Beidou digital twin system and hoisting method for precision hoisting of bridge steel components

Abstract

The present application belongs to the field of bridge engineering construction monitoring and intelligent control, aiming at the problems of low efficiency and poor real-time performance in the field hoisting of segment steel members in long-span bridge structure, a Beidou digital twin system and hoisting method for bridge steel member precision hoisting are proposed. The system includes a multi-source data acquisition module, a coordinate conversion module, a posture solving module, a three-dimensional virtual model construction module, a hoisting motion state monitoring module, and a digital twin module. The posture solving module calculates the standard attitude vector using a comprehensive deep learning model according to the degree of freedom data, attitude information, and motion speed, and performs attitude solving based on the standard attitude vector to obtain the attitude angle of the bridge steel member. The hoisting motion state monitoring module can perform safety warning and attitude adjustment when the bridge steel member is abnormal. The digital twin module can synchronously update the attitude information and three-dimensional virtual model of the bridge steel member, and obtain the current spatial trajectory and historical state of the bridge steel member.

Description

Technical Field

[0001] This application relates to the field of bridge engineering construction monitoring and intelligent control technology, specifically to a Beidou digital twin system and hoisting method for precision hoisting of bridge steel components. Background Technology

[0002] In recent years, with the rapid development of long-span bridge engineering, steel structural components have been widely used in bridge superstructures due to their high strength, high rigidity, and ease of industrial manufacturing. During bridge construction, the hoisting accuracy of steel components directly affects the overall quality of the structural installation and construction safety. Currently, commonly used hoisting methods rely mainly on manual visual inspection combined with laser rangefinders or total stations for attitude monitoring. This method is inefficient, lacks real-time performance, and is highly dependent on operators in high-altitude operations or complex environments, posing significant error risks and safety hazards.

[0003] Meanwhile, the application of the BeiDou Navigation Satellite System in high-precision positioning has gradually matured, and the accuracy of sensors such as inclinometers and IMUs has continued to improve, providing a technological foundation for spatial attitude perception during the hoisting of bridge steel components. The rise of digital twin technology has further provided new pathways for the virtual-real mapping, dynamic control, and visual management of components. However, there is still a lack of a system solution that integrates BeiDou positioning, attitude calculation, digital twin models, and hoisting control. There is an urgent need to develop a precision hoisting system with real-time monitoring, intelligent early warning, and automatic adjustment capabilities to meet the higher requirements of modern bridge engineering for construction accuracy and safety. Summary of the Invention

[0004] In view of the above problems, this application provides a Beidou digital twin system and method for precision hoisting of bridge steel components, which aims to adjust the attitude of bridge steel components during the hoisting process in order to overcome or at least partially solve the above problems.

[0005] In a first aspect, embodiments of this application provide a BeiDou digital twin system for precision hoisting of bridge steel components, the system comprising:

[0006] The multi-source data acquisition module is used to acquire the degrees of freedom, attitude information and motion velocity of bridge steel components in three-dimensional space in real time;

[0007] The coordinate transformation module is used to transform the collected degrees of freedom data from the geodetic coordinate system to the local coordinate system;

[0008] The attitude calculation module calculates the standard attitude vector based on the degree of freedom data and attitude information in the local coordinate system and the motion speed using a comprehensive deep learning model, and performs attitude calculation based on the standard attitude vector to obtain the attitude angle of the bridge steel component.

[0009] The 3D virtual model construction module is used to build a 3D virtual model based on the 3D laser scanning point cloud of bridge steel components, which can reflect the changes in the actual hoisting process of bridge steel components in real time.

[0010] The hoisting motion status monitoring module determines the attitude deviation of the bridge steel components based on the attitude angle and issues an early warning message when an abnormality occurs. Based on the early warning message, it determines the adjustment amount of the hoisting points and adjusts the attitude of the bridge steel components.

[0011] The digital twin module is used to synchronously update the attitude information and three-dimensional virtual model of the bridge steel components during the entire hoisting process, and to obtain the current spatial trajectory and historical status of the bridge steel components in real time based on the three-dimensional virtual model.

[0012] The archiving module is used to record and save the attitude deviations, early warning information, lifting point adjustment amounts, and three-dimensional laser scanning point cloud data of bridge steel components.

[0013] Optionally, the hoisting motion status monitoring module includes:

[0014] The early warning submodule determines whether the attitude of the bridge steel component is abnormal or whether there is an abnormal spatial position offset based on the attitude angle obtained from the attitude angle, and issues an early warning message when an abnormality occurs.

[0015] The lifting point tension adjustment submodule determines the lifting point adjustment amount based on the received warning information, adjusts the lifting point position or sling length of the crane, and corrects the posture of the bridge steel components in real time.

[0016] Optionally, a standard attitude vector is calculated using a comprehensive deep learning model based on the degree-of-freedom data and attitude information in the local coordinate system, as well as the motion velocity. Then, attitude calculation is performed based on the standard attitude vector to obtain the attitude angles of the bridge steel components, including:

[0017] The state vector is composed of the degrees of freedom data in the local coordinate system, attitude information, and motion velocity.

[0018] Input the state vector into the trained integrated deep learning model to obtain the standard pose vector;

[0019] The current attitude angle of the bridge steel component is obtained by attitude calculation based on the standard attitude vector.

[0020] Optionally, the integrated deep learning model includes a Transformer network module, a multilayer perceptron module, and an output layer. The output layer contains three output nodes: a position output node, a velocity output node, and a pose output node.

[0021] Optionally, the process of establishing the comprehensive deep learning model includes:

[0022] The input sample data is standardized, and the standard pose vector of the next time step is used as the label of the sample data.

[0023] Based on the first pruning hyperparameter, the connection relationships of each neuron within the MLP module of the first integrated deep learning model and the connection relationships between each neuron within the MLP module and the position output node, velocity output node and attitude output node are pruned to obtain the second integrated deep learning model; wherein, the first pruning hyperparameter is the pruning hyperparameter obtained by the first integrated deep learning model after the previous model training.

[0024] Based on the first repair hyperparameter, the connection relationships of each neuron inside the MLP module of the second integrated deep learning model and the connection relationships between each neuron inside the MLP module and the position output node, velocity output node and attitude output node are repaired to obtain the third integrated deep learning model; wherein, the first repair hyperparameter is the repair hyperparameter obtained by the first integrated deep learning model after the previous model training.

