Construction management method and device for upward bridge fabrication machine

Abstract

The invention discloses a construction management method and device for an upward bridge fabrication machine, and belongs to the technical field of bridge construction management, and the method comprises the steps: carrying out the equipment assembly initialization of the upward bridge fabrication machine, and building a twin model through the upward bridge fabrication machine and an application scene; carrying out bridge hoisting path optimization based on a twin model; establishing a control constraint, taking the calibration path as an optimization target, and establishing a first control parameter; a steel wire rope tension sensor, a gyroscope, an acceleration sensor and a displacement sensor which are mounted on the ascending bridge fabrication machine are used for real-time data acquisition; establishing a second control parameter; and after the first control parameter is compensated by using a second control parameter, the ascending bridge fabrication machine is controlled to execute a bridge hoisting task. The technical problems that in construction of an existing ascending bridge fabrication machine, path planning is extensive, and control parameters are difficult to adapt to complex environment changes, so that hoisting precision is not high, and stability is poor are solved.

Description

Technical Field

[0001] The invention relates to the technical field of bridge construction management, and in particular to a construction management method and device for an upward bridge-building machine. Background Art

[0002] The upward bridge-building machine is an important special bridge construction equipment. It uses its own lifting system to lift and assemble prefabricated beam segments into bridges. Compared with traditional bridge-building machines, the upward bridge-building machine has the advantages of strong adaptability and high construction efficiency, and is especially suitable for the construction of high piers and large-span bridges.

[0003] The existing construction of upward bridge-building machines mainly relies on manual experience, which results in limited hoisting accuracy and difficulty in fully ensuring safety. In addition, traditional methods have poor adaptability to complex environments (such as wind), which can easily cause shaking and instability during the hoisting process. At the same time, the existing technology has deficiencies in monitoring, early warning, and optimization of hoisting paths, making it difficult to effectively deal with potential safety hazards and improve construction efficiency. Therefore, it is urgent to develop an intelligent construction management method that can overcome the above limitations in order to improve the construction efficiency, safety, and environmental adaptability of upward bridge-building machines. Summary of the invention

[0004] The present application provides a construction management method and device for an upward bridge-building machine, aiming to solve the technical problems in the construction of existing upward bridge-building machines, such as rough path planning and difficulty in adapting control parameters to complex environmental changes, which leads to low lifting accuracy and poor stability.

[0005] In view of the above problems, the present application provides a construction management method and device for an upward bridge-building machine.

[0006] The first aspect disclosed in the present application provides a construction management method for an upward bridge-building machine, the method comprising: after the upward bridge-building machine is initialized for equipment assembly, a twin model is established using the upward bridge-building machine and an application scenario; based on the twin model, a bridge hoisting path is optimized, a calibration path is configured with the hoisting path optimization result, and the calibration path is segmented to establish a position importance identifier of the path segmentation result; after establishing control constraints according to the path segmentation result and the position importance identifier, the calibration path is used as an optimization target, and control parameter fitting under the control constraint is performed based on a parameter optimization network to establish a first control parameter; real-time data collection is performed using a wire rope tension sensor, a gyroscope, an acceleration sensor, and a displacement sensor installed on the upward bridge-building machine to establish a real-time status data set; wind speed monitoring is performed on the application scenario to establish real-time wind speed data, and the real-time status data set and the real-time wind speed data are used as input data to perform hoisting stability optimization through a reinforcement learning channel to establish a second control parameter; after compensating the first control parameter with the second control parameter, the upward bridge-building machine is controlled to perform a bridge hoisting task.

[0007] Another aspect disclosed in the present application provides a construction management device for an upward bridge-building machine, the device comprising a twin model establishment module, which is used to establish a twin model using the upward bridge-building machine and an application scenario after the upward bridge-building machine is initialized for equipment assembly; an importance identification establishment module, which is used to optimize the bridge hoisting path based on the twin model, configure a calibration path with the hoisting path optimization result, and segment the calibration path to establish a position importance identification of the path segmentation result; a first control parameter establishment module, which is used to establish a control constraint based on the path segmentation result and the position importance identification, and then use the calibration path as the optimization target to perform optimization based on the parameter optimization network. The control parameters are fitted under the control constraints to establish the first control parameters; a state data set establishment module is used to use the wire rope tension sensor, gyroscope, acceleration sensor, and displacement sensor installed on the upward bridge-building machine to collect real-time data and establish a real-time state data set; a second control parameter establishment module is used to monitor the wind speed of the application scenario and establish real-time wind speed data, and use the real-time state data set and the real-time wind speed data as input data to perform hoisting stability optimization through a reinforcement learning channel to establish a second control parameter; a hoisting task execution module is used to control the upward bridge-building machine to perform a bridge hoisting task after compensating the first control parameter with the second control parameter.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages: By adopting the technical solutions of twin model-based path optimization, real-time sensor data acquisition and reinforcement learning dynamic compensation, the technical problems of extensive path planning and difficulty in adapting control parameters to complex environmental changes in the construction of existing upward bridge-building machines, resulting in low lifting accuracy and poor stability, are solved. The technical effect of optimizing path planning, improving the adaptability of control parameters and achieving high-precision and high-stability lifting tasks is achieved.

