Intelligent Construction Control Method and System for Girder Erection Crane Based on Machine Vision

Abstract

The present invention provides an intelligent construction control method and system for a bridge erecting machine based on machine vision, which relates to the field of intelligent control technology. Image information of the construction site is collected by a plurality of high-definition cameras arranged on the bridge erecting machine; the image information is input into a pre-trained deep learning model to identify and locate the targets in the construction scene, and combined with the graph structure of the construction scene topology after spatial-aware graph convolution operation, a digital twin model of the construction scene is constructed; the intelligent decision-making agent generates an initial construction strategy according to the current construction scene state, optimizes the initial construction strategy by using an improved Monte Carlo tree search, and generates an optimal construction strategy through multi-objective solution; after the construction is completed, based on the digital twin model and the whole-process data, a graph neural network is used to analyze the construction quality and efficiency, and a multi-dimensional evaluation report and optimization suggestions are generated.

Description

Technical Field

[0001] The present invention relates to intelligent control technology, and particularly to an intelligent construction control method and system for a bridge erecting machine based on machine vision. Background Art

[0002] Traditional control of bridge erecting machines mainly relies on manual operation and simple mechanical control systems, with the following limitations:

[0003] Firstly, it is difficult to guarantee the accuracy and consistency of manual operation. In a complex construction environment, operators need to pay attention to multiple parameters and targets simultaneously, which easily leads to operation errors or judgment deviations. Especially in the case of long-term continuous operation, the fatigue of operators will further increase the risk of errors.

[0004] Secondly, traditional mechanical control systems lack the ability to perceive the construction environment in real time. The environment faced by a bridge erecting machine during construction is dynamically changing, including weather conditions, terrain changes, obstacles at the construction site, etc. Simple mechanical control systems are difficult to adapt to these changes and cannot make timely adjustments and responses.

[0005] Thirdly, it is difficult to achieve high-precision positioning and attitude control with traditional methods. In the construction of large bridges, precise positioning and attitude control are crucial for ensuring the correct installation of components. However, relying solely on manual measurement and simple mechanical control, it is difficult to meet the high-precision standards required by modern bridge engineering. Summary of the Invention

[0006] Embodiments of the present invention provide an intelligent construction control method and system for a bridge erecting machine based on machine vision, which can solve the problems in the prior art.

[0007] In the first aspect of the embodiments of the present invention,

[0008] An intelligent construction control method for a bridge erecting machine based on machine vision is provided, including:

[0009] Collecting image information of the construction site through multiple high-definition cameras arranged on the bridge erecting machine; inputting the image information into a pre-trained deep learning model to identify and locate targets in the construction scene, including various components of the bridge erecting machine, bridge components, and the surrounding environment; based on the recognition results of the targets in the construction scene, combining with the graph structure of the topological structure of the construction scene after spatial perception graph convolution operation, constructing a digital twin model of the construction scene;

[0010] Using a reinforcement learning algorithm, an intelligent decision-making agent is trained based on a digital twin model and a predefined reward and punishment function; the intelligent decision-making agent generates an initial construction strategy according to the current construction scene state, optimizes the initial construction strategy using an improved Monte Carlo tree search, and generates an optimal construction strategy through multi-objective solution; the optimal construction strategy is converted into intelligent control instructions, and the intelligent control instructions are sent to the control system of the bridge erecting machine to achieve precise control of each actuator of the bridge erecting machine; during the construction process, data of various sensors of the bridge erecting machine are continuously collected, the data of each sensor and the feedback data of the actuator are fused to generate a state estimation of the bridge erecting machine; based on the state estimation, a nonlinear model predictive control algorithm is used to perform closed-loop control on the bridge erecting machine;

[0011] After the construction is completed, based on the digital twin model and the whole-process data, a graph neural network is used to analyze the construction quality and efficiency, and a multi-dimensional evaluation report and optimization suggestions are generated.

[0012] In an alternative embodiment,

[0013] Collecting image information of the construction site through multiple high-definition cameras arranged on the bridge erecting machine; inputting the image information into a pre-trained deep learning model to identify and locate the targets in the construction scene, including the steps of each component of the bridge erecting machine, bridge components and the surrounding environment:

[0014] Collecting image information of the construction site through multiple high-definition cameras arranged on the bridge erecting machine; performing adaptive histogram equalization preprocessing on the image information; inputting the preprocessed image information into a pre-trained improved MaskR-CNN deep learning model, and a spatial pyramid pooling module is introduced into the feature extraction network of the improved Mask R-CNN deep learning model;

[0015] An attention mechanism module is added after the region of interest alignment layer of the improved Mask R-CNN deep learning model, and the feature is highlighted and the irrelevant background information is suppressed by generating a spatial attention map and a channel attention map;

[0016] Using the improved Mask R-CNN deep learning model to identify and locate the targets in the construction scene, including each component of the bridge erecting machine, bridge components and the surrounding environment;

[0017] Comparing the recognition results of the targets in the construction scene with a dedicated data set that pre-builds the components of the bridge erecting machine and bridge components; using the focal loss function as the loss function during the training process of the improved Mask R-CNN deep learning model, and introducing an online case mining technology to dynamically select difficult samples for training;

[0018] Perform quantization and pruning operations on the trained improved Mask R-CNN deep learning model, converting the 32-bit floating-point numbers in the model into 8-bit integers; use the processed improved Mask R-CNN deep learning model to perform real-time object detection and recognition on the construction scene, and output the category, location, and contour information of the object; transmit the detection and recognition results to the control system of the bridge erector.

[0019] In an alternative embodiment,

[0020] Based on the recognition results of the objects in the construction scene, combined with the graph structure of the construction scene topology structure after spatial-aware graph convolution operations, the steps to construct a digital twin model of the construction scene include:

[0021] Construct a graph structure representing the construction scene topology, where the nodes of the graph structure represent the recognized objects or elements in the preset model, and the edges of the graph structure represent the spatial relationships between the nodes;

[0022] Based on the constructed graph structure, initialize the node feature matrix;

[0023] Apply spatial-aware graph convolution operations to the graph structure, and the spatial-aware graph convolution operations include:

[0024] Calculate the Euclidean distance between each pair of nodes in the graph structure;

[0025] Based on the calculated Euclidean distance, use the Gaussian kernel function to calculate the spatial distance weights between the nodes, obtaining the spatial distance weight matrix;

[0026] Construct the adjacency matrix of the graph structure and add self-connections to the adjacency matrix;

[0027] Calculate the degree matrix of the adjacency matrix after adding self-connections;

[0028] Multiply the spatial distance weight matrix by the adjacency matrix to obtain the weighted adjacency matrix;

[0029] Apply the weighted adjacency matrix to the node feature matrix and multiply it by the learnable weight matrix;

[0030] Apply a non-linear activation function to the product result to obtain the updated node feature matrix;

[0031] Based on the updated node feature matrix, perform feature extraction on the two-dimensional object recognition results and three-dimensional model data in the construction scene;

[0032] Input the extracted two-dimensional and three-dimensional features into the multi-modal feature fusion module, and the multi-modal feature fusion module uses a fully connected layer to process the features and calculates the fusion weights through an attention mechanism;

[0033] Perform weighted fusion on the processed features according to the fusion weights to obtain fused features;

[0034] Based on the fused features and the graph structure obtained through spatial-aware graph convolution operations, construct a digital twin model of the construction scenario.

[0035] In an alternative embodiment,

[0036] The steps for the intelligent decision-making agent to generate an initial construction strategy according to the current construction scenario state, optimize the initial construction strategy using an improved Monte Carlo tree search, and generate an optimal construction strategy through multi-objective solution include:

[0037] Use the intelligent decision-making agent to generate an initial construction strategy based on the current construction scenario state;

[0038] Optimize the initial construction strategy using an improved Monte Carlo tree search to generate an optimization result; the improved Monte Carlo tree search includes introducing a dynamic action generator, using the UCT algorithm with adaptive exploration parameters, implementing tree parallelization, and applying heuristic pruning;

[0039] In the UCT algorithm with the adaptive exploration parameter, introduce a dynamically adjusted exploration parameter c:

[0040] c = c_base * (1 + α * exp(-β * t / T));

[0041] where c_base is the base exploration parameter, α and β are adjustment factors, t is the current iteration number, and T is the total iteration number;

[0042] Perform multi-objective solution on the optimization result of the initial construction strategy, including defining objective functions for construction efficiency and energy consumption, and performing multi-objective optimization using an improved NSGA-III algorithm. The improved NSGA-III algorithm includes adaptive crossover and mutation, local search, and reference point adaptation; generate an optimal construction strategy according to the multi-objective optimization solution result.

[0043] In an alternative embodiment,

[0044] The steps for the improved NSGA-III algorithm to include adaptive crossover and mutation, local search, and reference point adaptation include:

[0045] Introduce an adaptive crossover and mutation mechanism based on population diversity, dynamically adjust the crossover probability and mutation probability according to the diversity metric of the current population. The diversity metric is calculated using the entropy weight method, and increase the intensity of genetic operations when the population diversity decreases;

[0046] After each generation of iteration of the NSGA-III algorithm, evaluate the objective function values of all individuals in the current population, identify and construct the non-dominated solution set, select multiple solutions from the non-dominated solution set for local search, perform multiple simulated annealing iterations on each selected solution, the simulated annealing iteration adopts an exponentially decaying temperature schedule, generate neighborhood solutions in each iteration and calculate the acceptance probability according to the energy difference, if the solution obtained by local search dominates the original solution, replace the original solution with the new solution;

[0047] Adopt a reference point adaptive mechanism, including uniformly generating initial reference points in the standardized objective space, periodically evaluating the association degree of each reference point during the evolution process, using the density peak clustering algorithm to identify high-density regions in the objective space for low-association reference points, moving the low-association reference points to the nearest high-density region and maintaining dispersion, and at the same time introducing a regularization term to limit the large-scale migration of reference points.