[0025] The standardized sample data is input into the third integrated deep learning model to obtain the predicted pose vector for the next time step.

[0026] The model loss value of the third integrated deep learning model is calculated based on the predicted pose vector and the standard pose vector.

[0027] The model parameters and hyperparameters of the first integrated deep learning model are adjusted by adjusting the model loss value to obtain the trained integrated deep learning model.

[0028] Optionally, the system further includes an intelligent data processing center, which comprises:

[0029] The data processing center is used to receive multi-source data and early warning information collected by the multi-source data acquisition module, and to clean, fuse, analyze and store the multi-source data, remove invalid data and interference information, extract feature parameters, and analyze the early warning information to send the obtained lifting point adjustment amount to the archiving module.

[0030] The model computation center is used to drive the dynamic updating of the 3D virtual model based on feature parameters, so as to realize the mapping between the 3D virtual model and the actual hoisting state.

[0031] Secondly, embodiments of this application provide a method for hoisting steel components for bridges, based on the aforementioned system implementation, the method comprising:

[0032] Real-time acquisition of the degrees of freedom, attitude information, and motion velocity of bridge steel components in three-dimensional space;

[0033] The collected degrees of freedom data are transformed from the geodetic coordinate system to the local coordinate system;

[0034] The standard attitude vector is obtained based on the degree of freedom data and attitude information in the local coordinate system and the motion velocity. The attitude is then calculated based on the standard attitude vector to obtain the current attitude angle of the bridge steel component.

[0035] Based on the three-dimensional laser scanning point cloud of bridge steel components, a three-dimensional virtual model is constructed, and the motion state and attitude of the bridge steel components during the actual hoisting process are synchronously mapped through rigid body transformation.

[0036] Calculate the position offset and attitude deviation of the bridge steel components. If the position offset or attitude deviation exceeds the threshold, issue an early warning.

[0037] Based on the early warning information, determine the lifting point adjustment amount, adjust the lifting point position of the crane or the length of the sling, and correct the posture of the bridge steel components in real time;

[0038] The attitude information of the bridge steel components is updated synchronously with the three-dimensional virtual model, and the current spatial trajectory and historical status of the bridge steel components are obtained in real time based on the three-dimensional virtual model.

[0039] Optionally, the positional offset and attitude deviation of the bridge steel components are calculated. If the positional offset or attitude deviation exceeds a threshold, an early warning message is issued, including:

[0040] Calculate the spatial distance between the current position of the bridge steel component and the target position, and use it as the position offset;

[0041] Calculate the Euclidean difference between the current attitude angle and the target attitude angle of the bridge steel component, and use it as the attitude deviation;

[0042] Determine whether the attitude deviation is greater than the attitude angle threshold or the position offset is greater than the offset threshold;

[0043] If the position offset is greater than the offset threshold, a position offset warning message is issued; if the attitude deviation is greater than the attitude angle threshold, proceed to the next step.

[0044] If the tilt component of the attitude deviation is greater than the tilt threshold, a lateral tilt warning is issued; if the pitch component of the attitude deviation is greater than the pitch threshold, a longitudinal tilt warning is issued; if the roll component of the attitude deviation is greater than the tilt threshold, a roll warning is issued.

[0045] Optionally, based on the warning information, the adjustment amount of the lifting point is determined, and the position of the lifting point or the length of the sling is automatically adjusted, including:

[0046] Calculate the lifting point adjustment amount based on the lifting point position and attitude information in the early warning information;

[0047] The displacement adjustment component and the sling length adjustment component are obtained based on the lifting point adjustment amount;

[0048] Adjust the position of the lifting point based on the displacement adjustment component;

[0049] Adjust the sling length based on the sling length adjustment component.

[0050] Optionally, the formula for calculating the adjustment amount of the lifting point is:

[0051] ;

[0052] In the formula, The adjustment amount for the output suspension point; This refers to the proportional control gain coefficient. The pseudo-inverse Jacobian matrix is ​​used to arrange the suspension points, describing the relationship between the attitude and the displacement of the suspension points; This represents the proportion of the center of gravity shift. Let the centroid position vector of the component be denoted as . This is the current component's center of gravity offset vector.

[0053] Compared with the prior art, the specific beneficial effects of the present invention are as follows:

[0054] First, this invention integrates BeiDou high-precision positioning and tilt sensing technology to achieve real-time perception and virtual-real synchronization of the six-degree-of-freedom spatial attitude of the hoisted components, effectively improving the visualization level and control precision of the hoisting process. Furthermore, it automatically triggers an alarm and links the sling control system to correct the attitude when the component's attitude is abnormal, significantly reducing safety risks caused by skewness, rotation, etc., during hoisting. In summary, this system has advantages such as flexible deployment, rapid response, and closed-loop control, making it particularly suitable for precision hoisting operations of large-volume, complex bridge steel components, and has good engineering application prospects and promotional value.

[0055] Secondly, this invention constructs a comprehensive deep learning model for generating standard pose vectors. It sets up three independent output nodes, each outputting the angular elements of the pose vector in one of the three directions. Because the output nodes are independent, the prediction accuracy of each angular element is fully guaranteed, thereby improving the overall prediction accuracy of the pose vector. Furthermore, during the training of the comprehensive deep learning model, each training round does not retain the results of the previous pruning. Instead, it re-prunes the comprehensive deep learning model, which contains the original connections but retains only the other model parameters adjusted after the previous training. This pruning method avoids the comprehensive deep learning model from crashing, preventing convergence and thus improving the stability of the training. In addition, since the final trained comprehensive deep learning model reduces the connections between nodes compared to the original model, it significantly improves response speed and the efficiency of obtaining predicted pose vectors. Attached Figure Description

[0056] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments of this application will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0057] Figure 1 This is a system module framework diagram of the present invention;

[0058] Figure 2 This is a schematic diagram of the steel components (steel box girder) of the bridge;

[0059] Figure 3 This is a flowchart of the method proposed in this invention;

[0060] Figure 4 This is a flowchart of the workflow of each module in the system of the present invention;

[0061] Figure 5 This is an architecture diagram of the comprehensive deep learning model of the present invention;

[0062] Figure 6 It is a convergence curve of the model loss value as the number of training iterations increases;

[0063] Figure 7 It is a coordinate graph showing the predicted value and standard value in the X direction of the predicted attitude vector;

[0064] Figure 8 It is a coordinate graph showing the predicted value and standard value in the Y direction of the predicted attitude vector;

[0065] Figure 9 It is a coordinate graph showing the predicted value and standard value in the Z direction of the predicted attitude vector;

[0066] Among them: 1-BeiDou satellite; 2-small BeiDou RTK positioning module; 3-steel box girder; 4-crane; 5-intelligent data processing center; 6-alarm judgment and control equipment; 7-high-definition camera; 8-high-precision inclinometer; 9-IMU / velocity sensor; 10-top plate; 11-bottom plate; 12-partition. Detailed Implementation

[0067] Exemplary embodiments of this application will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of this application are shown in the drawings, it should be understood that this application may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of this application and to fully convey the scope of this application to those skilled in the art.