[0009] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 A schematic flow chart of a construction management method of an upward bridge-building machine is provided for an embodiment of the present application; Figure 2 A structural schematic diagram of a construction management device for an upward bridge-building machine is provided for an embodiment of the present application.

[0011] Explanation of the accompanying drawings: twin model establishing module 11, importance identification establishing module 12, first control parameter establishing module 13, state data set establishing module 14, second control parameter establishing module 15, lifting task execution module 16. DETAILED DESCRIPTION

[0012] The overall idea of ​​the technical solution provided by this application is as follows: The embodiments of the present application provide a construction management method and device for an upward bridge-building machine. The present application constructs an intelligent construction management method for an upward bridge-building machine by introducing technologies such as digital twins, parameter optimization networks, real-time sensor data acquisition, and reinforcement learning. The core idea is to comprehensively improve the accuracy and stability of hoisting tasks through path optimization, dynamic parameter fitting, environmental anomaly monitoring, and real-time compensation adjustment. At the same time, combined with wind speed prediction and early warning mechanisms, it can achieve safe control and intelligent operation in complex construction scenarios to adapt to changing environmental requirements.

[0013] After introducing the basic principles of the present application, various non-limiting implementation methods of the present application will be specifically described below in conjunction with the drawings in the specification.

[0014] Embodiment 1, as Figure 1 As shown, the embodiment of the present application provides a construction management method for an upward bridge-building machine, the method comprising: Step S100: After the upward bridge-building machine is initialized for equipment assembly, a twin model is established using the upward bridge-building machine and an application scenario.

[0015] Specifically, an upward bridge-building machine is a type of bridge construction machinery, which is characterized by advancing upward along the built bridge deck, hoisting and splicing prefabricated beam segments section by section. Equipment assembly initialization refers to the installation, debugging and parameter configuration of each component of the bridge-building machine before it is put into use to ensure its normal operation. For example, check whether the hydraulic system, electrical system, control system, etc. are normal, and set the initial mechanical arm position, lifting capacity and other parameters. The application scenario refers to the actual working environment of the bridge-building machine, including the type of bridge, topography, meteorological conditions, etc. The twin model refers to the use of a digital mirror of a physical model, through the fusion of multi-source information such as sensor data, physical models and historical data, to build a digital model in a virtual space that is highly consistent with the physical entity. This digital model can simulate, simulate and predict the behavior and state of physical entities.

[0016] First, it is necessary to accurately model the geometry of the real bridge-building machine, including its structure, size, kinematic parameters, etc. This can be obtained through 3D scanning, CAD drawings, etc. At the same time, it is necessary to collect data on the application scenario, including topography, pier location, size and weight of prefabricated beam segments, etc. For example, a laser scanner is used to scan the terrain of the bridge site to obtain high-precision 3D point cloud data, and then a terrain model is created through professional modeling software (such as Rhino, SolidWorks). For the bridge-building machine itself, accurate geometric parameters can be obtained by consulting design drawings or conducting on-site measurements, and modeling can be performed in CAD software.

[0017] Based on the collected data, a physical model of the bridge-building machine and the application scenario is constructed in a virtual environment. This requires the use of a physics engine (such as Unity, Unreal Engine, Gazebo) to simulate physical phenomena such as gravity, friction, and collision. For example, in Unity, the 3D model of the bridge-building machine can be imported and its physical properties such as mass and inertia can be set. At the same time, the terrain model is imported and the friction coefficient of the ground is set. Through the physics engine, the movement and force of the bridge-building machine during the lifting process can be simulated.

[0018] Connect the virtual model with the real bridge-building machine, collect the status data of the bridge-building machine in real time through sensors (such as displacement sensors, inclination sensors, force sensors, etc.), and synchronize it to the virtual model. Conversely, the control instructions in the virtual model can also be sent to the real bridge-building machine. For example, a force sensor is installed on the real bridge-building machine to monitor the tension of the wire rope. These data will be transmitted to the virtual model in real time, and the force condition of the wire rope will be displayed in the virtual model.

[0019] The twin model is used to conduct various simulation experiments, such as simulating the hoisting process under different hoisting paths and wind conditions, to optimize the hoisting plan and evaluate construction risks. For example, a wind force of 10 m / s is simulated in a virtual environment to observe the swing of the bridge crane during the hoisting process, so as to evaluate the wind resistance of the hoisting plan.

[0020] By establishing a twin model, the trial and error costs in actual construction can be effectively reduced, and construction efficiency and safety can be improved. For example, by pre-planning the path and assessing the wind impact in a virtual environment, safety accidents and construction delays caused by unreasonable paths or excessive wind can be avoided in actual construction.

[0021] Step S200: Optimizing the bridge hoisting path based on the twin model, configuring a calibration path based on the optimization result of the hoisting path, segmenting the calibration path, and establishing a position importance identifier of the path segmentation result.