[0048] In an optional embodiment,

[0049] During the construction process, continuously collect the data of various sensors of the bridge erecting machine, fuse the data of various sensors and the feedback data of the actuators to generate a state estimate of the bridge erecting machine; the steps of performing closed-loop control on the bridge erecting machine based on the state estimate using a non-linear model predictive control algorithm include:

[0050] Use an inertial measurement unit, a global positioning system, a laser rangefinder, a strain sensor and an actuator encoder to collect the state information of the bridge erecting machine, and the state information includes acceleration, angular velocity, position, relative distance, structural stress, hydraulic cylinder position and motor speed;

[0051] Construct a state vector including position, acceleration, angular velocity and force / moment, and establish a discrete-time non-linear state equation and an observation equation based on the state vector;

[0052] Use an extended Kalman filter to perform state estimation on the discrete-time non-linear state equation and the observation equation, including performing a prediction step and an update step, where the prediction step includes state prediction and covariance prediction, and the update step includes calculating the Kalman gain, state update and covariance update;

[0053] Based on the state estimation result, construct a non-linear model predictive control optimization problem, and the optimization objectives include tracking a predetermined trajectory, minimizing energy consumption and satisfying safety constraints;

[0054] Adopt a real-time iteration strategy to solve the non-linear model predictive control optimization problem, and divide the solution process into a preparation stage and a feedback stage, where the preparation stage includes linearizing the system dynamics and constructing a quadratic programming sub-problem, and the feedback stage includes updating the initial state, solving the quadratic programming sub-problem and applying the control input;

[0055] Based on the solution result of the nonlinear model predictive control, an optimal control input is generated, and the optimal control input is converted into a control signal for the execution mechanism of the bridge erecting machine to perform closed-loop control on the bridge erecting machine.

[0056] In an alternative embodiment,

[0057] After the construction is completed, based on the digital twin model and the whole-process data, the steps of using the graph neural network to analyze the construction quality and efficiency and generate a multi-dimensional evaluation report and optimization suggestions include:

[0058] Represent the bridge erecting machine and the bridge structure as a heterogeneous graph, where the nodes represent components or monitoring points, the edges represent the physical connections or logical relationships between components, and each node and edge are attached with multi-dimensional feature vectors; apply a dedicated feature transformation matrix to the multi-dimensional feature vectors of different types of nodes and edges to generate the initial node and edge representations;

[0059] Based on the initial node and edge representations, construct an improved graph neural network model, including a node-level attention mechanism and an edge-level attention mechanism, where the node-level attention is used to learn the importance weights between different features, and the edge-level attention is used to learn the influence degree of different neighbor nodes;

[0060] Introduce a temporal convolutional network after the graph convolutional layer of the graph neural network model, use causal convolution and dilated convolution to capture temporal features, and fuse them with the output of the graph convolutional layer;

[0061] Use the historical data of the bridge erecting machine construction system to train the graph neural network model, and iterate the training until convergence;

[0062] Use the trained graph neural network model to perform quality and efficiency analysis on the new bridge erecting machine construction system data to obtain the prediction results of various indicators; based on the prediction results, generate a multi-dimensional evaluation report including the overall score, analysis of various indicators, identification of abnormal events and their root cause analysis; according to the problems and potential risks identified in the multi-dimensional evaluation report, combined with historical construction experience, generate optimization suggestions and improvement measures.

[0063] In the second aspect of the embodiments of the present invention,

[0064] Provide an intelligent construction control system for a bridge erecting machine based on machine vision, including:

[0065] The first unit is used to collect image information of the construction site through a plurality of high-definition cameras arranged on the bridge erecting machine; input the image information into a pre-trained deep learning model to identify and locate the targets in the construction scene, including various components of the bridge erecting machine, bridge components and the surrounding environment; based on the recognition results of the targets in the construction scene, combined with the graph structure of the construction scene topology structure after spatial-aware graph convolution operation, construct a digital twin model of the construction scene;

[0066] The second unit is used to train an intelligent decision-making agent based on a digital twin model and a predefined reward and punishment function by using a reinforcement learning algorithm; the intelligent decision-making agent generates an initial construction strategy according to the current construction scene state, optimizes the initial construction strategy by using an improved Monte Carlo tree search, and generates an optimal construction strategy through multi-objective solution; converts the optimal construction strategy into intelligent control instructions, and sends the intelligent control instructions to the control system of the bridge erecting machine to achieve precise control of each actuator of the bridge erecting machine; during the construction process, continuously collect the data of each sensor of the bridge erecting machine, fuse the data of each sensor and the feedback data of the actuator to generate a state estimation of the bridge erecting machine; based on the state estimation, use a non-linear model predictive control algorithm to perform closed-loop control on the bridge erecting machine;

[0067] The third unit is used to, after the construction is completed, analyze the construction quality and efficiency by using a graph neural network based on the digital twin model and the whole-process data, and generate a multi-dimensional evaluation report and optimization suggestions.

[0068] In the third aspect of the embodiments of the present invention,

[0069] A kind of electronic device is provided, including:

[0070] A processor;

[0071] A memory for storing instructions executable by the processor;

[0072] Wherein, the processor is configured to call the instructions stored in the memory to execute the method described above.

[0073] In the fourth aspect of the embodiments of the present invention,

[0074] A computer-readable storage medium is provided, on which computer program instructions are stored, and when the computer program instructions are executed by a processor, the method described above is implemented.

[0075] The construction strategy generation method based on an intelligent decision-making agent, improved Monte Carlo tree search and multi-objective optimization in this article has many beneficial effects. First of all, it can automatically generate high-quality construction strategies according to the complex construction scene state, greatly reducing the burden of manual decision-making. Secondly, the improved MCTS significantly improves the efficiency and quality of strategy optimization through technologies such as dynamic action generation, adaptive exploration and parallelization. Moreover, multi-objective optimization considers multiple objectives such as construction efficiency and energy consumption, making the generated strategies more comprehensive and practical. Finally, the adaptability and flexibility of this method enable it to adapt to various complex construction environments and requirements, providing strong technical support for the intelligent and refined construction management of bridge erecting machines.

[0076] The method for analyzing the construction quality and efficiency of bridge erecting machines based on graph neural networks adopted in this paper has beneficial effects in many aspects. First of all, it can comprehensively consider the complex topological relationships between bridge erecting machines and bridge structures, capture the mutual influences between components, and provide more accurate analysis results than traditional methods. Secondly, the introduced attention mechanism and temporal convolutional network enable the model to adaptively focus on important features and long-term dependencies, improving the accuracy and robustness of the analysis. Moreover, this method can achieve end-to-end learning, directly obtaining evaluation results from raw data and reducing the interference of human factors. Finally, the generated multi-dimensional evaluation reports and optimization suggestions provide intuitive and comprehensive decision-making support for construction management personnel, helping to improve construction quality, efficiency, and safety. The application of this method can not only optimize current construction projects, but also continuously improve construction technologies and management strategies by accumulating and analyzing a large amount of data, promoting the technological progress of the entire bridge construction industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0077] Figure 1 It is a schematic flowchart of the intelligent construction control method for bridge erecting machines based on machine vision according to an embodiment of the present invention;

[0078] Figure 2 It is a schematic structural diagram of the bridge erecting machine according to an embodiment of the present invention. In the figure,

[0079] 1. Right end beam assembly; 2. Front support system; 3. Transport vehicle; 4. Cross beam; 5. Camera; 6. Trolley system; 8. Middle support; 9. Auxiliary support; 10. Front support system; 11. Left end beam assembly; 12. Main beam;

[0080] Figure 3 It is a schematic structural diagram of the intelligent construction control system for bridge erecting machines based on machine vision according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0081] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0082] The technical solutions of the present invention will be described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments.

[0083] Figure 1The flowchart of the intelligent construction control method for a bridge erecting machine based on machine vision according to an embodiment of the present invention is shown as Figure 1 follows. The method includes:

[0084] S1. Collect image information of the construction site through multiple high-definition cameras installed on the bridge erecting machine; input the image information into a pre-trained deep learning model to identify and locate the targets in the construction scene, including various components of the bridge erecting machine, bridge components, and the surrounding environment; based on the recognition results of the targets in the construction scene, combine with the graph structure of the construction scene topology after spatial perception graph convolution operation to construct a digital twin model of the construction scene;

[0085] S2. Use the reinforcement learning algorithm to train an intelligent decision-making agent based on the digital twin model and a predefined reward and punishment function; the intelligent decision-making agent generates an initial construction strategy according to the current construction scene state, optimizes the initial construction strategy using an improved Monte Carlo tree search, and generates an optimal construction strategy through multi-objective solution; convert the optimal construction strategy into intelligent control instructions, and send the intelligent control instructions to the control system of the bridge erecting machine to achieve precise control of each actuator of the bridge erecting machine; during the construction process, continuously collect data of various sensors of the bridge erecting machine, fuse the data of various sensors and the feedback data of the actuators to generate a state estimation of the bridge erecting machine; based on the state estimation, use the non-linear model predictive control algorithm to perform closed-loop control on the bridge erecting machine;

[0086] S3. After the construction is completed, analyze the construction quality and efficiency based on the digital twin model and the whole-process data, and generate a multi-dimensional evaluation report and optimization suggestions.