[0068] Example 1: The BeiDou digital twin system for precision hoisting of bridge steel components provided in this application includes:

[0069] (1) Multi-source data acquisition module

[0070] The multi-source data acquisition module is used to acquire in real time the degree of freedom data, attitude information and motion velocity of bridge steel components in three-dimensional space.

[0071] Optionally, combined Figures 1-2 The multi-source data acquisition module includes small Beidou RTK positioning modules 2 installed at the four corners of the steel box girder 3 and high-precision inclinometers 8 and IMU / velocity sensors 9 on the steel box girder 3. These modules together constitute the perception layer, which can collect the degree of freedom data, attitude information and motion speed of the bridge steel components in three-dimensional space in real time, with positioning accuracy down to the centimeter level. In this embodiment, the bridge steel components can specifically be steel box girder 3, which includes a top plate 10, a bottom plate 11 and a partition plate 12 set between the top plate 10 and the bottom plate 11.

[0072] The small Beidou RTK positioning module 2 uses RTK differential technology to receive data sent by Beidou satellite 1 to obtain the degree of freedom data of bridge steel components. Its positioning accuracy is better than ±1cm and supports three-point or four-point synchronous measurement.

[0073] The high-precision inclinometer 8 is a three-axis inclinometer used to acquire real-time attitude information of bridge steel components. Its angle accuracy is better than ±0.1° and it has a temperature compensation function.

[0074] IMU / velocity sensor 9 is used to collect the movement speed of bridge steel components.

[0075] Finally, the attitude changes of the bridge steel components are comprehensively sensed by using a high-precision inclinometer 8, a small Beidou RTK positioning module 2, and an IMU / velocity sensor 9, namely, the displacement and attitude deviation of the components caused by the force during the hoisting process.

[0076] (2) Coordinate transformation module

[0077] The coordinate transformation module is used to transform the collected degrees of freedom data from the geodetic coordinate system (CGCS2000 coordinate system) to the local coordinate system;

[0078] Optionally, the degree-of-freedom data acquired in real time by the small Beidou RTK positioning module 2 is three-dimensional coordinates in the CGCS2000 coordinate system. This module is used to convert the three-dimensional coordinates in the CGCS2000 coordinate system to a local coordinate system defined by the project.

[0079] (3) Attitude calculation module

[0080] The attitude calculation module calculates the standard attitude vector based on the degree of freedom data and attitude information in the local coordinate system and the motion speed using a comprehensive deep learning model, and performs attitude calculation based on the standard attitude vector to obtain the attitude angle of the bridge steel component.

[0081] Specifically, based on the degrees of freedom data and attitude information in the local coordinate system, and the motion velocity, a standard attitude vector is calculated using a comprehensive deep learning model. Attitude calculation is then performed based on this standard attitude vector, including:

[0082] Combine the degrees of freedom data with attitude information and motion velocity to form a state vector;

[0083] The state vector is input into the integrated deep learning model to obtain the standard pose vector;

[0084] The attitude angles of bridge steel components are obtained by attitude calculation based on standard attitude vectors.

[0085] Furthermore, such as Figure 5As shown, this application employs a comprehensive deep learning model with the Transformer model as its backbone. This comprehensive deep learning model includes a Transformer network module, a Multilayer Perceptron (MLP) module, and an output layer. The MLP module connects the Transformer network module and the output layer. The MLP module contains at least eight hidden layers; except for the first layer connected to the Transformer network module, which has one node, the number of nodes in each hidden layer is arbitrarily selected between 6 and 10. The output layer contains three output nodes: a position output node, a velocity output node, and a pose output node. The activation function for each output node is a softmax function with different weight parameters. During the training phase, the connections between each node (neuron) in the MLP module are also considered as model parameters.

[0086] The softmax functions with different weight parameters are as follows:

[0087] ;

[0088] Where N is the number of samples (1000). Let X be the weight hyperparameter in the X direction. Let be the input value in the X direction of the i-th sample data. Let be the input value in the X direction of the j-th sample data. It is worth noting that this is not the input value of the model, but the input value of the output node. In essence, there is no substantial difference between the input values ​​of the three nodes; the difference is only determined by the network node of the last layer of the MLP that connects to this output node.

[0089] Furthermore, the process of establishing the comprehensive deep learning model includes:

[0090] First, the input sample data is standardized using an input paradigm, which is formatted as [position, velocity, attitude], where attitude includes three angles in the X, Y, and Z directions. The sample data can be collected from a standard component at the experimental location, and its velocity is represented as angular velocities in the X, Y, and Z directions. For the same standard component, the standard attitude vector of the next time step is used as the label for the sample data.

[0091] Optionally, based on the first pruning hyperparameter, the connections between nodes (neurons) within the MLP module of the first integrated deep learning model, as well as the connections between each node within the MLP module and the position output node, velocity output node, and attitude output node of the output layer, are randomly pruned to obtain the second integrated deep learning model. The first pruning hyperparameter is defined as "the number of prunes between any two layers does not exceed 5"; the first integrated deep learning model can refer to an integrated deep learning model at any training stage.

[0092] Optionally, after model pruning, the connections within the MLP modules of the pruned integrated deep learning model (i.e., the second integrated deep learning model) and between the MLP modules and the output layer can be repaired to obtain a pruned and repaired integrated deep learning model (i.e., the third integrated deep learning model). The repair is based on a first repair hyperparameter, which is defined as "the number of repairs between any two layers does not exceed 5". It should be noted that although the maximum values ​​of the first repair hyperparameter and the first pruning hyperparameter are equal, these two hyperparameters are independent and do not affect each other.