[0022] Specifically, bridge hoisting path optimization refers to finding the optimal bridge hoisting path through a certain algorithm in the twin model to achieve specific optimization goals, such as the shortest path, the most stable path, the most time-saving path, etc. The calibration path refers to the reference path determined according to the optimization results and used as a reference for the actual hoisting operation. Path segmentation refers to decomposing the calibration path into several small line segments or path segments for more precise control and analysis. Position importance identification refers to assigning different weights or importance identifications to each line segment or path segment after the path segmentation to reflect their criticality in the hoisting process. For example, key positions such as lifting points and beam drop points have higher weights.

[0023] Based on the twin model, a path optimization algorithm (such as A* algorithm, genetic algorithm, ant colony algorithm, etc.) is used to search for the optimal lifting path. The optimization objectives can be varied, for example: Shortest path: reduce the lifting distance and improve construction efficiency. Most stable path: reduce swinging and shaking during the lifting process to improve safety. Obstacle avoidance path: avoid obstacles (such as bridge piers, other equipment) to ensure the smooth progress of the lifting process. For example, when using the A* algorithm for path search, it is necessary to define a suitable heuristic function, such as the straight-line distance to the target point. The algorithm will comprehensively consider the path length and the heuristic function value to search for an optimal path. In the twin model, different paths can be simulated, and the stability and force conditions during the lifting process can be calculated according to the model's physical engine to select the optimal path.

[0024] The best path obtained by optimization is used as the calibration path, which will serve as the benchmark for the actual lifting operation. This step is equivalent to mapping the optimization results in the virtual environment to the actual operation. For example, if a smooth and stable lifting path is obtained by optimization in the virtual environment, then this path will be saved for subsequent path segmentation and control parameter generation. The calibration path is divided into several small line segments or path segments. The segmentation method can be equidistant segmentation or segmentation based on the curvature of the path, with denser segmentation in places with larger curvature. For example, a curved path can be divided into several straight line segments, and the length of each straight line segment can be adjusted as needed. In places with larger path curvature, a smaller segmentation length can be used to improve control accuracy.

[0025] According to the path segmentation results, different weights or importance identifiers are assigned to each line segment or path segment. For example, the lifting point and the beam drop point are key locations in the lifting process, so their weights are higher. In the straight part of the path, the weight can be relatively low. The importance identifier can be expressed in a numerical value or symbol. For example, a value between 0 and 1 can be used to represent the importance, with 1 being the most important and 0 being the least important. The importance of the lifting point and the beam drop point can be set to 0.9, while the importance of the straight part can be set to 0.5.

[0026] For example, during the hoisting process of a cross-river bridge, the impact of water flow needs to be considered. Through the twin model for path optimization, a hoisting path with the least impact of water flow can be found and configured as the calibration path. The calibration path is then segmented, and the path segments close to the water surface are given a higher importance mark to strengthen the monitoring and control of these areas. During the hoisting process of a high-pier bridge, the impact of wind needs to be considered. Through the twin model for path optimization, a hoisting path with the least impact of wind force can be found. The calibration path is then segmented, and the path segments in the high-altitude working area are given a higher importance mark to improve the safety of high-altitude working.

[0027] Through this step, an optimized and highly operable lifting path can be obtained, and key positions on the path can be focused on to improve the accuracy, stability and safety of the lifting.

[0028] Step S300: After establishing control constraints according to the path segmentation result and the position importance identifier, the calibrated path is used as an optimization target, and control parameter fitting under the control constraints is performed based on a parameter optimization network to establish first control parameters.

[0029] Specifically, control constraints refer to various restrictions on the movement of the bridge-building machine, including speed constraints, acceleration constraints, position accuracy constraints, attitude constraints, etc. These constraints are to ensure the safety and stability of the lifting process. For example, limit the maximum lifting speed, limit the maximum swing angle of the wire rope, etc. Parameter optimization network refers to a machine learning model used to learn and optimize control parameters. It can automatically adjust the parameters of the model according to input data (such as calibration path, control constraints) to achieve the best control effect. Common parameter optimization networks include neural networks, genetic algorithms, etc. Control parameter fitting refers to the calculation of a series of control parameters through the parameter optimization network according to the calibration path and control constraints, so that the bridge-building machine can move accurately along the calibration path and meet various constraints.

[0030] According to the path segmentation results and position importance identification, combined with the actual lifting requirements, establish corresponding control constraints. For example: Speed ​​constraint: In the straight part of the path, a higher lifting speed can be allowed; when turning or approaching the target position, the speed needs to be reduced to ensure stability. According to the position importance identification, more stringent speed constraints should be set for high-importance positions. Position accuracy constraint: At key positions such as lifting points and beam dropping points, higher position accuracy is required to ensure accurate installation of the beam segment. Posture constraint: Limit the swing angle of the wire rope to prevent the beam segment from shaking too much. These constraints can be expressed in the form of mathematical formulas or rules. For example, the speed constraint can be expressed as V≤V max(s) , where v is the actual speed, V max(s) is the maximum velocity associated with the path position s.