[0087] In an alternative embodiment,

[0088] The steps of collecting image information of the construction site through multiple high-definition cameras installed on the bridge erecting machine; inputting the image information into a pre-trained deep learning model to identify and locate the targets in the construction scene, including various components of the bridge erecting machine, bridge components, and the surrounding environment include:

[0089] Collect image information of the construction site through multiple high-definition cameras installed on the bridge erecting machine; perform adaptive histogram equalization preprocessing on the image information; input the preprocessed image information into a pre-trained improved Mask R-CNN deep learning model, and introduce a spatial pyramid pooling module into the feature extraction network of the improved Mask R-CNN deep learning model;

[0090] Add an attention mechanism module after the region of interest alignment layer of the improved Mask R-CNN deep learning model to highlight features and suppress irrelevant background information by generating a spatial attention map and a channel attention map;

[0091] Use the improved Mask R-CNN deep learning model to identify and locate the targets in the construction scene, including the components of the bridge erection machine, bridge components, and the surrounding environment;

[0092] Compare the recognition results of the targets in the construction scene with a pre-constructed dedicated dataset containing the components of the bridge erection machine and bridge components; During the training process of the improved Mask R-CNN deep learning model, use the focal loss function as the loss function, and introduce the online case mining technology to dynamically select difficult samples for training;

[0093] Perform quantization and pruning operations on the trained improved Mask R-CNN deep learning model, converting the 32-bit floating-point numbers in the model into 8-bit integers; Use the processed improved Mask R-CNN deep learning model to perform real-time target detection and recognition on the construction scene, and output the category, location, and contour information of the targets; Transmit the detection and recognition results to the control system of the bridge erection machine.

[0094] Exemplarily, the method for identifying the targets in the bridge erection machine construction scene based on the improved Mask R-CNN is an advanced computer vision technology, aiming to improve the safety and efficiency in the bridge erection machine construction process. This method realizes high-precision target detection and recognition through multiple steps.

[0095] First, install multiple high-definition cameras on the bridge erection machine to ensure full coverage of the construction site. These cameras usually adopt a 4K resolution and a frame rate of not less than 30fps to capture high-quality image information. The installation positions of the cameras include but are not limited to key parts such as the main beam, cross beam, and lifting tackle to obtain images with different angles and fields of view.

[0096] The collected original images are preprocessed by adaptive histogram equalization. This step is crucial for processing images under complex lighting conditions. Specifically, the image is divided into 8x8 small blocks, and histogram equalization is performed on each small block separately, and then the results of adjacent blocks are merged using the bilinear interpolation method. This method can effectively enhance the local contrast, enabling the targets in the shadow area or strong light area to be clearly recognized.

[0097] The preprocessed images are input into the improved Mask R-CNN deep learning model. The main network of this model adopts the ResNet-101 architecture, and a spatial pyramid pooling module is introduced on this basis. The spatial pyramid pooling module contains max pooling operations of four scales: 1x1, 2x2, 4x4, and 8x8. Each pooling is followed by a 1x1 convolutional layer, and finally all the outputs are concatenated. This design can effectively capture multi-scale features and improve the detection ability of the model for targets of different sizes.

[0098] After aligning the layers in the region of interest, an attention mechanism module is added. The spatial attention mechanism generates a 2D attention map to highlight important spatial location information. Specifically, when implementing, average pooling and max pooling are performed on the feature map, and then the results are concatenated and passed through a 7x7 convolutional layer to obtain the spatial attention map. The channel attention mechanism generates a 1D vector to emphasize important feature channels. When implementing, global average pooling and global max pooling are performed on the feature map, and then the channel attention vector is obtained through two fully connected layers. The combination of these two attention mechanisms can effectively highlight important features and suppress irrelevant background information.

[0099] During the training process of the model, the focal loss function is used as the loss function. The focal loss function solves the class imbalance problem by adjusting the weights of easy and hard samples. In practical applications, the alpha parameter is set to 0.25 and the gamma parameter is set to 2.0. In addition, online case mining technology is introduced to dynamically select hard samples for training. Specifically, in each mini-batch, the losses of all samples are calculated, and the top 33% samples with the highest losses are selected as cases, and only these samples are used to update the model parameters.

[0100] After training is completed, quantization and pruning operations are performed on the model. The quantization process converts 32-bit floating-point numbers to 8-bit integers. The specific steps include determining the quantization range, calculating quantization parameters, quantizing weights and activation values, and re-calibrating.

[0101] The optimized model is used for real-time object detection and recognition in the construction scene. The model output includes the category of the object (such as main beam, cross beam, spreader, etc.), the location (represented by bounding box coordinates), and the contour information (represented by a pixel-level mask). This information is transmitted through a high-speed network to the control system of the bridge erecting machine to achieve real-time monitoring and auxiliary decision-making.

[0102] This article significantly improves the safety of the bridge erecting machine construction process through high-precision object detection and recognition, and can timely detect potential dangers and give early warnings. Secondly, real-time monitoring and analysis improve construction efficiency and optimize the process flow. Thirdly, the optimization of the model enables it to run on an embedded system, realizing edge computing and reducing data transmission latency.

[0103] In an alternative embodiment,

[0104] Based on the recognition results of the objects in the construction scene, combined with the graph structure of the construction scene topology after spatial-aware graph convolution operation, the steps of constructing a digital twin model of the construction scene include:

[0105] Construct a graph structure representing the construction scene topology, where the nodes of the graph structure represent the recognized objects or elements in the preset model, and the edges of the graph structure represent the spatial relationships between the nodes;

[0106] Initialize the node feature matrix based on the constructed graph structure;

[0107] Apply a spatial-aware graph convolution operation to the graph structure, where the spatial-aware graph convolution operation includes:

[0108] Calculate the Euclidean distance between each pair of nodes in the graph structure;

[0109] Based on the calculated Euclidean distance, use the Gaussian kernel function to calculate the spatial distance weights between nodes, obtaining a spatial distance weight matrix;

[0110] Construct the adjacency matrix of the graph structure and add self-connections to the adjacency matrix;

[0111] Calculate the degree matrix of the adjacency matrix after adding self-connections;

[0112] Multiply the spatial distance weight matrix by the adjacency matrix to obtain a weighted adjacency matrix;

[0113] Apply the weighted adjacency matrix to the node feature matrix and multiply it by a learnable weight matrix;

[0114] Apply a non-linear activation function to the product result to obtain an updated node feature matrix;

[0115] Based on the updated node feature matrix, perform feature extraction on the 2D object recognition results and 3D model data in the construction scenario;

[0116] Input the extracted 2D and 3D features into a multi-modal feature fusion module, where the multi-modal feature fusion module uses a fully connected layer to process the features and calculates the fusion weights through an attention mechanism;

[0117] Perform weighted fusion on the processed features according to the fusion weights to obtain fused features;

[0118] Based on the fused features and the graph structure that has undergone the spatial-aware graph convolution operation, construct a digital twin model of the construction scenario.

[0119] Exemplarily, the method for constructing a digital twin model of a construction scenario based on object recognition results and spatial-aware graph convolution is an innovative technical solution aimed at accurately restoring a complex bridge erection machine construction environment. This method integrates advanced technologies such as computer vision, graph neural networks, and multi-modal feature fusion, achieving an efficient conversion from 2D images to 3D digital twin models.

[0120] First, construct a graph structure representing the topological structure of the construction scene. The nodes in the graph structure represent the identified targets or elements in the preset model, such as main beams, cross beams, support frames, etc. The edges of the graph structure represent the spatial relationships between the nodes, such as adjacent, connected, or supported. For example, in a typical bridge erecting machine construction scene, there may be 20 nodes, respectively representing the main structural components and surrounding environmental elements.

[0121] Next, initialize the node feature matrix. The initial features of each node can include information such as its position coordinates, dimensions, category, etc. For a graph structure with 20 nodes, the dimension of the initial feature matrix may be 20x64, where 64 is the feature dimension of each node.

[0122] Then, apply the spatial-aware graph convolution operation to the graph structure. The core of this operation is to consider the spatial distance information between nodes, so as to more accurately capture the mutual relationships of various elements in the construction scene. In specific implementation, first calculate the Euclidean distance between each pair of nodes in the graph structure. For example, for two nodes with three-dimensional coordinates (1, 2, 3) and (4, 5, 6) respectively, the Euclidean distance between them is approximately 5.2.

[0123] Based on the calculated Euclidean distance, use the Gaussian kernel function to calculate the spatial distance weights between nodes. The Gaussian kernel function is in the form of an exponential function, where the distance is the negative part of the exponent, and the sigma parameter controls the decay rate of the function. Usually, sigma is set to half of the average node distance. This step obtains the spatial distance weight matrix, and the matrix size is 20x20 (assuming there are 20 nodes).

[0124] Then construct the adjacency matrix of the graph structure and add self-connections to the adjacency matrix. The adjacency matrix is a 20x20 matrix (assuming there are 20 nodes), where the element is 1 indicating that two nodes are connected, and 0 indicating not connected. Adding self-connections means setting the diagonal elements of the adjacency matrix to 1.

[0125] Calculate the degree matrix of the adjacency matrix after adding self-connections. The degree matrix is a diagonal matrix, and the elements on the diagonal represent the degrees of the corresponding nodes (the number of edges connected to them). For example, if a certain node is connected to 3 other nodes, then the corresponding diagonal element in the degree matrix is 4 (including self-connections).

[0126] Multiply the spatial distance weight matrix by the adjacency matrix to obtain the weighted adjacency matrix. This step introduces the spatial distance information into the graph structure, making the connection weights between nodes that are closer in space larger.

[0127] Apply the weighted adjacency matrix to the node feature matrix and multiply it with the learnable weight matrix. The size of the learnable weight matrix depends on the input feature dimension and the desired output feature dimension. For example, it may be a 64x128 matrix, which maps the node features from 64 dimensions to 128 dimensions.

[0128] Apply a non-linear activation function, such as the ReLU function, to the product result to obtain the updated node feature matrix. This step introduces non-linearity and enhances the expressive power of the model. The dimension of the updated node feature matrix may become 20x128.