[0093] Then, the standardized sample data is input into the third integrated deep learning model to obtain the predicted pose vector for the next time step output by the third integrated deep learning model.

[0094] Next, the model loss value of the third integrated deep learning model is calculated based on the predicted pose vector and the standard pose vector. The model loss value includes the cross-entropy loss value and / or the mean squared error loss value. It is important to note that both the cross-entropy loss value and the mean squared error loss value are calculated and summed across the three dimensions XYZ, as shown in the following formula:

[0095] ;

[0096] ;

[0097] ;

[0098] ;

[0099] in, This refers to the X component in the standard attitude vector. Let X be the X component in the predicted pose vector output by the model, and N be the number of samples. The cross-entropy loss is in the X dimension. , , The mathematical forms are exactly the same; only the mathematical forms shown here are actually used. .

[0100] The same applies here; therefore, the above formula only shows the mean squared error loss in the X dimension. .

[0101] Finally, the model parameters and hyperparameters of the first integrated deep learning model are adjusted by the model loss value to obtain the first integrated deep learning model with updated model parameters, the first pruning hyperparameters with updated parameters, and the first repair hyperparameters with updated parameters.

[0102] Repeat the step of standardizing the input sample data and using the standard pose vector of the next time step as the label of the sample data to start the next model training.

[0103] When the model loss value meets the preset convergence condition, or when the training times of the first integrated deep learning model reach the preset number of times, the adjustment of the model parameters and hyperparameters of the first integrated deep learning model is stopped, and the trained integrated deep learning model is obtained.

[0104] Furthermore, each training iteration of the model involves pruning the complete integrated deep learning model, but all model parameters and hyperparameters retain the values ​​obtained from the previous training iteration. Optionally, in this application, the learning rate is set to 0.0001, the number of training epochs is set to 10, and the number of sample data in each epoch is set to 1000.

[0105] No pruning is performed when using the model. The state vector of the target bridge steel component is input into the trained integrated deep learning model to obtain the model output attitude vector. Then, each value in the attitude vector is rounded to obtain the standard attitude vector.

[0106] Figures 6-9 It is the prediction result of a comprehensive deep learning model, in which, Figure 6 The blue curve is formed by connecting the model loss values ​​corresponding to each epoch, and the red curve is the fitting curve formed by connecting the model loss values ​​corresponding to each 10 epochs. Figures 7-9 The blue dots represent the positions where the predicted attitude vector corresponds to the standard attitude vector, and the red dashed line indicates that the predicted attitude vector is completely equal to the standard attitude vector. It can be seen that the predicted attitude vector is very close to the standard attitude vector, especially in the small angle range, the prediction accuracy is very high, which proves the feasibility of this method.

[0107] In this application, during model training, the results of the previous pruning are not retained in the next training iteration. Instead, the integrated deep learning model, which contains the original connections but retains only the other model parameters adjusted after the previous training, is pruned again. Furthermore, since the integrated deep learning model used in this method is a lightweight model, this pruning method can avoid the problem of model crashing and failing to converge caused by traditional pruning methods.

[0108] (4) Three-dimensional virtual model construction module

[0109] The 3D virtual model building module is used to build a 3D virtual model based on the 3D laser scanning point cloud of bridge steel components, which can reflect the changes in the actual hoisting process in real time.

[0110] Optionally, a three-dimensional virtual model of the bridge steel components can be established by three-dimensional laser scanning, and the three-dimensional virtual model adopts a rigid body skeleton driving method to achieve attitude synchronization.

[0111] (5) Lifting motion status monitoring module

[0112] The hoisting motion status monitoring module, installed on the alarm discrimination and control device 6, interacts with the intelligent data processing center and the PLC control module of the crane 4 to realize safety warnings and attitude adjustments when the crane lifts bridge steel components; it includes the following sub-modules:

[0113] The early warning submodule is used to issue early warning information when the attitude or spatial position of the hoisted bridge steel components is abnormal.

[0114] The warning information includes lateral tilt warning information, longitudinal tilt warning information, rollover warning information, and position offset warning information;

[0115] The lifting point tension adjustment submodule is used to adjust the lifting point position or sling length of crane 4 according to the received warning information, and correct the posture of bridge steel components in real time.

[0116] Furthermore, after receiving the warning information, the lifting point tension adjustment submodule analyzes and processes the warning information, calculates the lifting point adjustment amount, and generates position movement commands and length adjustment commands. These commands are then sent to the crane's PLC module and the electric sling extension mechanism. Finally, the crane's PLC module receives the position movement command, parses it, and controls the electric slide rail or trolley mechanism of the lifting beam to adjust the lifting point position according to the adjustment amount. Next, the electric sling extension mechanism receives and parses the length adjustment command, adjusting the sling length according to the adjustment amount, thereby achieving posture adjustment. The lifting point adjustment amount includes position translation and sling extension, used to determine the adjustment amount of the lifting point position and sling length, respectively. In addition, a high-definition camera 7 can be installed on the crane 4 to capture environmental photos of the current lifting site.

[0117] (6) Digital Twin Module

[0118] The digital twin module is used to continuously and synchronously update the attitude information and three-dimensional virtual model of the bridge steel components throughout the entire hoisting process, and to obtain the current spatial trajectory and historical status of the bridge steel components in real time based on the three-dimensional virtual model.

[0119] Furthermore, by continuously synchronizing and updating the posture information with the 3D virtual model, construction personnel can monitor the current spatial trajectory and historical status of the bridge steel components in real time through the visualization platform. The visualization platform is developed using Web 3D (with the Three.js framework) technology to achieve multi-angle rotation, trajectory tracing, and interactive control.

[0120] (7) Archiving module

[0121] The archiving module is used to record and save the attitude deviation, early warning information, lifting point adjustment amount and three-dimensional laser scanning point cloud data of bridge steel components;

[0122] Specifically, after the bridge steel components are in place, the system will archive the entire process data, and the archived data will be saved in either PDF report or BIM / Laser point cloud format.

