold, i.e. This triggers the intelligent scheduling subsystem to re-plan the scheduling. The task priority adjustment model after re-scheduling is as follows: ,in This represents the priority of the j-th lifting task; The house number representing the location of the j-th task; This represents the current progress deviation rate of the house; This represents the weight of the component corresponding to the j-th task; This represents the maximum weight of all components to be hoisted. , Represents the weighting coefficient, and When schedule deviation takes priority Based on the adjusted task priorities, the intelligent scheduling subsystem re-executes the ant colony algorithm in step S3 to plan the optimal path. At the same time, it pushes the real-time attitude of the tower crane, the surrounding environment, and the location information of the components to be lifted from the digital twin system to the remote cab or the unmanned driving system to assist in control decisions.

[0090] In the preferred embodiment, the digital archive records all operating instructions, changes in tower crane operating parameters, and changes in construction status. The digital archive is used for traceability of the construction process.

[0091] In actual deployment, the first step is to build and initialize the digital twin system. This involves selecting a game engine such as Unity or Unreal Engine, or a professional visualization engine (such as ContextCapture), and importing BIM models of multiple prefabricated houses. The BIM models must include information such as component attributes and construction plans. The engine tools are then used to construct a virtual environment that corresponds 1:1 to the physical construction site. A multi-source data dynamic mapping module is then developed within the engine, and mapping functions are written. The code logic is implemented. The calculation, while pre-setting the parameter mapping coefficient matrix. , The components need to be calibrated in advance based on the BIM model component attributes and tower crane mechanical characteristics; deploy RFID tag identification modules to ensure that the system can read RFID tag data to update the prefabricated component hoisting and positioning status, and complete the basic construction of the digital twin system.

[0092] The second step involves deploying an intelligent monitoring system, developing a visual interface using the Qt framework or WebGL, integrating a construction progress deviation analysis algorithm module, and writing... and The calculation code is used to calculate the construction progress of multiple buildings and identify critical path tasks; a multi-threshold collaborative alarm module is developed to... The alarm logic is written into the code, and SMS notification functionality is integrated (which can call the Twilio API or domestic SMS service interface), and a wind speed exceeding threshold can be set. Tower crane overload threshold Progress lag threshold ; Build a data receiving module to ensure that the intelligent monitoring system can acquire data collected by the sensors in real time. , Data, and statistics on RFID tags Extracted from BIM model .

[0093] The third step involves closed-loop optimization and deployment of the digital archive module. This includes developing a schedule deviation-driven rescheduling module within the intelligent scheduling subsystem and writing... Task priority calculation code, setting weight coefficients , An interface was established with the ant colony algorithm in step S3 to ensure that the algorithm can be triggered to replan the path after the priority is adjusted; a data push module was developed to enable the digital twin system to push the tower crane's real-time attitude, surrounding environment, and location information of the components to be lifted to the remote cab or unmanned system; a digital archive module was deployed to store all operation instructions, tower crane operating parameter changes, and construction status changes using MySQL or MongoDB databases, and a data writing and query interface was developed to support traceability of the construction process.

[0094] The fourth step involves system integration and operation. The digital twin system is launched to test the dynamic mapping effect of multi-source data, check the 1:1 alignment accuracy between the virtual and physical scenes, and verify the RFID tag identification and component status update functions. The intelligent monitoring system is then launched to simulate different wind speeds, lifting weights, and progress deviation scenarios, testing the accuracy of alarm triggering and the timeliness of SMS notifications. The visualization of construction progress and critical path identification are checked for normal operation. A progress deviation exceeding the threshold scenario is triggered to verify whether the closed-loop optimization module can adjust task priorities and trigger the ant colony algorithm to replan the path, and to check the accuracy of data push. The digital archive module is tested to confirm that operation commands, parameter changes, and other data can be recorded and queried normally. After successful integration and testing, the system is officially put into use, and its operational status is continuously monitored to ensure stable collaboration among all modules.

[0095] The digital twin and intelligent monitoring collaboration and closed-loop optimization mechanism in step S5 provides assurance for the visualization, safety, and dynamic adjustment of tower crane construction. Its beneficial effects are mainly reflected in three aspects: First, it enables visualization and real-time synchronization of construction scenarios. The 1:1 virtual environment built by the digital twin system accurately synchronizes the tower crane status, component location, and environmental information through a multi-source data dynamic mapping algorithm. Operators can intuitively grasp the overall construction situation without going to the site. Especially in the scenario of simultaneous construction of multiple buildings, it can clearly control the progress and equipment status of each area, reducing decision-making delays caused by information asymmetry.