[0129] Based on the updated node feature matrix, extract features from the 2D object recognition results and 3D model data in the construction scene. The 2D features may include the bounding box coordinates and class probabilities of the objects, and the 3D features may include point cloud data, volume information, etc.

[0130] Input the extracted 2D and 3D features into the multi-modal feature fusion module. This module uses fully connected layers to process the 2D and 3D features respectively, and then calculates the fusion weights through an attention mechanism. The attention mechanism can adaptively adjust the importance of different modal features. For example, in some cases, it may rely more on 2D features, while in other cases, it may rely more on 3D features.

[0131] Perform weighted fusion on the processed features according to the calculated fusion weights to obtain the fused features. The fused features integrate 2D image information and 3D structure information, providing rich inputs for constructing an accurate digital twin model.

[0132] Finally, based on the fused features and the graph structure obtained through spatial-aware graph convolution operations, construct a digital twin model of the construction scene. This digital twin model not only contains the geometric information of each component, but also contains the topological and spatial relationships between them, and can accurately reflect the state of the actual construction scene.

[0133] This method for constructing a digital twin model based on spatial-aware graph convolution and multi-modal feature fusion has several beneficial effects. First, it can more accurately capture the spatial relationships between elements in the construction scene, improving the geometric accuracy of the model. Second, by fusing 2D and 3D features, this method fully utilizes information from different modalities, enhancing the robustness and adaptability of the model. Third, the spatial-aware graph convolution operation enables the model to effectively process large-scale and complex construction scenes, with good scalability. Finally, the constructed digital twin model provides a solid foundation for subsequent construction monitoring, safety warning, and intelligent decision-making, helping to improve the efficiency and safety of bridge erector construction.

[0134] In an alternative embodiment,

[0135] The intelligent decision-making agent generates an initial construction strategy based on the current construction scene state, optimizes the initial construction strategy using an improved Monte Carlo tree search, and generates an optimal construction strategy through multi-objective solving. The steps include:

[0136] Using the intelligent decision-making agent to generate an initial construction strategy based on the current construction scene state;

[0137] Using an improved Monte Carlo tree search to optimize the initial construction strategy and generate an optimization result; the improved Monte Carlo tree search includes introducing a dynamic action generator, using the UCT algorithm with adaptive exploration parameters, implementing tree parallelization, and applying heuristic pruning;

[0138] In the UCT algorithm with the adaptive exploration parameter, introduce a dynamically adjusted exploration parameter c:

[0139] c = c_base * (1 + α * exp(-β * t / T));

[0140] where c_base is the base exploration parameter, α and β are adjustment factors, t is the current iteration number, and T is the total iteration number;

[0141] Perform multi-objective solving on the optimization result of the initial construction strategy, including defining objective functions for construction efficiency and energy consumption, and using an improved NSGA-III algorithm for multi-objective optimization. The improved NSGA-III algorithm includes adaptive crossover and mutation, local search, and reference point adaptation; generate an optimal construction strategy according to the multi-objective optimization solution result.

[0142] Exemplarily, the method of the intelligent decision-making agent combining the improved Monte Carlo tree search and the multi-objective optimization algorithm to generate an optimal construction strategy is an advanced artificial intelligence technology, aiming to improve the efficiency and safety of the bridge erecting machine construction. This method combines multiple technologies such as reinforcement learning, heuristic search, and evolutionary algorithms to achieve an intelligent conversion from the current construction scene state to the optimal construction strategy.

[0143] First, use the intelligent decision-making agent to generate an initial construction strategy based on the current construction scene state. The intelligent decision-making agent is trained using a deep reinforcement learning model, such as the Proximal Policy Optimization algorithm. The agent receives the state information of the construction scene, including the positions, postures, construction progress, etc. of each component, and outputs a series of initial construction actions. For example, for a typical bridge erecting machine construction scene, the initial strategy may include 20 consecutive construction actions, and each action includes specific parameters such as main beam movement, cross beam adjustment, and spreader operation.

[0144] Next, the improved Monte Carlo Tree Search (MCTS) is used to optimize the initial construction strategy. The improved MCTS contains multiple innovative points. First, a dynamic action generator is introduced. This generator dynamically generates a set of feasible actions according to the current state, rather than using a fixed action space. For example, in some states, only part of the operations may be safe or effective, and the dynamic generator will adjust the optional actions accordingly.

[0145] The UCT (game tree search) algorithm with adaptive exploration parameters is used in MCTS. The core of the UCT algorithm is to balance exploration and exploitation, and the exploration parameter c plays a key role. In this method, the value of c is dynamically adjusted with the search process:

[0146] c = c_base * (1 + α * exp(-β * t / T));

[0147] where c_base is the base exploration parameter, α and β are adjustment factors, t is the current iteration number, and T is the total iteration number.

[0148] Specifically, the calculation of the c value takes into account the base exploration parameter, the current iteration number, and the total iteration number. For example, assume that the base exploration parameter is 1.4, the adjustment factor α is 0.5, β is 2, and the total iteration number is 1000. Then, at the 100th iteration, the c value is approximately 1.9, and at the 900th iteration, the c value is approximately 1.45. This dynamic adjustment makes the algorithm more inclined to explore in the early stage of the search and pay more attention to exploitation in the later stage.

[0149] Another improvement of MCTS is to achieve tree parallelization. Utilizing the advantages of multi-core processors, multiple search trees are expanded simultaneously, and each search tree is responsible for a different action sequence. This greatly improves the search efficiency, enabling a larger decision space to be explored within a limited time. For example, on an 8-core processor, 8 search trees can be expanded simultaneously, and each tree explores a different initial action.

[0150] In addition, MCTS also applies heuristic pruning techniques. Based on domain knowledge and historical data, some obviously suboptimal or unsafe action sequences are pruned to reduce the search space. For example, if a certain action sequence will cause component collisions, that sequence and its subsequences will be directly pruned and no longer searched.

[0151] The optimization result of MCTS is a series of improved construction action sequences, which can generally better balance efficiency and safety compared to the initial strategy.

[0152] Perform multi-objective optimization on the results optimized by MCTS to further optimize the construction strategy. First, define multiple objective functions. Typical objectives include construction efficiency and energy consumption. Construction efficiency can be quantified by the time required to complete a specific construction task, while energy consumption can be measured by the cumulative power or fuel consumption.

[0153] The multi-objective optimization uses an improved NSGA-III (Non-dominated Sorting Genetic Algorithm III) algorithm. The improvements of this algorithm include three aspects: adaptive crossover and mutation, local search, and reference point adaptation. Adaptive crossover and mutation dynamically adjust the crossover rate and mutation rate according to the population diversity. For example, when the population diversity is low, increase the mutation rate to improve the exploration ability; when the diversity is high, increase the crossover rate to enhance exploitation.

[0154] The local search strategy fine-tunes some excellent individuals after each generation of evolution. Specifically, the simulated annealing algorithm can be used to perform local optimization on the selected individuals to further improve the quality of the solution. Reference point adaptation dynamically adjusts the position and density of the reference point according to the current population distribution to ensure that the algorithm can evenly explore the entire Pareto front.

[0155] Through multi-objective optimization, a set of non-dominated solutions, that is, the Pareto optimal solution set, is finally obtained. Each solution represents a construction strategy that achieves different balances between construction efficiency and energy consumption. Decision-makers can select the most suitable strategy from this set of solutions according to specific needs. For example, in a time-constrained situation, a strategy with higher efficiency but slightly higher energy consumption may be selected; while in regular construction, a strategy with a more balanced efficiency and energy consumption may be chosen.

[0156] The construction strategy generation method based on intelligent decision-making agents, improved Monte Carlo tree search, and multi-objective optimization in this paper has beneficial effects in many aspects. First, it can automatically generate high-quality construction strategies according to the complex construction scenario state, greatly reducing the burden of manual decision-making. Second, the improved MCTS significantly improves the efficiency and quality of strategy optimization through technologies such as dynamic action generation, adaptive exploration, and parallelization. Third, multi-objective optimization considers multiple objectives such as construction efficiency and energy consumption, making the generated strategies more comprehensive and practical. Finally, the adaptability and flexibility of this method enable it to adapt to various complex construction environments and requirements, providing strong technical support for the intelligent and refined construction management of bridge erection machines.

[0157] In an optional embodiment,

[0158] The steps of the improved NSGA-III algorithm including adaptive crossover and mutation, local search, and reference point adaptation include:

[0159] An adaptive crossover and mutation mechanism based on population diversity is introduced, and the crossover probability and mutation probability are dynamically adjusted according to the diversity measure of the current population. The diversity measure is calculated using the entropy weight method, and the intensity of genetic operations is increased when the population diversity decreases;

[0160] After each generation of iteration of the NSGA-III algorithm, the objective function values of all individuals in the current population are evaluated, the non-dominated solution set is identified and constructed, multiple solutions are selected from the non-dominated solution set for local search, and multiple simulated annealing iterations are performed on each selected solution. The simulated annealing iteration uses an exponentially decaying temperature schedule, generates neighborhood solutions in each iteration, and calculates the acceptance probability according to the energy difference. If the solution obtained by local search dominates the original solution, the original solution is replaced with the new solution;

[0161] A reference point adaptive mechanism is adopted, including uniformly generating initial reference points in the standardized objective space, periodically evaluating the association degree of each reference point during the evolution process, using the density peak clustering algorithm to identify high-density regions in the objective space for low-association reference points, moving the low-association reference points to the nearest high-density region and maintaining dispersion, and at the same time introducing a regularization term to limit the large-scale migration of reference points.

[0162] Exemplarily, the improved NSGA-III algorithm is an efficient multi-objective optimization method. For the complex bridge erection machine construction strategy optimization problem, innovative mechanisms such as adaptive crossover and mutation, local search, and reference point adaptation are introduced. These improvements aim to improve the convergence speed and solution quality of the algorithm while maintaining the diversity and uniform distribution of the population.