[0123] (8) Intelligent Data Processing Center

[0124] This system also includes an intelligent data processing center 5, which is the system's processing layer. The intelligent data processing center 5 may include a data processing center and a model computation center. Wherein:

[0125] The data processing center receives multi-source data and alarm information collected by the multi-source data acquisition module, and cleans, merges, analyzes and stores the multi-source data, removes invalid data and interference information, and extracts feature parameters; at the same time, after analyzing the early warning information, it sends the obtained lifting point adjustment amount to the archiving module.

[0126] The model computation center is used to drive the dynamic updating of the 3D virtual model based on feature parameters, so as to achieve accurate mapping between the 3D virtual model and the actual hoisting status.

[0127] Example 2: Combination Figures 3-4 This application also provides a method for precision hoisting of bridge steel components, which is based on the aforementioned BeiDou digital twin system for precision hoisting of bridge steel components. The method includes the following steps:

[0128] Step 1: Real-time acquisition of the degrees of freedom, attitude information, and motion velocity of the bridge steel components;

[0129] Step 2: Transform the collected degrees of freedom data from the geodetic coordinate system to the local coordinate system;

[0130] Specifically, since the degree-of-freedom data collected by the BeiDou RTK positioning module 2 is in 3D coordinates under the CGCS2000 (geocentric coordinate system), it needs to be converted to coordinates under the local coordinate system of the construction project through coordinate transformation. When converting between two different 3D Cartesian coordinate systems, a seven-parameter model (seven-parameter coordinate transformation) is typically used. The formula for the seven-parameter coordinate transformation is:

[0131] ;

[0132] ;

[0133] ;

[0134] ;

[0135] ;

[0136] in (x, y, z) are the three-dimensional coordinates in the geodetic coordinate system; (x, y, z) are the three-dimensional coordinates in the local coordinate system; , , () represents the translation parameters in three directions; The scaling factor; , , () represents the rotation angle about the X, Y, and Z axes; It is a rotation matrix (using ZYX order Euler angles). , , Rotation around the X-axis Angle rotation matrix, rotation about the Y-axis Rotation matrix of angle, rotation about Z-axis Rotation matrix of the angle.

[0137] Step 3: Obtain the standard attitude vector based on the degree of freedom data and attitude information and motion velocity in the local coordinate system, and perform attitude calculation based on the standard attitude vector to obtain the current attitude angle (Eulerian angle triplet) of the bridge steel component.

[0138] Optionally, attitude calculation based on the standard attitude vector can be achieved by calculating Euler angles from the rotation matrix. First, the orientation matrix of the local coordinate system is directly calculated using the geometric point coordinates of the component or the surface features of the component. Then, in Based on the standard attitude vector, the rotation order is selected to obtain the rotation matrix, and then the rotation matrix is ​​decomposed into Euler angles (attitude angles).

[0139] ;

[0140] in, The direction matrix is ​​a 3×3 local coordinate system.

[0141] Euler angles can be calculated using the following formula:

[0142] ;

[0143] In the formula: This refers to the longitudinal direction of bridge steel components, such as the main axis of a steel box girder or hoisting member; This direction is perpendicular to the length of the bridge steel member but parallel to the main plane of the member (such as the web), and is used to form a right-handed coordinate system. The direction of the normal vector of the bridge steel member is the direction perpendicular to the principal plane of the member (the normal vector can be obtained by fitting the plane using the least squares method). The roll angle is the rotation around the X-axis; The pitch angle is rotated around the Y-axis; The yaw angle is the rotation around the Z-axis; Let be the component of the j-th axis in the local coordinate system in the i-th direction in the global coordinate system. This represents the component of the local z-axis in the global y-direction. This represents the component of the local z-axis in the global z-direction. This represents the component of the local z-axis in the global x-direction. This represents the component of the local y-axis in the global x-direction. This represents the component of the local x-axis in the global x-direction.

[0144] Step 4: Based on the 3D laser scanning point cloud of the bridge steel components, construct a 3D virtual model, and use rigid body transformation to synchronously map the 3D virtual model with the motion state and attitude of the bridge steel components during the actual hoisting process;

[0145] Specifically, the BIM model or 3D laser scan point cloud of the bridge steel components is first imported. A lightweight 3D virtual model is then constructed using Revit / Unity. The coordinate system of the virtual model is then set to align with the center of gravity of the actual component, determining the XYZ directions. The positions of the actual lifting points are marked on the 3D virtual model as the application points for subsequent forces and movements. Real-time position and angle information is then input into the 3D virtual model, driving it to undergo rigid body transformation. The rigid body transformation formula is used to achieve synchronous mapping between the physical and virtual postures, thereby simulating the motion state and mechanical properties of the bridge steel components during hoisting. This 3D virtual model allows construction personnel to easily observe positional shifts on computers or tablets and replay previous trajectories.

[0146] When constructing a 3D virtual model, 3D modeling software, such as Autodesk Revit, can be used to perform high-precision geometric modeling of each type of prefabricated component based on the design drawings, geometric dimensions, material properties, and assembly process requirements of the prefabricated bridge components. This allows for precise depiction of the component's outline, connection nodes, embedded parts, and other details, ensuring the accuracy and integrity of the model.

[0147] To achieve posture synchronization between the virtual model and the physical object, the component posture is mapped synchronously using the following rigid body transformation formula:

[0148] ;

[0149] ;

[0150] In the formula: It is a 4×4 pose transformation matrix; It is a 3×3 rotation matrix, obtained from the attitude angles; The geometric center of the component is located by a 3×1 translation vector. To map the location points to the digital twin scene; These are the local coordinate points of the component;

[0151] Rigid body transformation enables synchronous mapping of physical and virtual postures. Furthermore, by introducing a dynamic simulation model, the effects of gravity and inertial forces on component motion can be simulated, ensuring the virtual environment accurately reflects the physical process. Therefore, a dynamic model of the hoisting system is established, considering factors such as gravity, friction, and inertial forces, to simulate the motion state and mechanical characteristics of equipment and components during hoisting, including the dynamic responses to actions such as lifting, translation, rotation, and lowering. By driving the dynamic model with real-time data, the virtual simulation can reflect the changes in the actual hoisting process in real time, achieving synchronization between the virtual and real worlds.

[0152] Step 5: Calculate the position offset and attitude deviation of the bridge steel components. If the position offset or attitude deviation is greater than the threshold, issue an early warning message.