[0096] Secondly, it improves the accuracy of construction safety and progress control. The multi-threshold collaborative alarm algorithm of the intelligent monitoring system sets early warning thresholds from multiple dimensions such as wind speed, lifting weight, and progress, avoiding false alarms from a single threshold. After the alarm is triggered, it promptly notifies the management personnel, enabling rapid handling of safety hazards. The construction progress deviation analysis algorithm accurately calculates the progress deviation rate and identifies critical path tasks, ensuring that the management personnel focus on core tasks and prevent overall progress delays.

[0097] Third, it enables dynamic optimization and process traceability. The schedule deviation-driven rescheduling algorithm can adjust task priorities according to the actual progress, triggering the intelligent scheduling subsystem to replan the optimal path and ensure construction efficiency. Digital archives record all operation and status data, which facilitates the traceability of the subsequent construction process. If problems occur, the cause can be quickly located. At the same time, it provides data reference for similar projects and promotes the standardization of construction management.

[0098] The above embodiments are merely preferred technical solutions of the present invention and should not be considered as limitations on the present invention. The scope of protection of the present invention should be limited to the technical solutions described in the claims, including equivalent substitutions of the technical features described in the claims. That is, equivalent substitutions and improvements within this scope are also within the scope of protection of the present invention.

Claims

1. A tower crane intelligent scheduling and remote control method, characterized in that: The method comprises: S1, build a distributed system architecture containing a physical layer, a perception control layer, a data processing decision layer, an application layer and a communication layer, and deploy corresponding equipment; S2, collect environment, tower crane state and BIM model related multi-source data, and transmit to the data fusion processing center for fusion processing; S3, based on the ant colony algorithm, the intelligent scheduling subsystem is used to plan the optimal lifting path and task sequence, and the scheduling result is output; S4, select the corresponding control mode, issue the lifting instruction to execute the task, and trigger the forced braking in emergency; S5, map the construction scene through the digital twin system, the intelligent monitoring system monitors the state and alarms, and the feedback data realizes closed-loop optimization and records the digital archives.

2. The tower crane intelligent dispatching and remote control method according to claim 1, characterized in that: In step S1, the physical layer deploys intelligent tower cranes with closed-loop control systems, fabricated houses and prefabricated components. The intelligent tower crane is installed around the fabricated houses arranged in a triangular shape; The perception control layer is configured with environment perception sensors, tower crane state perception sensors and execution control terminals. The environment perception sensors include high-precision laser radars and wide-angle high-definition cameras. The wide-angle high-definition cameras are equipped with AI identification programs for identifying the hooking state. The tower crane state perception sensors include rotary encoders, trolley position sensors, lifting height sensors, torque sensors, anemometers, inclination sensors and vibration sensors; The execution control terminal includes a tower crane control system and a high-performance edge computing unit. The edge computing unit is installed on the tower crane body for data preprocessing and local obstacle avoidance calculation; The data processing decision layer builds an intelligent scheduling subsystem and a data fusion processing center; The application layer is configured with a remote cab, a digital twin and an intelligent monitoring system; The communication layer is deployed with industrial-grade switches and 5G base stations.