[0163] First, an adaptive crossover and mutation mechanism based on population diversity is introduced. This mechanism dynamically adjusts the intensity of genetic operations to adapt to the evolutionary state of the population. Specifically, the entropy weight method is used to calculate the diversity measure of the population. The entropy weight method considers the distribution of each decision variable in the population, calculates the information entropy of each variable, and then comprehensively obtains the overall diversity index. For example, assume there is a population of 100 individuals, and each individual has 10 decision variables. For each decision variable, its value range is equally divided into 10 intervals, the number of individuals falling into each interval is counted, and the information entropy of this variable is calculated. Then, the information entropies of all variables are weighted averaged to obtain the diversity measure of the population.

[0164] Dynamically adjust the crossover probability and mutation probability according to the calculated diversity metric. When the diversity metric is lower than a preset threshold (such as 0.6), increase the mutation probability to improve the exploration ability of the population. For example, the base mutation probability of 0.01 can be increased to 0.05. At the same time, appropriately reduce the crossover probability, such as from 0.9 to 0.8, to reduce the destruction of existing genes. On the contrary, when the diversity metric is higher than another threshold (such as 0.8), increase the crossover probability and reduce the mutation probability to strengthen the utilization of existing excellent genes.

[0165] After each generation of iteration in the NSGA-III algorithm, introduce a local search mechanism to further improve the quality of the solutions. First, evaluate the objective function values of all individuals in the current population, and construct the non-dominated solution set through non-dominated sorting. From this non-dominated solution set, select multiple (such as 10) representative solutions for local search. The selection criteria can be based on the distribution of the solutions in the objective space to ensure coverage of different regions of the Pareto front.

[0166] Perform multiple simulated annealing iterations on each selected solution. The simulated annealing algorithm adopts an exponentially decaying temperature schedule, with the initial temperature set to 100 and the temperature decreasing by 1% in each iteration. In each iteration, generate a neighborhood solution of the current solution through a small random perturbation. For example, for a solution containing 20 decision variables, randomly select 1 - 3 variables and adjust them within ±5% of their current values to generate a new candidate solution.

[0167] Calculate the energy difference based on the objective function values of the new solution and the current solution, and calculate the acceptance probability based on the energy difference and the current temperature. If the new solution is better than the current solution, directly accept it; otherwise, accept the inferior solution with a certain probability, and this probability decreases as the temperature decreases. Through this mechanism, the algorithm can jump out of the local optimum in local search. Perform 100 simulated annealing iterations on each selected solution. If the new solution obtained by local search is better than the original solution in any objective and not inferior to the original solution in other objectives (i.e., the new solution dominates the original solution), then replace the original solution with the new solution.

[0168] To better guide the population to converge to the Pareto front and maintain the diversity of solutions, adopt a reference point adaptive mechanism. First, uniformly generate initial reference points in the normalized objective space. For example, for a three-objective optimization problem, 100 uniformly distributed reference points can be generated on the unit hyperplane.

[0169] During the evolution process, periodically (such as every 10 generations) evaluate the degree of association of each reference point. The degree of association is measured by calculating the number of individuals closest to each reference point. For reference points with an association degree lower than the threshold (such as 3), it is considered that they are in a low-density area and need to be adjusted.

[0170] Use the density peak clustering algorithm to identify high-density regions in the target space. This algorithm first calculates the local density of each solution and the distance to high-density points, and then identifies the density peak points, which have both high local density and are far from other high-density points. These density peak points represent the high-density regions in the target space.

[0171] Move the reference points with low correlation to the nearest high-density region. When moving, it is necessary to maintain the dispersion between reference points to avoid multiple reference points clustering in the same region. Specifically, the reference points can be moved to the edge rather than the center of the high-density region to maintain the coverage range. At the same time, a regularization term is introduced to limit the large-scale migration of reference points. For example, the maximum movement distance for each adjustment can be restricted to no more than 20% of the average distance between reference points.

[0172] The improved NSGA-III algorithm adopted in this paper has many beneficial effects in the optimization of the bridge erection machine construction strategy. First, the adaptive crossover and mutation mechanism can dynamically adjust the intensity of genetic operations according to the population state, effectively balancing the exploration and exploitation capabilities of the algorithm and accelerating the convergence speed. Second, the local search strategy finely adjusts the excellent solutions through the simulated annealing algorithm, significantly improving the quality of the solutions, especially in complex decision spaces, and can find better solutions. Third, the reference point adaptive mechanism enables the algorithm to better adapt to the distribution characteristics of the target space, improving the diversity and uniformity of the solutions, and providing a more comprehensive Pareto optimal solution set for decision-makers. Finally, these improvements enable the algorithm to perform excellently in dealing with high-dimensional, multi-objective construction strategy optimization problems, and can quickly generate high-quality and diverse construction plans with limited computing resources, providing strong decision-making support for intelligent and refined bridge erection machine construction management.

[0173] In an optional embodiment,

[0174] During the construction process, continuously collect the data of various sensors of the bridge erection machine, fuse the data of various sensors and the feedback data of the actuators to generate a state estimate of the bridge erection machine; the steps of performing closed-loop control on the bridge erection machine using the nonlinear model predictive control algorithm based on the state estimate include:

[0175] Use an inertial measurement unit, a global positioning system, a laser rangefinder, a strain sensor, and an actuator encoder to collect the state information of the bridge erection machine, and the state information includes acceleration, angular velocity, position, relative distance, structural stress, hydraulic cylinder position, and motor speed;

[0176] Construct a state vector including position, acceleration, angular velocity, and force / moment, and establish a discrete-time nonlinear state equation and an observation equation based on the state vector;

[0177] The extended Kalman filter is used for state estimation of discrete-time nonlinear state equations and observation equations, including performing a prediction step and an update step. The prediction step includes state prediction and covariance prediction, and the update step includes calculating the Kalman gain, state update, and covariance update;

[0178] Based on the state estimation results, a nonlinear model predictive control optimization problem is constructed. The optimization objectives include tracking a predetermined trajectory, minimizing energy consumption, and satisfying safety constraints;

[0179] A real-time iterative strategy is used to solve the nonlinear model predictive control optimization problem. The solution process is divided into a preparation stage and a feedback stage. The preparation stage includes linearizing the system dynamics and constructing a quadratic programming subproblem, and the feedback stage includes updating the initial state, solving the quadratic programming subproblem, and applying the control input;

[0180] Based on the solution results of the nonlinear model predictive control, an optimal control input is generated, and the optimal control input is converted into a control signal for the bridge erecting machine actuator to perform closed-loop control on the bridge erecting machine.

[0181] Exemplarily, the intelligent control system of the bridge erecting machine is an integration of complex multi-sensor fusion and advanced control algorithms. The system realizes accurate estimation of the state of the bridge erecting machine by continuously collecting and processing multi-source sensor data, and based on these estimation results, uses the nonlinear model predictive control algorithm for closed-loop control to ensure the accuracy, safety, and energy efficiency of the construction process.

[0182] First, the system uses a variety of sensors to collect the state information of the bridge erecting machine. The inertial measurement unit (IMU) is used to measure the acceleration and angular velocity of the bridge erecting machine, with a typical sampling frequency of 100 Hz and an accuracy of up to 0.01 m / s 2 and 0.01 ° / s. The global positioning system (GPS) provides the absolute position information of the bridge erecting machine. Usually, the RTK-GPS technology is adopted, which can achieve centimeter-level positioning accuracy and an update frequency of 10 Hz. The laser rangefinder is used to measure the relative distance between the bridge erecting machine and the surrounding environment, with a measurement range of up to 100 m and an accuracy of ±1 mm. Strain sensors are installed at key structural parts of the bridge erecting machine to monitor the structural stress, and the sampling frequency is usually 1 kHz. The actuator encoder records the hydraulic cylinder position and motor speed, with a resolution of up to 0.1 mm and 1 rpm.

[0183] After preliminary processing, these sensor data are used to construct a state vector that includes position, acceleration, angular velocity, and forces / torques. For example, a typical state vector may contain 18 variables: 3 spatial position coordinates, 3 Euler angles, 3 linear velocity components, 3 angular velocity components, and 6 principal forces / torques. Based on this state vector, discrete-time nonlinear state equations and observation equations are established. The state equations describe the evolution of the system state, taking into account the kinematic and dynamic characteristics of the bridge erecting machine. The observation equations, on the other hand, establish the relationship between the sensor measurements and the system state.

[0184] To obtain an accurate state estimate from the sensor data contaminated by noise, an Extended Kalman Filter (EKF) is used for state estimation. The execution of the EKF consists of two main steps: prediction and update. In the prediction step, based on the previous state estimate and control input, the state equation is used to predict the state and state covariance at the current time. For example, if the estimated position of the bridge erecting machine at the previous time is (10m, 5m, 2m) and the velocity is (0.5m / s, 0.3m / s, 0.1m / s), then the position after 0.1 seconds can be predicted to be approximately (10.05m, 5.03m, 2.01m).

[0185] In the update step, the Kalman gain is first calculated, which is a key parameter that balances the prior estimate and measurement information. Then, based on the difference between the actual measurement and the predicted value, the state estimate and covariance are updated. For example, if the GPS measurement shows that the actual position is (10.06m, 5.04m, 2.02m), the EKF will correct the predicted position according to the Kalman gain to obtain a more accurate state estimate.

[0186] Based on the state estimation results of the EKF, a Nonlinear Model Predictive Control (NMPC) optimization problem is constructed. The optimization objectives of NMPC generally include three aspects: tracking a predefined trajectory, minimizing energy consumption, and satisfying safety constraints. Tracking a predefined trajectory can be defined as the sum of the squares of the deviations between the actual trajectory and the reference trajectory. Minimizing energy consumption can be approximated by minimizing the sum of the squares of the control inputs. Safety constraints include speed limits, acceleration limits, joint angle limits, etc.