[0031] Select a suitable parameter optimization network according to the specific control problem. For example, a deep neural network can be used to learn complex nonlinear control laws, or a genetic algorithm can be used to search for the optimal control parameter combination. The calibration path and control constraints are used as input data and input into the parameter optimization network for training or optimization. By adjusting the network parameters, the control parameters output by it can enable the bridge-building machine to move accurately along the calibration path and meet various constraints. For example, if a neural network is used for fitting, the position, speed and other information of the path point can be used as input, and the voltage, current and other information of the control motor can be used as output. By training the neural network, it can learn the mapping relationship between the path point and the control parameters. If a genetic algorithm is used, the control parameters can be encoded into chromosomes, and the optimal chromosome, that is, the optimal control parameter combination, can be searched through operations such as selection, crossover, and mutation. The control parameters output by the parameter optimization network are saved as the first control parameters.

[0032] Through this step, a set of preliminary control parameters can be automatically generated based on the calibration path and control constraints, so that the bridge crane can move according to the predetermined trajectory and meet various safety and performance requirements. This method can better handle complex nonlinear control problems and can adapt to different lifting scenarios and constraints.

[0033] Step S400: Utilize the wire rope tension sensor, gyroscope, acceleration sensor and displacement sensor installed on the upward bridge-building machine to collect real-time data and establish a real-time status data set.

[0034] Specifically, a wire rope tension sensor refers to a sensor used to measure the tension on a wire rope. Its output signal reflects the force applied to the wire rope, which is essential for monitoring load changes and safety during the lifting process. A gyroscope is a sensor used to measure the angular velocity or angular acceleration of an object. In a bridge-building machine, a gyroscope can be used to measure the rotation angle and angular velocity of a boom or spreader to monitor its posture changes. An accelerometer is a sensor used to measure the acceleration of an object. It can measure the acceleration of a bridge-building machine in three axial directions, and is used to detect vibration, shock, and other conditions, as well as to assist in calculating speed and displacement. A displacement sensor is a sensor used to measure changes in the position of an object. In a bridge-building machine, a displacement sensor can be used to measure the lifting height, horizontal movement distance, etc. of a hook or spreader, so as to accurately control its position. A real-time status data set refers to a data set formed by integrating, storing, and processing various sensor data collected in real time, and is used to monitor, analyze, and control the operating status of a bridge-building machine.

[0035] The wire rope tension sensor is installed on the wire rope. The gyroscope and acceleration sensor are usually installed on the key parts of the boom or the spreader. The displacement sensor is installed at the location where the displacement needs to be measured, such as the lifting mechanism of the hook. When installing, the sensor's range, accuracy and installation method need to be considered to ensure the accuracy and reliability of the measurement results. For example, the tension sensor needs to select a suitable range to cover the maximum tension that occurs during the lifting process.

[0036] Build a data acquisition system, including sensor signal conditioning circuit, analog-to-digital converter (ADC), data acquisition card, data storage device, etc. The sensor signal conditioning circuit is used to amplify, filter and convert the weak signal output by the sensor to adapt it to the input range of the ADC. The ADC converts the analog signal into a digital signal, and the data acquisition card transmits the digital signal to a computer or other data processing device. The data storage device is used to store the collected data.

[0037] Set the parameters for data acquisition, including sampling frequency, acquisition channels, data storage format, etc. The sampling frequency needs to be determined based on the frequency components of the measured signal, and usually needs to satisfy the Nyquist sampling theorem. The acquisition channels correspond to the number and type of sensors. The data storage format can be selected as text file, CSV file, binary file, etc.

[0038] The collected raw data is processed, such as filtering, correction, unit conversion, etc. Filtering is used to remove noise interference, correction is used to eliminate the error of the sensor itself, and unit conversion is used to convert the raw data into engineering units. The processed data is stored in a database or file to establish a real-time status data set. For example, the Kalman filter can be used to fuse the data of the accelerometer and gyroscope to obtain more accurate attitude information. You can use a database (such as MySQL, InfluxDB) or a time series database (such as TimescaleDB) to store time series data.

[0039] Through real-time data collection and the establishment of status data sets, comprehensive monitoring and precise control of the operating status of the bridge-building machine can be achieved, improving the safety and efficiency of construction: through real-time monitoring of key parameters, potential safety hazards can be discovered in a timely manner.

[0040] Step S500: Monitor the wind speed of the application scenario, establish real-time wind speed data, use the real-time status data set and the real-time wind speed data as input data, perform hoisting stability optimization through a reinforcement learning channel, and establish a second control parameter.