3. The tower crane intelligent dispatching and remote control method according to claim 2, characterized in that: The data preprocessing and local obstacle avoidance of the edge computing unit include: S101, data preprocessing step, the edge computing unit receives the multi-source data collected by the perception control layer sensors, including: S1011, data screening, the edge computing unit screens out the construction scene obstacle data collected by the high-precision laser radar and wide-angle high-definition camera in the environment perception sensor. This obstacle data includes house corners, temporary facilities and other components. At the same time, the edge computing unit screens out the real-time running data of the tower crane collected by the rotary encoder, trolley position sensor, lifting height sensor, torque sensor, anemometer, inclination sensor and vibration sensor in the tower crane state perception sensor. This running data includes rotary angle, trolley position, hook height, stress torque, wind speed, tower body perpendicularity and vibration parameters. Invalid data, which is abnormal value beyond the normal range caused by sensor failure, is excluded; S1012, data space-time alignment, based on the time synchronization signal of the communication layer 5G base station, add a timestamp for each type of data screened, and unify the three-dimensional coordinate data of obstacles collected by the laser radar and the two-dimensional image obstacle data collected by the camera into a polar coordinate system with the tower machine rotation center as the origin. The establishment of the polar coordinate system takes the tower machine rotation center as the origin and takes the direction pointing to the assembled house A as the positive direction of the polar axis. The three-dimensional coordinate (x, y, z) of the obstacle collected by the laser radar is converted into polar coordinates The formula is 、 、 , where r is the polar radius, θ is the polar angle and the range is [0, 2π), h is the height, the space coordinates of multi-source data are aligned, and the standardized data consistent in time and space are obtained; S1013, data denoising and compression, the standardized tower vibration parameters, wind speed data are denoised by using moving average filtering algorithm, and the filtering formula is , wherein is the parameter value after denoising at t time, N is the length of the sliding window, the length of the sliding window N is dynamically adjusted according to the sensor sampling frequency, when the sensor sampling frequency , N takes 10; when , N takes 8; when , N takes 5, is the original parameter value at t time, is the sensor sampling interval; the laser radar point cloud data is compressed by using the voxel grid downsampling algorithm, the key contour features of the obstacles are retained, the data transmission and calculation load are reduced, and the effective data after preprocessing is obtained; S102, local obstacle avoidance calculation step, based on the preprocessed valid data, the local obstacle avoidance step of the edge computing unit includes: S1021, obstacle modeling, according to the polar coordinate data of high-precision laser radar after preprocessing, the polar coordinate model of static obstacle in the construction scene is constructed, the static obstacle includes four corners of the house, fixed temporary facilities, and each obstacle is represented as a region set in polar coordinates wherein is the polar radius of the obstacle to the tower center of rotation, is the polar angle of the obstacle, , is the polar radius range of the obstacle, , is the polar angle range of the obstacle; In combination with the real-time position of the dynamic obstacle recognized by the wide-angle high-definition camera AI discrimination program, the dynamic obstacle includes a hoisting component and a construction personnel, the AI program recognizes the type of the dynamic obstacle through a target detection algorithm, and calculates the polar radius r of the dynamic obstacle to the slewing center of the tower crane based on a preset distance calibration model of the camera internal parameters, the camera internal parameters include a focal length f and a pixel size d, and the calculation formula is wherein H is the actual height of the dynamic obstacle, the component height is known, and the personnel height is set as 1.3-1.8 meters by default, is the pixel height of the dynamic obstacle in the camera image, the polar coordinate information of the dynamic obstacle is supplemented to the obstacle model, and a real-time obstacle set is updated ; S1022, tower crane motion trajectory prediction, based on the real-time data preprocessed by the tower crane state perception sensor, the current rotation angle of the tower crane is obtained , the current position polar radius of the trolley , the current height of the hook , combined with the next control instruction issued by the tower crane control system, the control instruction includes the rotation angular velocity ω, the trolley amplitude variation speed v, and the hook lifting speed , the tower crane hook motion trajectory in the future time is predicted, the value is 0.5-2 seconds and is set according to the tower crane running speed, the prediction formula of the trajectory in polar coordinates is 、 、 , wherein 、 、 , respectively the predicted polar radius, the predicted polar angle, and the predicted height of the hook at the moment S1023, collision detection, judge whether the predicted trajectory and the real-time obstacle set whether there is a collision risk, the collision determination condition is: if there is , so that and in the obstacle height range, the obstacle height range is obtained by laser radar data and is represented as , that is, it meets , it is determined that there is a collision risk; S1024, obstacle avoidance instruction generation, if it is determined that there is a collision risk, the edge computing unit calculates the obstacle avoidance adjustment amount based on the collision detection result, and generates a local obstacle avoidance control instruction: If the collision risk is caused by the swing direction, the swing angular velocity is adjusted, and the adjustment formula is wherein is the safety extreme angle for avoiding obstacles, and the safety extreme angle is determined by the following method: if is within the obstacle extreme angle range , when , ; otherwise , is a safety redundancy angle and takes a value of 5-10 degrees, which is set according to the complexity of the construction scene, is a sign function, which ensures that the tower crane swings in a direction away from the obstacle. If the collision risk is caused by the luffing direction of the trolley, adjust the luffing speed of the trolley, and the adjustment formula is wherein In order to avoid the safety extreme radius of the obstacle, the safety extreme radius Take the nearest extreme radius outside the extreme radius range of the obstacle; If the collision risk comes from the height direction, adjust the lifting speed of the hook, and the adjustment formula is wherein The safety height for avoiding obstacles is Take the nearest height outside the height range of the obstacle The adjusted , , As a local obstacle avoidance control instruction, it is sent to the tower crane control system to realize local obstacle avoidance of the tower crane, and the obstacle avoidance process data is uploaded to the data fusion processing center.