[0187] The solution of the NMPC optimization problem adopts a real-time iterative strategy, which divides the solution process into a preparation stage and a feedback stage. In the preparation stage, the system dynamics are first linearized. For example, linearization approximation can be performed using Taylor expansion near the current operating point. Then, a quadratic programming subproblem is constructed based on the linearized model. This subproblem is a local approximation of the original nonlinear problem, but it can be quickly solved using an efficient quadratic programming solver.

[0188] In the feedback stage, first, update the initial state using the latest state estimation results. Then, solve the constructed quadratic programming sub-problem to obtain a series of optimal control inputs. For example, for an NMPC problem with a prediction horizon of 2 seconds and a control interval of 0.1 seconds, 20 control inputs will be obtained each time the solution is calculated. Finally, apply the first control input obtained from the solution to the system, and use the remaining control inputs as the initial guess for the next optimization.

[0189] Based on the solution results of NMPC, generate optimal control inputs and convert them into specific control signals for the execution mechanisms of the bridge erecting machine. For example, the control inputs may include the traveling speed of the main girder, the pitching angular velocity of the cross beam, the hook speed of the hoisting mechanism, etc. These control signals are converted into corresponding voltage or current signals through a digital-to-analog converter and sent to the drivers of each execution mechanism, thereby realizing the closed-loop control of the bridge erecting machine.

[0190] The intelligent control system of the bridge erecting machine based on multi-sensor fusion and non-linear model predictive control adopted in this paper has many beneficial effects. First, multi-sensor fusion and EKF state estimation provide a highly accurate and robust estimation of the state of the bridge erecting machine, laying a solid foundation for subsequent control. Second, the NMPC algorithm can simultaneously consider the non-linear characteristics of the system, multi-objective optimization, and various constraints, and generate an optimal control strategy. This not only ensures that the bridge erecting machine can accurately track the predetermined trajectory but also minimizes energy consumption while ensuring safety. Third, the adoption of the real-time iteration strategy enables NMPC to obtain an effective solution within limited computational time, meeting the requirements of real-time control. Finally, this closed-loop control method can effectively cope with external disturbances and model uncertainties, significantly improving the accuracy, efficiency, and safety of bridge erecting machine construction, and providing advanced technical support for large-scale infrastructure construction.

[0191] In an alternative embodiment,

[0192] After the construction is completed, the steps of analyzing the construction quality and efficiency using a graph neural network based on the digital twin model and the whole-process data to generate a multi-dimensional evaluation report and optimization suggestions include:

[0193] Represent the bridge erecting machine and the bridge structure as a heterogeneous graph, where the nodes represent components or monitoring points, the edges represent the physical connections or logical relationships between components, and each node and edge are attached with multi-dimensional feature vectors; apply specialized feature transformation matrices to the multi-dimensional feature vectors of different types of nodes and edges to generate the initial node and edge representations;

[0194] Based on the initial node and edge representations, construct an improved graph neural network model, including a node-level attention mechanism and an edge-level attention mechanism, where the node-level attention is used to learn the importance weights between different features, and the edge-level attention is used to learn the influence degree of different neighbor nodes;

[0195] After the graph convolution layer of the graph neural network model, a temporal convolutional network is introduced. Causal convolution and dilated convolution are used to capture temporal features and fuse them with the output of the graph convolution layer;

[0196] Use the historical data of the bridge erecting machine construction system to train the graph neural network model, and iterate the training until convergence;

[0197] Use the trained graph neural network model to conduct quality and efficiency analysis on the new data of the bridge erecting machine construction system to obtain the prediction results of various indicators; based on the prediction results, generate a multi-dimensional evaluation report including overall score, analysis of various indicators, abnormal event identification and its root cause analysis; according to the problems and potential risks identified in the multi-dimensional evaluation report, combined with historical construction experience, generate optimization suggestions and improvement measures.

[0198] Exemplarily, after the bridge erecting machine construction is completed, using the digital twin model and the whole-process data for quality and efficiency analysis is a complex and important task. This technical solution proposes an innovative method based on graph neural network, which can comprehensively analyze the construction quality and efficiency, generate a multi-dimensional evaluation report and optimization suggestions.

[0199] First, represent the bridge erecting machine and the bridge structure as a heterogeneous graph. In this graph, nodes can represent different components or monitoring points. For example, a node may represent a main girder, a cross beam, a support leg, or a specific stress monitoring point. Edges represent the physical connections or logical relationships between components, such as the connection between the main girder and the cross beam, or the spatial relationship between adjacent monitoring points. Each node and edge is attached with a multi-dimensional feature vector, containing rich information. For nodes, the features may include position coordinates, geometric dimensions, material properties, stress and strain values, etc. For example, a node representing a main girder may have the following features: [10.5, 2.3, 1.8, 500, 7850, 210, 150], representing length (m), width (m), height (m), mass (kg), density (kg / m 3 ), elastic modulus (GPa), and maximum stress (MPa) respectively. The features of the edge may include connection type, stiffness coefficient, force transmission characteristics, etc.

[0200] To process these heterogeneous features, specialized feature transformation matrices are applied to different types of nodes and edges. The purpose of this step is to transform the original features into vector representations with the same dimension, facilitating subsequent graph neural network processing. For example, a 100x7 transformation matrix can be used to transform the above 7-dimensional main girder features into a 100-dimensional vector representation. This process is essentially a feature embedding that can capture the potential relationships between features.

[0201] Based on the initial node and edge representations, an improved graph neural network model is constructed. This model contains two key attention mechanisms: node-level attention and edge-level attention. The node-level attention mechanism is used to learn the importance weights between different features. For example, when evaluating the quality of the main girder, the stress value may be more important than the density, so it will be assigned a higher weight. The edge-level attention mechanism is used to learn the influence degree of different neighbor nodes. For example, when analyzing the abnormality of a monitoring point, the directly connected components may have a greater influence than the distant components.

[0202] A temporal convolutional network is introduced after the graph convolutional layer of the graph neural network model. This design aims to capture temporal features. Causal convolution is used to ensure that the model only uses past and current information for prediction, avoiding information leakage. Dilated convolution, on the other hand, increases the receptive field, enabling the model to effectively capture long-term dependencies. For example, 3 layers of dilated convolution with dilation rates of 1, 2, and 4 can be used to cover a longer time span without increasing the number of parameters. The output of the temporal convolutional network is fused with the output of the graph convolutional layer to obtain a feature representation that combines spatial and temporal information.

[0203] The graph neural network model is trained using the historical data of the bridge erecting machine construction system. The training data includes various sensor data collected during construction, construction records, and the final quality assessment results. The training process uses the mini-batch stochastic gradient descent method, with each batch containing 64 samples. The initial learning rate is set to 0.001, and a learning rate decay strategy is used. The model training is carried out iteratively until the loss function on the validation set converges or reaches the preset number of iterations (such as 1000 rounds).

[0204] The trained graph neural network model is used to analyze the quality and efficiency of the new bridge erecting machine construction system data. The model can predict various indicators, such as structural stress distribution, deformation, construction efficiency, etc. For example, for a new construction project, the model may predict that the maximum stress of the main girder is 145 MPa, the deformation is 5 mm, and the construction efficiency is 95%.

[0205] Based on the prediction results, a multi-dimensional evaluation report is generated. This report includes an overall score, analysis of each indicator, identification of abnormal events and their root cause analysis. The overall score can use a 100-point system, comprehensively considering various aspects of performance. The analysis of each indicator details the predicted values, historical comparisons, and industry standard comparisons of each key indicator. For example, the report may indicate that the stress of the main girder is within the normal range but 5% higher than the historical average, and it is recommended to conduct further monitoring. The abnormal event identification section lists all indicators or abnormal patterns that exceed the preset threshold, and uses the interpretability features of the graph neural network for root cause analysis. For example, if the stress of a certain node is detected to be abnormally high, the model can trace the key factors and propagation paths that affect this node.

[0206] Based on the problems and potential risks identified in the multi-dimensional assessment report, combined with historical construction experience, optimization suggestions and improvement measures are generated. These suggestions may include adjusting construction techniques, optimizing resource allocation, strengthening the monitoring of specific parts, etc. For example, if the report indicates that stress concentration frequently occurs in a certain type of connector, the system may suggest improving the connection design or using higher-strength materials.

[0207] The method for analyzing the construction quality and efficiency of the bridge erecting machine based on the graph neural network adopted in this paper has beneficial effects in many aspects. First of all, it can comprehensively consider the complex topological relationship between the bridge erecting machine and the bridge structure, capture the mutual influence between components, and provide more accurate analysis results than traditional methods. Secondly, the introduced attention mechanism and temporal convolutional network enable the model to adaptively focus on important features and long-term dependencies, improving the accuracy and robustness of the analysis. Moreover, this method can achieve end-to-end learning, directly obtaining the evaluation results from the original data, reducing the interference of human factors. Finally, the generated multi-dimensional assessment report and optimization suggestions provide intuitive and comprehensive decision-making support for construction management personnel, helping to improve construction quality, efficiency, and safety. The application of this method can not only optimize the current construction project but also continuously improve construction techniques and management strategies by accumulating and analyzing a large amount of data, promoting the technological progress of the entire bridge construction industry.

[0208] In an alternative embodiment,

[0209] As Figure 2 shown, the structure of the bridge erecting machine of the present invention includes: a cross beam, a camera, and a transport vehicle.