[0041] Specifically, real-time wind speed data refers to wind speed data continuously collected and recorded by wind speed monitoring equipment, usually including the size and direction of wind speed. Reinforcement learning channel refers to the path or method of optimization using reinforcement learning algorithm. Reinforcement learning is a machine learning method that learns how to take the best action in a specific environment to obtain the maximum cumulative reward through the interaction between the agent and the environment. In this example, the control system of the bridge-building machine is used as the agent, the hoisting process and its environment (including wind speed, the state of the bridge-building machine itself, etc.) are used as the environment, and the stability and efficiency of the hoisting are used as the reward. Hoisting stability optimization refers to finding a control strategy that can maximize the hoisting stability under different wind speed conditions through reinforcement learning algorithm. The second control parameter refers to the control parameter obtained by reinforcement learning optimization. Unlike the first control parameter (the parameter obtained based on the calibration path and the control constraint fitting), it takes into account the influence of wind speed and aims to improve the wind resistance and stability of the hoisting.

[0042] First, install wind speed sensors (such as cup anemometers and ultrasonic anemometers) near the bridge-building machine or around the bridge site to monitor the size and direction of the wind speed in real time. Process and store the collected wind speed data to establish a real-time wind speed data set. For example, a weather station or special meteorological monitoring equipment can be used to monitor wind speed, and the data can be transmitted to a computer through a data acquisition system. Integrate the motion model of the bridge-building machine, the physical model of the hoisting process, and the wind speed model into a reinforcement learning environment. This environment needs to be able to simulate the hoisting process of the bridge-building machine at different wind speeds, and give corresponding rewards based on the stability and efficiency of the hoisting.

[0043] Design a suitable reward function to evaluate the stability and efficiency of the lifting process. The reward function should be able to reflect the key indicators of the lifting process, such as: Swing amplitude: the smaller the swing amplitude, the higher the reward. Displacement deviation: the smaller the deviation between the actual position and the target position, the higher the reward. Lifting time: the shorter the lifting time, the higher the reward. Safety constraints: if the safety constraints are violated (for example, the wire rope tension exceeds the threshold), a negative reward or penalty is given.

[0044] The control system is trained using reinforcement learning algorithms (such as Q-learning, Deep Q-Network (DQN), Proximal Policy Optimization (PPO), etc.). During the training process, the control system (agent) of the bridge crane interacts with the environment, selects corresponding control actions (such as adjusting the motor speed, wire rope length, etc.) according to the current state (including real-time state data sets and real-time wind speed data), and continuously adjusts its own control strategy according to the feedback (rewards) from the environment until convergence is achieved or the expected performance indicators are achieved. For example, a deep neural network can be used as a policy network for reinforcement learning, with state information as input and control actions as output. By continuously interacting with the environment, the network will gradually learn the optimal control strategy under different wind speeds.

[0045] After the training is completed, the control strategy obtained by reinforcement learning is extracted as the second control parameter, which is used to output the corresponding control instructions according to the current state and wind speed information.

[0046] By introducing reinforcement learning to optimize the hoisting stability, the performance of the hoisting system in complex wind environments can be significantly improved. Construction delays caused by weather conditions can be reduced. By learning control strategies that adapt to different wind conditions, the reliance on manual intervention can be reduced and the degree of automation can be improved.

[0047] Step S600: After compensating the first control parameter with the second control parameter, the upward bridge-building machine is controlled to perform a bridge hoisting task.

[0048] Specifically, the real-time status data of the bridge-building machine is continuously obtained, including wire rope tension, boom angle, position, etc. According to the current working conditions and calibration path, the calculation model of the first control parameter is used to calculate the control instructions at the current moment. Based on the reinforcement learning model, the current real-time state (including external factors such as wind speed) is input into the model to obtain the second control parameter. The second control parameter is used as a correction to compensate the first control parameter. The compensation method can be a simple weighted average or a more complex fusion algorithm, such as Kalman filtering. The compensated control parameters are sent to the actuator (such as a motor, hydraulic valve) to control the action of the bridge-building machine.

[0049] For example, when lifting a bridge at high altitude and in strong winds, the first control parameter cannot fully adapt to environmental changes, resulting in oscillation or position deviation during the lifting process. At this time, the second control parameter obtained through reinforcement learning can adjust the control instructions in real time, so that the lifting system can better adapt to wind interference and improve the stability of the lifting.

[0050] By compensating the first control parameter, the control performance of the upward bridge-building machine can be effectively improved, so that it can complete the lifting task more stably and accurately under complex working conditions.

[0051] Furthermore, the control parameter fitting under the control constraints is performed based on the parameter optimization network to establish the first control parameters, including: obtaining the initial position of the bridge, taking the initial position as the basic position, and initializing the parameter optimization network; obtaining the speed constraint and the accuracy constraint in the control constraints, configuring the preview distance according to the speed constraint and the accuracy constraint, and selecting the preview point on the calibration path by using the preview distance, taking the basic position as the starting point to perform approximate fitting of the preview point, and establishing the fitting control parameters; performing continuous fitting by using the parameter optimization network, and establishing the first control parameters according to all the fitting control parameters.