4. The tower crane intelligent dispatching and remote control method of claim 1, characterized in that: In step S2, the environmental data includes the house construction progress data detected by the high-precision laser radar, and the precast component state and position, the construction personnel state and position, the hook position and state, and the hoisting point position data collected by the wide-angle high-definition camera; the tower crane state data includes the slewing angle, the trolley running position, the hook lifting height, the torque acting on the tower crane, the wind speed and direction, the tower body verticality, and the tower crane vibration parameters; The BIM model related data includes the size, weight, installation position of the precast component, and the construction progress data of each house; The fusion processing includes time and space alignment, filtering, and fusion of multi-source data to generate unified and accurate data.

5. The tower crane intelligent dispatching and remote control method of claim 1, characterized in that: In step S3, the scheduling target of the ant colony algorithm is to minimize the total operation time, and the objective function model is: wherein, is the task sequence, is the start, is the total number of components, i.e. the total number of tasks, is the transfer time from task to task , is the fixed operation time of task , including picking and installation; The initialization of the ant colony algorithm comprises: setting the number of ants to be 1-4 times the number of hoisting tasks, an information factor , a heuristic factor , an information volatility rate , an information constant , a maximum iteration number times; creating an information matrix of (task number+1)×(task number+1), and setting the initial value of the matrix to 0.1; From task To task Transition time The computational model is ； In the formula, is the task polar angle of the pickup position, is the task polar angle of the installation position, is the tower rotation angular velocity (degree / second), is the task polar radius of the pickup position, is the task polar radius of the installation position, is the tower luffing speed.

6. The tower crane intelligent dispatching and remote control method according to claim 5, characterized in that: In step S3, the state transition rule of the ant colony algorithm is: ； wherein is the number of iterations, is the number of iterations, is the number of iterations, is the number of iterations, is the number of iterations, is the number of iterations, is the number of iterations, is the number of iterations, is the number of iterations, is the number of iterations, is the number of iterations, is the number of iterations, is the number of iterations, is the number of iterations, is the number of iterations, is the number of iterations, The process of planning the optimal hoisting path and task sequence includes: A virtual starting point, i.e., the tower crane center, is set for each ant, the visited task list and task sequence are cleared, the candidate tasks with incomplete preconditions are filtered out from the unvisited tasks, and when there is no selectable task, the task with the least missing preconditions is selected and the number of selections is recorded; the movement time from the current position to the pickup point of each candidate task is calculated, which is the larger value of the rotation time and the amplitude change time; Based on the above state transition rule, the task selection probability is calculated, the task is selected by the roulette method and added to the path sequence, the task is marked as visited and the current position is updated to the installation point of the task, and the above process is repeated until all tasks are visited; The total path time of each ant is calculated, which is the cumulative value of the movement time between tasks and the fixed operation time of each task.

7. The tower crane intelligent dispatching and remote control method of claim 5, characterized in that: In step S3, the pheromone update mechanism of the ant colony algorithm includes global update, and the specific formula is: ； ； ； In the formula, For the first Task during the next iteration To the mission The concentration of pheromones along the path, For path pheromone increment, The number of ants. For the first Only ants on the path The increase in pheromones left on the surface For the first The total path length after an ant completes the entire path, which is the total operation time; Before outputting the scheduling result, it also includes comparing the total path time of all ants in this round to determine the optimal path in this round, and if the optimal path in this round is better than the historical optimal path, the global optimal path is updated; The pheromone of all paths is volatilized and incrementally updated according to the above pheromone update mechanism; The planning termination condition is to reach the maximum iteration number The scheduling result includes the optimal task sequence, the total time estimation, and the performance index, and the performance index includes the tower crane utilization rate, the time saving rate, and the constraint satisfaction statistics.