[0153] Optionally, step 5 includes the following sub-steps:

[0154] Step 5.1: Calculate the spatial distance between the current position of the bridge steel component and the target position, as the position offset. ;

[0155] Step 5.2: Calculate the Euclidean difference between the current attitude angle and the target attitude angle of the bridge steel component, as the attitude deviation;

[0156] Specifically, the formula for calculating the Euclidean difference is:

[0157] ;

[0158] in, The current component attitude angle (e.g., Euler angle triplet). To construct the expected attitude angle, This is for attitude deviation;

[0159] Step 5.3: Determine whether the attitude deviation is greater than the attitude angle threshold, or whether the position offset is greater than the offset threshold;

[0160] If the position offset is greater than the offset threshold If the position is offset, a position offset warning message will be issued to indicate that the bridge steel component has shifted beyond the allowable range in spatial position.

[0161] If the attitude deviation is greater than the attitude angle threshold, proceed to step 5.4;

[0162] Step 5.4: If the tilt component in the attitude deviation is greater than the tilt threshold If the pitch component of the attitude deviation is greater than the pitch threshold, a lateral tilt warning will be issued; If the roll angle component in the attitude deviation is greater than the tilt angle threshold, a longitudinal tilt angle warning will be issued; If so, a flip warning message will be issued;

[0163] Optionally, the values ​​of the tilt angle threshold, pitch angle threshold, and offset threshold can be determined according to the type of component.

[0164] Step 6: Determine the adjustment amount of the lifting points based on the early warning information, automatically adjust the position of the lifting points or the length of the slings, and correct the posture of the bridge steel components in real time;

[0165] Specifically, when the system triggers an alarm, the lifting point tension adjustment submodule on crane 4, based on the attitude angle information or position offset information included in the received alarm message, and through the lifting point geometric constraints, can calculate the required lifting point adjustment amount using the following formula:

[0166] ;

[0167] In the formula, This is the output adjustment amount for the suspension point (the scaling length of the sling or the displacement of the suspension point). This refers to the proportional control gain coefficient. Let be the pseudo-inverse Jacobian matrix of the suspension point position, which describes the relationship between attitude and suspension point displacement; This represents the proportion of the center of gravity shift. The centroid position vector of the component; This is the current component's centroid offset vector;

[0168] Finally, based on the lifting point adjustment amount, the crane displacement adjustment component and sling length adjustment component are obtained, and the position of the lifting point and the length of the sling are adjusted respectively.

[0169] Step 7: Synchronously update the Euler angles of the bridge steel components with the virtual model, and obtain the current spatial trajectory and historical status of the bridge steel components in real time based on the virtual model;

[0170] Specifically, throughout the hoisting process, construction personnel can monitor the current spatial trajectory and historical status of the components in real time through a visualization platform. Due to the large data volume of prefabricated components and scene models, to ensure smooth dynamic display, this invention employs Level of Detail (LOD) technology to lightweight the constructed 3D virtual model. While maintaining the model's visual effects and key details, this significantly reduces the model's data volume, improving loading speed and rendering efficiency within the system to meet the performance requirements of real-time dynamic display. Therefore, the visualization platform is developed using Web 3D technology to achieve multi-angle rotation, trajectory tracing, and interactive control. To improve efficiency, LOD optimization can be achieved using the following formula:

[0171] ;

[0172] In the formula: The number of vertices currently displayed; This represents the number of vertices in the original model. This is the line-of-sight function.

[0173] In step 7, the distance between the component and the observation point in the image is calculated first, and then the number of vertices to be displayed is determined according to the above formula, so as to balance smoothness and clear visibility.

[0174] Furthermore, the attitude deviation information, early warning information, lifting point adjustment amount, and 3D laser scanning point cloud data during the hoisting process can be archived and saved; the archived data can be obtained by the following formula:

[0175] ;

[0176] ;

[0177] ;

[0178] in, The integral of the attitude deviation (overall construction accuracy). For alarm frequency and duration, This refers to the attitude deviation during the measurement process. Let be the attitude deviation of the i-th measurement; An alarm is triggered when the actual deviation exceeds a preset attitude deviation threshold. To adjust energy consumption estimates, The change in control quantity over time t is used to describe the change in control force or torque applied to correct deviations; T is the total measurement time; k is the total number of measurements.

[0179] Archived data can support construction quality assessment and digital engineering management, and can be used for subsequent hoisting analysis, construction quality assessment and digital archive management.

[0180] Experimental Case: To verify the effectiveness of the system and method proposed in this invention, the process of hoisting a large precast box girder of a cross-sea bridge of a certain urban railway, in which this invention is applied, is used as an example.

[0181] A new bridge on a certain urban railway crosses a deep-water channel and is being constructed in parallel with an existing highway bridge. The new bridge is located to the east of the old bridge, only about 50 meters away. The construction environment is complex and space is limited, and the precision and safety requirements for offshore hoisting operations are extremely high.

[0182] As a crucial component of the new bridge, the approach bridge section in the deep water area on the north bank requires the high-altitude precision hoisting of large precast box girders. Facing multiple challenges, including strong currents, deep water, and significant interference from construction work on nearby operational bridges, a live-ship hoisting experiment was conducted using the system and method proposed in this invention. The experiment achieved the precise placement of a 60-meter box girder using a large floating crane platform, aiming to verify the system's practicality, safety, and high-precision control capabilities in confined and complex sea areas.

[0183] The experimental principle is as follows:

[0184] 1. High-precision 3D coordinate acquisition

[0185] The BeiDou high-precision positioning module uses real-time differential technology (RTK) to receive signals from multiple satellites and compare them with data from ground reference stations (control points) to achieve dynamic positioning accuracy at the centimeter or even millimeter level. Through continuous sampling and data fitting, the continuously changing three-dimensional coordinates (X, Y, Z) of any monitoring point on a bridge component in space can be obtained.

[0186] 2. Coordinate system transformation and attitude calculation

[0187] To eliminate the deviation between the equipment coordinate system and the project design coordinate system, the system presets a seven-parameter transformation model (3 translations + 3 rotations + 1 scale factor) to transform the real-time measured spatial coordinates to the engineering coordinate system. Combining the coordinate changes of each measuring point, the system uses a geometric attitude analysis algorithm to calculate the translational deviation and attitude angles (tilt, pitch, roll) of the beam in real time, constructing a holographic model of the beam's attitude changes.