[0052] Specifically, the initial position refers to the initial position and posture of the bridge-building machine when it starts the lifting operation. The basic position uses the initial position as a reference for the subsequent fitting process. The speed constraint refers to limiting the maximum speed of the bridge-building machine to ensure safety and stability. The accuracy constraint refers to the accuracy that the bridge-building machine needs to achieve when it reaches the target position, such as the allowable range of position error and posture error. The preview distance refers to a distance that is observed or "pre-aimed" in advance on the calibration path, with the current position as the reference. Through previewing, the direction of the path can be predicted in advance and corresponding control adjustments can be made. The preview point refers to the end point of the preview distance, that is, the target position of the preview. Approximate fitting refers to adjusting the control parameters to gradually approach the preview point of the bridge-building machine. The fitting control parameters refer to the control parameters output by the parameter optimization network during the approximate fitting process. Continuous fitting refers to repeating the process of previewing and approximate fitting until the entire calibration path is tracked.

[0053] First, obtain the initial position of the bridge (actually the bridge-building machine) (for example, obtained through GPS or laser positioning system), use it as the basic position, and initialize the parameter optimization network (for example, randomly initialize the network weights). Configure the preview distance according to the speed constraint and the accuracy constraint. For example, if the speed constraint is low, the preview distance can be shorter; if the accuracy constraint is high, the preview distance should also be shortened accordingly. Generally, the selection of the preview distance needs to be adjusted and optimized according to the actual situation. Take the basic position as the starting point and select a preview point on the calibration path. Use the parameter optimization network, take the basic position and the preview point as input, perform approximate fitting, and calculate the corresponding fitting control parameters. For example, a neural network can be used as a parameter optimization network, input information such as the current position and the preview point position, output the voltage, current and other parameters of the control motor, and make the bridge-building machine move to the preview point. The approximate fitting process can use various optimization algorithms, such as gradient descent method, genetic algorithm, etc.

[0054] The end point of the last approximate fitting is used as the new basic position. Repeat the above steps, that is, select a new preview point on the calibration path and perform approximate fitting until the entire calibration path is tracked. In each fitting process, the parameter optimization network will output a set of fitting control parameters. All fitting control parameters obtained in the continuous fitting process are combined to form the first control parameter. This parameter set contains the control information required to complete the entire lifting process.

[0055] Through this step, the parameter optimization network can be used to automatically generate a set of control parameters suitable for the entire lifting process based on the calibration path and control constraints. This method can adapt to calibration paths of different shapes and different control constraints. The lifting accuracy can be improved through continuous fitting and precision constraints. The anti-interference ability of the system can be improved by reasonably configuring the preview distance and control constraints.

[0056] Furthermore, the method of using the parameter optimization network to perform continuous fitting and establishing the first control parameter according to all the fitting control parameters includes: configuring a continuous space using the path segmentation result; establishing a continuity constraint in the continuous space; using the approximate fitting result as an update starting point, performing continuous fitting using the update starting point and the continuity constraint, and establishing a new fitting control parameter; when any continuous fitting triggers the constraint end point of the continuity constraint, the current continuous fitting is terminated and continuous fitting is re-executed; and establishing the first control parameter according to all the fitting control parameters.

[0057] Specifically, continuous space refers to the space defined by the path segmentation result, in which the bridge-building machine needs to move continuously to complete the lifting task. This space is formed by connecting discrete path points, and is not a continuous space in the mathematical sense. Continuity constraints refer to constraints set to ensure the smoothness and continuity of the bridge-building machine's movement, such as velocity continuity, acceleration continuity, etc. It ensures that the movement of the bridge-building machine will not experience sudden changes or jumps. Approximate fitting results refer to the fitting control parameters calculated by the parameter optimization network based on the current position and preview point, as well as the new position to which the bridge-building machine moves under the control of the parameters. This new position is used as the "update starting point". The update starting point refers to the starting position of the next approximate fitting, which is determined by the result of the previous approximate fitting. The constraint end point refers to the end point of the range of motion allowed by the continuity constraint. If the update starting point exceeds this range, continuous fitting needs to be restarted. Newly added fitting control parameters refer to the control parameters newly generated for each approximate fitting during the continuous fitting process.

[0058] According to the path segmentation results, the segmented line segments or path segments are connected in sequence to form a continuous motion space. This space is actually composed of a series of discrete points, and the points are connected by line segments. For example, if the path is divided into 10 line segments, then these 10 line segments constitute a continuous space, and the bridge-building machine needs to pass through these line segments in sequence. In the continuous space, continuity constraints are established. These constraints usually include: Position continuity: ensure that the bridge-building machine moves in sequence along the segmented path without skipping any path point. Speed ​​continuity: ensure that the speed of the bridge-building machine changes smoothly and avoids sudden changes in speed. For example, the derivative of the speed (acceleration) can be used to set an upper limit. Acceleration continuity: further constrain the rate of change of the speed, avoid sudden changes in acceleration, and ensure smoother movement. These constraints can be expressed in the form of mathematical formulas or rules. For example, speed continuity can be expressed as the speed difference between two adjacent control cycles is less than a certain threshold.