[0210] Among them, label 1 is the right-end girder assembly, which is used to provide the support structure at the right end of the bridge erecting machine, lateral stability, and the load-bearing function of the end equipment; label 2 is the front support system, which is used to support the front end of the main girder, provide stability in the advancing direction, and bear the front load distribution; label 3 is the transporter, which is used to transport bridge segments, achieve precise positioning of the bridge sections, and meet the movement requirements during the installation process; label 4 is the cross beam, which is used to provide lateral structural support, connect each support system, and ensure the overall structural stiffness; label 5 is the camera, which is used to obtain the image information of the construction environment where the bridge erecting machine is located, and can transmit it to the controller by wired or wireless means to perform corresponding control on the bridge erecting machine; label 6 is the overhead crane system, which is used to hoist bridge components, perform precise position adjustment, and complete the movement control in the horizontal and vertical directions; label 8 is the middle support bracket, which is used to support the middle part of the main girder, disperse the bearing pressure, and provide stable support during the construction process; label 9 is the auxiliary support, which is used to provide additional support force, increase the overall structural stability, and assist the support requirements under special working conditions; label 10 is the front support system, which is used in cooperation with the No. 2 support to enhance the front structural strength and provide a construction operation platform; label 11 is the left-end girder assembly, which is used as the left-end support structure of the bridge erecting machine, balance the overall force, and cooperate with the right end to form a complete support system; label 12 is the main girder, which is used as the main load-bearing component of the entire bridge erecting machine, provide longitudinal support and guidance, and at the same time bear the installation foundation of various mechanisms and equipment.

[0211] The main parameters are as follows: Rated lifting capacity: 800t; Suitable erection beam span: ≤32m; Whole machine passing-through speed: 3m / min; Hoisting speed of lifting beam: No load: 0 - 1m / min, Heavy load: 0 - 0.5m / min; Longitudinal movement speed of lifting beam: 0 - 3m / min; Longitudinal movement speed of guide girder: 0 - 3m / min; Transverse movement amount of machine arm: Column No. 1 ±250mm, Column No. 2 ±350mm; Transverse movement speed of machine arm: 0.3m / min; Suitable slope: ≤30‰; Minimum curve radius: 1600m; Installed capacity: 230kw; System pressure: 31.5Mpa; Working level: A3; Maximum overall dimensions: 82.6mx16.8mX12m;

[0212] Its working process includes:

[0213] Preparation stage: The bridge erecting machine is in place, spanning two bridge piers; System calibration, including the initialization of the machine vision system;

[0214] Beam segment transportation: The bridge erecting machine moves to the precast beam segment storage area (speed: 0 - 3m / min); The spreader descends to grab the beam segment (descending speed: 0 - 1m / min);

[0215] Beam segment installation: The bridge erection machine carries the beam segment to the installation location; uses the machine vision system for precise positioning; and slowly lowers the beam segment to the predetermined location (lowering speed: 0-0.5m / min);

[0216] Precise adjustment: The machine vision system monitors the beam segment position in real time; fine-tuning is performed through lateral and longitudinal sliding (±250mm and ±350mm); ensuring that the beam segment positioning error is within the allowable range;

[0217] Fixing and verification: Fixing of beam segments after they are in place; final inspection and data recording by machine vision system;

[0218] Prepare for the next cycle: the bridge erecting machine moves to the next working position (moving speed: 0.3m / min).

[0219] The technical effect of the bridge erection machine provided by the present invention is mainly reflected in the high-precision installation. The lateral and longitudinal sliding mechanisms cooperate with the machine vision system to achieve accurate positioning of the beam segment, with high installation accuracy, ensuring the high quality and stability of the bridge structure. The bridge erection machine has a large load capacity, with a rated lifting capacity of 800 tons, and can handle large prefabricated beam segments, meeting the needs of modern bridge engineering for large spans and heavy loads.

[0220] In terms of flexibility and adaptability, the maximum span can reach 32 meters, suitable for various bridge types, and the minimum curve radius is 1,600 meters, which can adapt to different bridge lines. Efficient construction is another major feature. The whole machine has a hole-passing speed of 3m / min, a walking speed of 0.3m / min, and a hoisting speed of 0-1m / min, which enables fast construction. The high degree of automation reduces manual operations and improves construction efficiency.

[0221] In terms of safety performance, the system pressure is 31.5MPa, ensuring stability under heavy loads. The machine vision system monitors in real time to reduce human errors and safety risks. Precision control is its important feature. The hoisting speed can be precisely controlled to ensure that the beam section is in place smoothly. The lateral and longitudinal moving speeds can reach 0-3m / min, which is convenient for fine adjustment.

[0222] The bridge erecting machine has large-size adaptability, with a maximum size of 82.6m×16.8m×12m, which is suitable for large bridge projects. It has strong environmental adaptability and working level A3, indicating strong environmental adaptability and stability. Intelligent construction is its significant advantage. Combined with the machine vision system, it realizes intelligent positioning, automatic adjustment and real-time monitoring, improves construction accuracy and reduces human errors.

[0223] With strong data recording and analysis capabilities, the machine vision system can record construction data for quality control and subsequent optimization. The modular design facilitates transportation, assembly and maintenance. Its versatility makes it not only used for beam segment installation, but also for the lifting and positioning of other bridge components.

[0224] This type of bridge erection machine structure greatly improves the efficiency, precision, and safety of bridge construction by combining advanced mechanical design and intelligent control systems, representing the development direction of modern bridge engineering technology.

[0225] Figure 3 It is a schematic structural diagram of the intelligent construction control system of the bridge erection machine based on machine vision in the embodiment of the present invention. As Figure 3 shown, the system includes:

[0226] The first unit is used to collect image information of the construction site through multiple high-definition cameras installed on the bridge erection machine; input the image information into a pre-trained deep learning model to identify and locate the targets in the construction scene, including various components of the bridge erection machine, bridge components, and the surrounding environment; based on the recognition results of the targets in the construction scene, combined with the graph structure of the construction scene topology after spatial-aware graph convolution operation, construct a digital twin model of the construction scene;

[0227] The second unit is used to train an intelligent decision-making agent based on the digital twin model and a predefined reward and punishment function using the reinforcement learning algorithm; the intelligent decision-making agent generates an initial construction strategy according to the current construction scene state, optimizes the initial construction strategy using an improved Monte Carlo tree search, and generates an optimal construction strategy through multi-objective solution; converts the optimal construction strategy into intelligent control instructions, and sends the intelligent control instructions to the control system of the bridge erection machine to achieve precise control of each actuator of the bridge erection machine; during the construction process, continuously collect the data of various sensors of the bridge erection machine, fuse the data of various sensors and the feedback data of the actuators to generate a state estimate of the bridge erection machine; based on the state estimate, use the nonlinear model predictive control algorithm to perform closed-loop control on the bridge erection machine;

[0228] The third unit is used to analyze the construction quality and efficiency based on the digital twin model and the whole-process data after the construction is completed, and generate a multi-dimensional evaluation report and optimization suggestions.

[0229] In the third aspect of the embodiment of the present invention,

[0230] Provided is an electronic device, including:

[0231] A processor;

[0232] A memory for storing instructions executable by the processor;

[0233] Wherein, the processor is configured to call the instructions stored in the memory to execute the method described above.

[0234] In the fourth aspect of the embodiment of the present invention,

[0235] Provided is a computer-readable storage medium having stored thereon computer program instructions which, when executed by a processor, implement the foregoing method.

[0236] The present invention may be a method, an apparatus, a system, and / or a computer program product. The computer program product may include a computer-readable storage medium having thereon computer-readable program instructions for performing various aspects of the present invention.

[0237] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the various embodiments of the present invention.

Claims

1. The intelligent construction control method of bridge erection machine based on machine vision is characterized in that: include: The image information of the construction site is collected by multiple high-definition cameras installed on the bridge erection machine; the image information is input into a pre-trained deep learning model to identify and locate targets in the construction scene, including various components of the bridge erection machine, bridge components and the surrounding environment; based on the recognition results of the targets in the construction scene, combined with the graph structure of the topological structure of the construction scene after the spatial perception graph convolution operation, a digital twin model of the construction scene is constructed; Using reinforcement learning algorithms, we train intelligent decision-making agents based on digital twin models of construction scenarios and predefined reward and punishment functions; The intelligent decision-making agent generates an initial construction strategy according to the current state of the construction scene, optimizes the initial construction strategy using an improved Monte Carlo tree search algorithm, and generates an optimal construction strategy through multi-objective solution; after the optimal construction strategy is converted into an intelligent control instruction, the intelligent control instruction is sent to the control system of the bridge-building machine; during the construction process, the data of each sensor of the bridge-building machine is continuously collected, and the data of each sensor and the feedback data of each actuator are integrated to generate a state estimation of the bridge-building machine; based on the state estimation, the nonlinear model predictive control algorithm is used to perform closed-loop control of the bridge-building machine; After the construction is completed, the construction quality and efficiency are analyzed using graph neural networks based on the digital twin model and the entire process data, generating multi-dimensional evaluation reports and optimization suggestions; The improved Monte Carlo tree search algorithm includes: A dynamic action generator is introduced, a UCT algorithm with adaptive exploration parameters is used, tree parallelization is implemented, and heuristic pruning is applied; in the UCT algorithm with adaptive exploration parameters, a dynamically adjusted exploration parameter c is introduced: c=c_base*(1+α*exp(-β*t / T)); Among them, c_base is the basic exploration parameter, α and β are the first adjustment factor and the second adjustment factor respectively, t is the current iteration number, and T is the total iteration number; The multi-objective solution includes: defining the objective functions of construction efficiency and energy consumption, using the improved NSGA-III algorithm for multi-objective optimization, the improved NSGA-III algorithm including adaptive crossover and mutation, local search and reference point adaptation; generating the optimal construction strategy according to the multi-objective optimization solution result; The use of graph neural network to analyze construction quality and efficiency includes: The bridge erection machine and the bridge structure are represented as heterogeneous graphs, and a special feature transformation matrix is ​​applied to the multi-dimensional feature vectors of different types of nodes and edges to generate initial nodes and edges; Based on the initial nodes and edges, an improved graph neural network model is constructed. A temporal convolutional network is introduced after the graph convolutional layer of the improved graph neural network model. Causal convolution and dilated convolution are used to capture the temporal features and fuse them with the output of the graph convolutional layer.