[0188] 3. Error Analysis and Closed-Loop Control Mechanism

[0189] The measured coordinates are frequently compared with the design coordinates, and the spatial errors in the front-back (X), left-right (Y), and up-down (Z) directions are calculated in real time. Simultaneously, overall eccentricity and rotation trends are considered to generate multiple maintenance correction commands. These commands are fed back to the lifting equipment through a human-machine interface, enabling precise fine-tuning of the lifting points and forming a spatial control closed loop of "real-time perception - intelligent analysis - closed-loop control".

[0190] Based on the above experimental principles, a 60-meter precast box girder was used as the hoisting component, and a four-point synchronous hoisting method was adopted. The specific hoisting experiment process is as follows:

[0191] First, four measurement control points were selected along the project route to construct a regional benchmark control network. Each control point was located in a stable foundation with excellent visibility, providing reliable benchmark support for subsequent measurement and positioning. Measurement points were set at the four corners of the concrete beams of piers 62-63 of the bridge, at locations 60cm from the long side and 20cm from the short side. A Beidou high-precision positioning module was installed at each measurement point. Specifically, bolts were first firmly bonded to the beam surface using PV adhesive, and after curing, the positioning module was connected and fixed using bolts. This installation method ensured the stability and measurement accuracy of the module while not interfering with subsequent hoisting operations. The original coordinates and target coordinates of the four known measurement control points were input into a coordinate transformation module for accurate conversion between the equipment coordinate system and the project coordinate system.

[0192] Secondly, the lifting operation is carried out based on the crane vessel. After the beam is lifted, the spatial coordinates of each monitoring point are transmitted in real time using a small Beidou RTK positioning module 2. The system automatically receives and displays the current parameter information of the beam, including the design coordinates, real-time coordinates, axis deflection angle, adjustment amount in the up / down / forward / left / right directions, and the front and top views of the beam. The design coordinates and real-time coordinates are shown in Table 1 and Table 2, respectively; the up / down adjustment value, forward / backward adjustment value, and left / right adjustment value are shown in Tables 3-5.

[0193] Table 1. Design coordinates of each measurement control point on the beam:

[0194] ;

[0195] Table 2 Real-time coordinates of various measurement control points on the beam:

[0196] ;

[0197] Table 3. Adjustment values:

[0198] ;

[0199] Table 4. Adjustment values ​​before and after:

[0200] ;

[0201] Table 5. Left and right adjustment values:

[0202] ;

[0203] During the hoisting process, the system continuously receives real-time coordinates from the BeiDou positioning module and compares them frequently with the preset design coordinates. Through precise coordinate difference calculation and spatial vector analysis, the system determines the positional offset of the beam in the X, Y, and Z directions, and accordingly calculates the attitude correction direction (the Euler angles of the bridge steel components) and the specific adjustment amount required for each hoisting point in real time.

[0204] These adjustments are fed back to the on-site operators through a human-machine interface, thereby controlling the hoisting equipment to execute adjustment commands. The adjustment commands not only include forward, backward, left, right, and up-down displacement corrections, but also take into account the overall axis deflection angle and tilting trend, achieving comprehensive control over the spatial attitude of the beam.

[0205] As the lifting points are continuously fine-tuned, the beam gradually approaches the designed position. The system updates the deviation in real time and adjusts the commands synchronously, forming a closed-loop control mechanism. The lifting operation ends when the deviations of the beam in all directions converge to the set minimum allowable range. All measurement data during the lifting process are recorded, including the coordinates and timestamps of each measuring point. The measured coordinates are compared and analyzed with the design coordinates to calculate the planar position error and elevation error. The final calculation results are as follows:

[0206] Planar position error: Front-to-back error is 4cm; left-to-right error is 4cm.

[0207] Elevation error (including a 5cm pad and a 2cm connector for the BeiDou high-precision module): 2.55cm.

[0208] A comprehensive analysis from three dimensions—mean error, maximum deviation, and spatial stability—showed no significant jumps or drifts in the data at each monitoring point during the experiment. The system exhibited a high response frequency and stable data transmission, ensuring millimeter-level dynamic correction control during component attitude adjustment. Even under complex construction environments such as interference from adjacent bridges and wind and wave disturbances, the system maintained high measurement accuracy and stability, fully demonstrating its excellent anti-interference capability and engineering adaptability.

[0209] The final results demonstrate that this system can successfully achieve millimeter- to centimeter-level precision control in the installation of large precast box girders. Specifically, after fully considering systematic factors such as 5cm spacers and approximately 2cm connectors, the elevation error was consistently controlled within 2.55cm, and the average planar position error was approximately 4cm. The overall hoisting accuracy far surpasses traditional empirical construction standards, fully meeting the high-precision requirements for precast component installation in the "Railway Bridge and Culvert Construction Quality Acceptance Standard." This provides reliable technical support and practical basis for future application in similar projects.

[0210] Although preferred embodiments of the present application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the embodiments of the present application.

[0211] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal device. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal device that includes said element.

[0212] The above provides a detailed description of the BeiDou digital twin system and method for precision hoisting of bridge steel components provided in this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A Beidou digital twin system for precision hoisting of bridge steel components, characterized in that, The method comprises the following steps: A multi-source data acquisition module is used to acquire the degree of freedom data, attitude information and movement speed of the bridge steel member in three-dimensional space in real time; A coordinate conversion module is used to convert the acquired degree of freedom data from the geodetic coordinate system to the local coordinate system; An attitude solving module is used to calculate the standard attitude vector by using a comprehensive deep learning model according to the degree of freedom data, attitude information and movement speed in the local coordinate system, and to solve the attitude by using the standard attitude vector to obtain the attitude angle of the bridge steel member; A three-dimensional virtual model construction module is used to establish a three-dimensional virtual model capable of reflecting the actual hoisting process changes of the bridge steel member in real time based on the three-dimensional laser scanning point cloud of the bridge steel member; A hoisting movement state monitoring module is used to determine the attitude deviation of the bridge steel member according to the attitude angle and to issue a warning information when an abnormality occurs, to determine the lifting point adjustment amount based on the warning information, and to adjust the attitude of the bridge steel member; A digital twin module is used to update the attitude information and the three-dimensional virtual model of the bridge steel member synchronously during the entire hoisting process of the bridge steel member, and to obtain the current spatial trajectory and the historical state of the bridge steel member based on the three-dimensional virtual model in real time; An archiving module is used to record and save the attitude deviation, the warning information, the lifting point adjustment amount and the three-dimensional laser scanning point cloud data of the bridge steel member.