[0059] Use the result of the last approximate fitting (i.e., the update starting point) as the new starting point, use the parameter optimization network to perform the next approximate fitting, and calculate the newly added fitting control parameters. Determine whether the update starting point exceeds the constraint end point of the continuity constraint. For example, determine whether the next preview point exceeds the end point of the current path segment, or whether the speed / acceleration exceeds the set threshold. If the update starting point exceeds the constraint end point of the continuity constraint, the current continuous fitting is terminated. This means that the fitting of the current path segment is completed and the fitting of the next path segment needs to be started. Re-execute the continuous fitting process using the end point of the previous path segment as the new basic position until the entire calibration path is tracked. Combine the newly added fitting control parameters obtained in all continuous fitting processes in sequence to form the final first control parameters.

[0060] By introducing continuity constraints, the smoothness and stability of the bridge-building machine during movement can be guaranteed, and unexpected situations caused by sudden changes in control parameters can be avoided. Through continuous fitting, the calibration path can be tracked more accurately. Continuity constraints can avoid sudden changes and jitters during movement. External interference and uncertainty can be better dealt with.

[0061] Furthermore, the method of performing hoisting stability optimization through a reinforcement learning channel and establishing a second control parameter includes: establishing a reward function through the reinforcement learning channel, and the reward function is as follows: ; in, Represents the reward value, and are the weight factors for force uniformity and hoisting stability, Characterizes the standard deviation of the wire rope tension, is the total number of wire ropes, Characterize any wire rope, Characterization The tension of the wire rope, is the maximum swing angle, is the threshold of the swing angle, Characterizes position deviation, is the maximum position deviation threshold; the reward function is used to perform hoisting stability optimization and establish the second control parameter.

[0062] Specifically, the reward function is a function used to evaluate the behavior of the system in the current state. The higher the reward function, the better the system performance. Force uniformity refers to whether the force on each wire rope is balanced. The better the uniformity, the more stable the system. Hoisting stability refers to the stability of the system during the hoisting process, including the swing amplitude and position deviation of the hoisted object.

[0063] The state space consists of various state variables of the system, such as wire rope tension, hoisting object angle, position, etc. The action space represents all actions that the system can take, that is, the range of control parameters. According to the definition in the formula, calculate the reward value. The higher the reward value, the better the system state. Let the reinforcement learning model be trained in a simulated environment, and learn the optimal control strategy through trial and error. Deploy the trained model to the actual bridge-building machine and output the corresponding control instructions according to the real-time status.

[0064] By designing a reasonable reward function, the force uniformity and hoisting stability are quantified, and the optimal control parameters are found through the reinforcement learning algorithm, so as to achieve the stability and efficiency of the hoisting process. This method has strong adaptability and robustness and can be applied to various complex hoisting scenarios.

[0065] Furthermore, the method also includes: performing data backtracking on the real-time wind speed data to establish a backtracked wind speed data set; performing a wind speed steady-state analysis using the backtracked wind speed data set to establish a first steady-state analysis result; obtaining a wind speed prediction result, the wind speed prediction result having a prediction confidence, performing a wind speed steady-state analysis based on the wind speed prediction result and the prediction confidence to establish a second steady-state analysis result; establishing pre-compensation parameters based on the first steady-state analysis result and the second steady-state analysis result; and performing control compensation of the upward bridge-building machine using the pre-compensation parameters.

[0066] Specifically, data backtracking is based on time series, which traces back historical wind speed data to a specific time point or time period to form a wind speed data set for analysis. The backtracking wind speed data set refers to the set of historical wind speed data obtained through backtracking, which is used to evaluate the changing trend and steady-state characteristics of the ambient wind speed. Wind speed steady-state analysis refers to the analysis of the stability of wind speed data (such as fluctuation range, rate of change) to determine whether the wind speed is in an acceptable steady-state range. Wind speed prediction results refer to the use of prediction algorithms (such as time series prediction models) to predict future wind speed changes. The prediction results include wind speed values ​​and their credibility. Prediction confidence refers to the evaluation of the reliability of wind speed prediction results, usually expressed in the form of probability or confidence interval. Pre-compensation parameters refer to parameters calculated by combining steady-state analysis results and prediction data, which are used to adjust the control strategy of the bridge-building machine to reduce the impact of wind speed changes on the lifting task.

[0067] First, obtain real-time wind speed data from the wind speed monitoring device, and generate a retrospective wind speed data set based on the time series back-tracking historical data. Based on the retrospective wind speed data, the wind speed steady-state analysis is performed through statistical analysis (such as standard deviation and rate of change calculation) to obtain the first steady-state analysis result (stability assessment of the current ambient wind speed). Subsequently, machine learning or deep learning algorithms (such as LSTM time series models) are used to predict future wind speed changes, and the reliability of the prediction results is evaluated in combination with credibility, and a second steady-state analysis is performed to generate the second steady-state analysis result.

[0068] Next, based on the results of the first and second steady-state analyses, the pre-compensation parameters are calculated by weight fusion. For example, when the first steady-state analysis shows that the wind speed fluctuates slightly, while the second steady-state analysis predicts a large change in wind speed, the safety margin of the hoisting speed can be increased, or the control strategy can be adjusted to reduce the swing amplitude. Finally, the pre-compensation parameters are input into the control system to make compensatory adjustments to the upward bridge-building machine to ensure the stability of the hoisting task.