8. The tower crane intelligent dispatching and remote control method of claim 1, characterized in that: In step S4, the control modes include remote manual driving and local automatic driving; The remote manual driving mode is used for regular hoisting tasks, and the operator manually controls the tower crane through the multi-channel high-definition video monitoring of the remote cab; The local automatic driving mode is used for preset path hoisting, load swing suppression, and forbidden area protection tasks, which respectively realize automatic cruising to the target coordinate point, real-time closed-loop control to eliminate swing, and autonomous obstacle avoidance combined with position sensors; Mode switching needs to be manually authorized by the remote operation station to activate or disable; Emergency situations include tower crane overload, overspeed, and boundary crossing, and the local PLC hard safety logic is used to achieve forced braking independent of the remote system; Issuing hoisting instructions means that the intelligent scheduling subsystem issues instructions to the tower crane control system, which controls the execution of hoisting tasks.

9. The tower crane intelligent dispatching and remote control method of claim 1, characterized in that: In step S5, the digital twin system constructs a virtual environment corresponding to the physical construction site 1:1 based on a game engine or a professional visualization engine, and realizes real-time synchronization between the physical scene and the virtual scene using a multi-source data dynamic mapping algorithm. The algorithm establishes a parameter mapping relationship between physical entities and virtual models, and updates the virtual models driven by real-time data. The specific formula is: ； In the formula, is The virtual scene model state parameter set at the moment includes the tower crane position, hook height, rotation angle, hoisted weight, and component in-place state. For The moment physical scene entity state parameter set is composed of tower crane state and component position data collected by sensors. for a collection of momentary environmental perception data, including wind speed, construction worker position; is a parameter mapping coefficient matrix, which is calibrated in advance according to the BIM model component attribute and tower crane mechanical characteristics, This is a mapping function, through which physical parameters are mapped... Convert the virtual scene into state parameters that can be recognized by the virtual model to achieve a 1:1 dynamic alignment between the virtual scene and the physical scene; The dynamic mapping content of the digital twin system includes tower crane position, hook height, rotation angle, current hoisting weight, running / idle state, automatic identification and update of the hoisting and positioning state of prefabricated components through RFID tags, and superimposed display of real-time wind speed and other environmental information. The intelligent monitoring system calculates and visualizes the real-time construction progress of multiple prefabricated buildings using the construction progress deviation analysis algorithm, identifies key path tasks based on the construction logic relationship in the BIM model, monitors the deviation between the key path tasks and the plan, and the progress deviation calculation model is: ； ； In the formula, is the moment the construction progress deviation rate of the house To Moment The actual number of completed components of the house, the in-place component statistics identified by the RFID tag; For The moment the The planned completion component quantity of the house is extracted from the BIM model construction plan; For The maximum progress deviation rate corresponding to the critical path task at the moment is determined by comparing the progress deviation rates of the four buildings, and the critical path task is determined to realize accurate monitoring of the critical path.

10. The tower crane intelligent dispatching and remote control method of claim 1, characterized in that: In step S5, the intelligent monitoring system realizes safety warning using a multi-threshold collaborative alarm algorithm. This algorithm sets multiple-dimensional alarm thresholds and establishes a threshold collaborative judgment logic to avoid false alarms caused by a single threshold. The specific alarm triggering model is: ； In the formula, is Alarm trigger result at the moment, 1 represents triggering alarm, 0 represents not triggering; For Real-time wind speed values are collected by the anemorumbometer at the moment. For wind speed threshold, according to the tower crane safety operation specification; For The actual hoisting weight of the tower crane is converted by the torque sensor. The tower machine overload threshold is set according to the rated load of the tower machine. For The The construction progress deviation rate of the house, The progress lag threshold is set according to the total project duration. When the alarm threshold is triggered, the system notifies the main project managers through SMS. The closed-loop optimization employs a schedule deviation-driven rescheduling algorithm. When the actual construction progress deviates from the plan by more than a threshold, i.e. This triggers the intelligent scheduling subsystem to re-plan the scheduling. The task priority adjustment model after the re-scheduling is triggered is as follows: ； In the formula, is the priority of the th task to be hoisted, is the house number to which the th task belongs, is the current progress deviation rate of the house, is the weight of the component corresponding to the th task, is the maximum weight of all components to be hoisted, , is the weight coefficient, , when the progress deviation is prioritized ; The intelligent scheduling subsystem re-executes the ant colony algorithm in step S3 to plan the optimal path based on the adjusted task priority, and pushes the real-time attitude of the tower crane, the surrounding environment, and the position information of the components to be hoisted to the remote cab or the unmanned driving system to assist in decision-making.

11. The tower crane intelligent dispatching and remote control method of claim 10, characterized in that: In step S5, the digital archive records all operation instructions, changes in tower crane operating parameters, and changes in construction status, which are used for construction process tracing.

Images