2. The system of claim 1, wherein, The hoisting movement state monitoring module comprises: A warning sub-module is used to determine whether the hoisted bridge steel member is abnormal in attitude or abnormal in spatial position deviation based on the attitude deviation of the bridge steel member obtained from the attitude angle, and to issue a warning information when an abnormality occurs; A lifting point tension adjustment sub-module is used to determine the lifting point adjustment amount according to the received warning information, to adjust the lifting point position or the length of the lifting rope of the crane, and to correct the attitude of the bridge steel member in real time.

3. The system of claim 2, wherein, The standard attitude vector is calculated by using a comprehensive deep learning model according to the degree of freedom data, attitude information and movement speed in the local coordinate system, and the attitude is solved based on the standard attitude vector to obtain the attitude angle of the bridge steel member, which comprises: The degree of freedom data, attitude information and movement speed in the local coordinate system are combined to form a state vector; The state vector is input into the trained comprehensive deep learning model to obtain the standard attitude vector; The attitude is solved based on the standard attitude vector to obtain the current attitude angle of the bridge steel member.

4. The system of claim 3, wherein, The comprehensive deep learning model comprises a Transformer network module, a multi-layer perception module and an output layer, and the output layer comprises three output nodes, namely a position output node, a speed output node and an attitude output node.

5. The system of claim 4, wherein, The establishment process of the comprehensive deep learning model comprises: The input sample data is standardized, and the standard attitude vector at the next time step is used as the label of the sample data; The connection relationship between each neuron in the MLP module and the connection relationship between each neuron in the MLP module and the position output node, the speed output node and the attitude output node are pruned based on a first pruning hyperparameter to obtain a second comprehensive deep learning model; wherein the first pruning hyperparameter is the pruning hyperparameter obtained after the last model training of the first comprehensive deep learning model. based on the first repair hyperparameters, repairing connection relationships between neurons inside an MLP module of the second integrated deep learning model and between each neuron inside the MLP module and a position output node, a speed output node, and a pose output node, to obtain a third integrated deep learning model; wherein the first repair hyperparameters are repair hyperparameters obtained by the first integrated deep learning model after a previous model training; inputting the standardized sample data into the third integrated deep learning model to obtain a predicted pose vector at a next time step; calculating a model loss value of the third integrated deep learning model according to the predicted pose vector and a standard pose vector; adjusting model parameters and hyperparameters of the first integrated deep learning model through the model loss value to obtain a trained integrated deep learning model.

6. The system of claim 5, wherein, The intelligent data processing center comprises: a data processing center configured to receive multi-source data and early warning information collected by the multi-source data acquisition module, clean, fuse, analyze, and store the multi-source data, eliminate invalid data and interference information, extract feature parameters, and analyze the early warning information to send a hoisting point adjustment amount obtained to the archiving module; a model operation center configured to drive a three-dimensional virtual model to be dynamically updated according to the feature parameters, and realize mapping of the three-dimensional virtual model and an actual hoisting state.

7. A method for hoisting a bridge steel member, characterized by, The method comprises: obtaining, in real time, degree of freedom data, pose information, and movement speed of a bridge steel member in a three-dimensional space; converting the collected degree of freedom data from a geodetic coordinate system to a local coordinate system; obtaining a standard pose vector based on the degree of freedom data in the local coordinate system, the pose information, and the movement speed, and performing pose calculation based on the standard pose vector to obtain a current pose angle of the bridge steel member; constructing a three-dimensional virtual model based on three-dimensional laser scanning point clouds of the bridge steel member, and performing synchronous mapping of a movement state and a pose of the bridge steel member in an actual hoisting process and the three-dimensional virtual model through rigid body transformation; calculating a position offset amount and a pose deviation of the bridge steel member, and issuing early warning information if the position offset amount or the pose deviation is greater than a threshold value; determining a hoisting point adjustment amount based on the early warning information, adjusting a hoisting point position or a sling length of a crane, and correcting a pose of the bridge steel member in real time; synchronously updating the pose information of the bridge steel member and the three-dimensional virtual model, and obtaining a current spatial trajectory and a historical state of the bridge steel member in real time based on the three-dimensional virtual model.

8. The method of claim 7, wherein, calculating a position offset amount and a pose deviation of the bridge steel member, and issuing early warning information if the position offset amount or the pose deviation is greater than a threshold value, comprising: calculating a spatial distance between a current position of the bridge steel member and a target position as the position offset amount; calculating an Euclidean difference value between a current pose angle of the bridge steel member and a target pose angle as the pose deviation; determining whether the pose deviation is greater than a pose angle threshold value or whether the position offset amount is greater than an offset amount threshold value; if the position offset amount is greater than the offset amount threshold value, issuing position offset early warning information; and if the pose deviation is greater than the pose angle threshold value, proceeding to the next step. If the inclination component in the attitude deviation is greater than the inclination threshold, a lateral inclination warning information is sent; if the pitch component in the attitude deviation is greater than the pitch threshold, a longitudinal inclination warning information is sent; if the roll component in the attitude deviation is greater than the inclination threshold, a roll-over warning information is sent.

9. The method of claim 8, wherein, According to the warning information, the adjustment amount of the lifting point is determined, and the position of the lifting point or the length of the sling is automatically adjusted, including: According to the lifting point position and attitude information in the warning information, the adjustment amount of the lifting point is calculated; Based on the adjustment amount of the lifting point, the displacement adjustment component and the sling length adjustment component are obtained; Based on the displacement adjustment component, the position of the lifting point is adjusted; Based on the sling length adjustment component, the length of the sling is adjusted.

10. The method of claim 9, wherein, The calculation formula of the adjustment amount of the lifting point is: ； wherein, is the output hoist point adjustment amount; is the proportional control gain coefficient; is the pseudo-inverse Jacobian matrix of the hoist point arrangement, describing the relationship between the pose and the hoist point displacement; is the center of gravity offset proportion, is the component center of gravity position vector, is the current component center of gravity offset vector.

Images