2. The method according to claim 1, characterized in that: The steps of collecting image information of the construction site by multiple high-definition cameras installed on the bridge erecting machine; inputting the image information into a pre-trained deep learning model to identify and locate targets in the construction scene, including various parts of the bridge erecting machine, bridge components and the surrounding environment include: The image information of the construction site is collected by multiple high-definition cameras arranged on the bridge erecting machine; the image information is preprocessed by adaptive histogram equalization; the preprocessed image information is input into a pre-trained improved Mask R-CNN deep learning model, and a spatial pyramid pooling module is introduced into the feature extraction network of the improved Mask R-CNN deep learning model; An attention mechanism module is added after the region of interest alignment layer of the improved Mask R-CNN deep learning model to highlight features and suppress irrelevant background information by generating spatial attention maps and channel attention maps; Using the improved Mask R-CNN deep learning model, the targets in the construction scene are identified and located, including the components of the bridge erecting machine, bridge components and the surrounding environment; The recognition results of the targets in the construction scene are compared with a pre-built special dataset containing bridge erection machine parts and bridge components; the focal loss function is used as the loss function in the training process of the improved Mask R-CNN deep learning model, and difficult samples are dynamically selected for training through online case mining; The trained improved Mask R-CNN deep learning model is quantized and pruned to convert the 32-bit floating-point numbers in the model into 8-bit integers. The processed improved Mask R-CNN deep learning model is used to perform real-time target detection and recognition in the construction scene, and the target category, location, and contour information are output. The detection and recognition results are transmitted to the control system of the bridge erection machine.

3. The method according to claim 1, characterized in that: Based on the recognition results of the targets in the construction scene, combined with the graph structure of the topological structure of the construction scene after the spatial perception graph convolution operation, the steps of constructing a digital twin model of the construction scene include: Construct a graph structure representing the topological structure of the construction scene, wherein the nodes of the graph structure represent the identified targets or elements in the preset model, and the edges of the graph structure represent the spatial relationship between the nodes; Initialize the node feature matrix based on the constructed graph structure; Applying a spatially-aware graph convolution operation to the graph structure, the spatially-aware graph convolution operation comprising: Calculate the Euclidean distance between nodes in the graph structure; Based on the calculated Euclidean distance, the spatial distance weights between nodes are calculated using the Gaussian kernel function to obtain a spatial distance weight matrix; Constructing an adjacency matrix of a graph structure, adding self-connection to the adjacency matrix, calculating the degree matrix of the adjacency matrix after adding self-connection, and then multiplying it with the spatial distance weight matrix to obtain a weighted adjacency matrix; Apply a weighted adjacency matrix to the node feature matrix and multiply it with a learnable weight matrix; Apply a nonlinear activation function to the product result to obtain the updated node feature matrix; Based on the updated node feature matrix, feature extraction is performed on the two-dimensional target recognition results and three-dimensional model data in the construction scene; The extracted two-dimensional features and three-dimensional features are input into the multimodal feature fusion module, which uses a fully connected layer to process features and calculates fusion weights through an attention mechanism; Perform weighted fusion on the processed features according to the fusion weights to obtain fusion features; Based on the fused features and the graph structure after the spatial-aware graph convolution operation, a digital twin model of the construction scene is constructed.

4. The method according to claim 1, characterized in that: The intelligent decision-making agent generates an initial construction strategy according to the current construction scene status, optimizes the initial construction strategy using an improved Monte Carlo tree search algorithm, and generates the optimal construction strategy through multi-objective solution. The steps also include: Using intelligent decision-making agents to generate initial construction strategies based on the current state of the construction scene; The improved Monte Carlo tree search algorithm is used to optimize the initial construction strategy and generate the optimization results; Generate the optimal construction strategy based on the multi-objective optimization solution results.

5. The method according to claim 4, characterized in that The improved NSGA-III algorithm includes the steps of adaptive crossover and mutation, local search and reference point adaptation, including: Introducing an adaptive crossover and mutation mechanism based on population diversity, dynamically adjusting the crossover probability and mutation probability according to the current population diversity metric, which is calculated using the entropy weight method, and increasing the intensity of genetic operations when population diversity decreases; In each iteration of the NSGA-III algorithm, the objective function values ​​of all individuals in the current population are evaluated, a non-dominated solution set is identified and constructed, multiple solutions are selected from the non-dominated solution set for local search, and multiple simulated annealing iterations are performed on each selected solution. The simulated annealing iteration adopts an exponentially decaying temperature schedule, and a neighborhood solution is generated in each iteration and the acceptance probability is calculated according to the energy difference. When the performance of the new solution obtained by the local search is better than the original solution, the new solution is used to replace the original solution. A reference point adaptive mechanism is adopted, including uniformly generating initial reference points in the standardized target space, periodically evaluating the correlation degree of each reference point during the evolution process, using the density peak clustering algorithm for low-correlation reference points to identify high-density areas in the target space, moving low-correlation reference points to the nearest high-density area while maintaining dispersion, and introducing regularization terms to limit the large-scale migration of reference points.

6. The method according to claim 1, characterized in that During the construction process, the data of each sensor of the bridge-building machine is continuously collected, and the data of each sensor and the feedback data of each actuator are integrated to generate a state estimation of the bridge-building machine; Based on state estimation, the steps of closed-loop control of the bridge erecting machine using the nonlinear model predictive control algorithm include: Using an inertial measurement unit, a global positioning system, a laser rangefinder, a strain sensor and an actuator encoder to collect state information of the bridge erecting machine, the state information including acceleration, angular velocity, position, relative distance, structural stress, hydraulic cylinder position and motor speed; Constructing a state vector including position, acceleration, angular velocity and force / torque, and establishing a discrete-time nonlinear state equation and an observation equation based on the state vector; An extended Kalman filter is used to perform state estimation on discrete-time nonlinear state equations and observation equations, including executing a prediction step and an update step, wherein the prediction step includes state prediction and covariance prediction, and the update step includes calculating Kalman gain, state update, and covariance update; Based on the state estimation results, a predictive control optimization problem based on a nonlinear model is constructed. The optimization objectives include tracking the predetermined trajectory, minimizing energy consumption, and satisfying safety constraints. A real-time iteration strategy is adopted to solve the predictive control optimization problem, and the solution process is divided into a preparation phase and a feedback phase, wherein the preparation phase includes linearizing the system dynamics and constructing a quadratic programming subproblem, and the feedback phase includes updating the initial state, solving the quadratic programming subproblem and applying the control input; Based on the solution results of nonlinear model predictive control, the optimal control input is generated. After the optimal control input is converted into the control signal of the bridge crane actuator, the bridge crane is controlled in a closed loop.

7. The method according to claim 1, characterized in that After the construction is completed, the steps of analyzing the construction quality and efficiency using graph neural networks based on the digital twin model and the whole process data, and generating a multi-dimensional evaluation report and optimization suggestions include: The nodes of the heterogeneous graph represent components or monitoring points, and the edges represent the physical connections or logical relationships between components. Each node and edge is accompanied by a multi-dimensional feature vector. The improved graph neural network model includes node-level attention mechanism and edge-level attention mechanism. Node-level attention is used to learn the importance weights between different features, and edge-level attention is used to learn the influence of different neighbor nodes. Use historical data from the bridge erection machine construction system to train the graph neural network model, and iterate the training until convergence; Use the trained graph neural network model to analyze the quality and efficiency of the new bridge-building machine construction system data and obtain the prediction results of various indicators. Based on the prediction results, generate a multidimensional evaluation report that includes overall score, analysis of various indicators, identification of abnormal events and their root cause analysis. According to the problems and potential risks identified in the multidimensional evaluation report, combined with historical construction experience, generate optimization suggestions and improvement measures.

8. A machine vision-based intelligent construction control system for bridge erection machines, used to implement the method described in any one of claims 1 to 7, characterized in that: include: The first unit is used to collect image information of the construction site through multiple high-definition cameras installed on the bridge erection machine; input the image information into a pre-trained deep learning model to identify and locate targets in the construction scene, including various components of the bridge erection machine, bridge components and the surrounding environment; based on the recognition results of the targets in the construction scene, combined with the graph structure of the topological structure of the construction scene after the spatial perception graph convolution operation, a digital twin model of the construction scene is constructed; The second unit is used to train intelligent decision-making agents using reinforcement learning algorithms based on digital twin models of construction scenarios and predefined reward and punishment functions; The intelligent decision-making agent generates an initial construction strategy according to the current state of the construction scene, optimizes the initial construction strategy using an improved Monte Carlo tree search algorithm, and generates the optimal construction strategy through multi-objective solution; after the optimal construction strategy is converted into an intelligent control instruction, the intelligent control instruction is sent to the control system of the bridge erection machine; during the construction process, the data of each sensor of the bridge erection machine is continuously collected, and the data of each sensor and the feedback data of each actuator are integrated to generate a state estimation of the bridge erection machine; Based on state estimation, the nonlinear model predictive control algorithm is used to perform closed-loop control of the bridge erecting machine. The third unit is used to analyze the construction quality and efficiency after the construction is completed, based on the digital twin model and the data of the whole process, using the graph neural network to generate a multi-dimensional evaluation report and optimization suggestions.

9. An electronic device, characterized in that: include: processor; a memory for storing processor-executable instructions; The processor is configured to call the instructions stored in the memory to execute the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having computer program instructions stored thereon, characterized in that: When the computer program instructions are executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images