[0069] Through the analysis method combining real-time wind speed data backtracking and prediction, this step can accurately evaluate the current wind speed status and future change trends, dynamically generate control compensation parameters, and effectively improve the stability and safety of hoisting tasks in complex environments. In particular, by introducing prediction trust and multi-level steady-state analysis, the reliability of pre-compensation parameters is enhanced, allowing the bridge crane to efficiently complete the hoisting task in wind speed changes, while reducing safety risks and construction errors caused by environmental factors.

[0070] Furthermore, the method also includes: obtaining compensation results and establishing a trigger warning threshold mapped to the compensation results; when using the compensation results to control the upward bridge-building machine to perform a bridge lifting task, collecting responses through the wire rope tension sensor, gyroscope, acceleration sensor, and displacement sensor to establish a response data set; performing response authentication on the response data set through the trigger warning threshold to establish a response warning.

[0071] Specifically, the compensation result refers to the adjustment parameters generated by combining real-time data analysis and control strategy optimization, which are used to adapt to environmental changes and optimize equipment operation. The trigger warning threshold is the critical value of the parameter generated by mapping the compensation result. When the actual parameters of the equipment during operation exceed this threshold, the warning mechanism is triggered. The response data set refers to the real-time operating parameter set collected by various sensors, which is used to analyze the operating status of the equipment. Response authentication refers to comparing the response data set with the trigger warning threshold to determine whether the operating parameters exceed the safe range and generate a warning signal.

[0072] First, the compensation results are obtained, that is, the control parameters optimized by dynamically changing conditions such as wind speed and path, such as adjustment speed, wire rope tension, etc. On this basis, the trigger warning threshold associated with the compensation results is established to define the safety boundary of equipment operation.

[0073] Subsequently, when performing the lifting task, the equipment operation data is collected in real time through the wire rope tension sensor, gyroscope, acceleration sensor and displacement sensor to form a response data set (such as tension, swing angle, acceleration and position deviation). The various parameters in the response data set are compared with the corresponding trigger warning threshold, and the response authentication mechanism is used to determine whether there are safety hazards. For example, if the actual tension is 460N, which exceeds the trigger threshold of 450N, a response warning signal is generated.

[0074] By obtaining compensation results and generating trigger warning thresholds, this method can achieve dynamic real-time monitoring and rapid response authentication during the execution of the task. When parameter changes that exceed the safety boundary occur during the lifting process, the system can generate warnings in time to remind operators or automatically adjust the equipment control strategy, thereby significantly improving the safety and stability of the lifting task, reducing potential accident risks and improving construction efficiency.

[0075] Furthermore, the method also includes: synchronously activating a visual sensing device, and utilizing the visual sensing device to perform scene attention monitoring of the application scenario, wherein the scene attention monitoring includes dynamic scene attention monitoring and dynamic scene attention monitoring; establishing a scene anomaly according to the scene attention monitoring result, generating a reconstruction control parameter through the scene anomaly, and utilizing the reconstruction control parameter to control the upward bridge-building machine to continue to perform the bridge lifting task.

[0076] Specifically, visual sensing equipment is used to collect images or video data of the construction environment, including cameras, depth cameras, lidar, etc., which can perceive scene information in real time. Dynamic scene attention refers to real-time monitoring of changing scenes, such as the movement trajectory of equipment or the range of workers' activities during lifting. Static scene attention refers to monitoring key features in fixed scenes, such as the location of obstacles in the lifting path. Scene anomalies refer to abnormal situations identified based on scene attention monitoring results, including abnormal equipment operation, sudden changes in the environment, etc. Reconstructed control parameters refer to control parameters regenerated in response to scene anomalies, which are used to adjust equipment operation and restore normal execution of tasks.

[0077] During the bridge hoisting task, the visual sensing equipment is activated synchronously to collect real-time images or video data of the application scene. Scene attention monitoring is divided into two parts: dynamic and static. Dynamic scene attention is used to analyze the stability and safety of the equipment's motion trajectory in real time, such as the swing angle of the wire rope and the offset of the hoisted object; static scene attention focuses on the key features of the fixed scene in the hoisting path, such as whether there are new obstacles around the bridge piers or whether the path markings are clear. Use image processing algorithms (such as OpenCV) and target detection models (such as YOLO or Mask R-CNN) to analyze the collected data in real time to identify potential anomalies.

[0078] When the scene monitoring results show an abnormality (such as path obstacles, new safety hazards), the system generates a scene abnormality and triggers the generation of reconstruction control parameters. This process can recalculate the adjusted operating parameters through control algorithms (such as dynamic programming or reinforcement learning models), such as the path to bypass obstacles or adjust the angle and direction of the equipment posture. Finally, the new reconstructed control parameters are used to adjust the operating state of the upward bridge-building machine so that it can continue to perform the lifting task.

[0079] Through scene attention monitoring by visual sensing equipment, this method can perceive the dynamic changes and potential risks in the lifting task in real time. After discovering the scene anomaly, the system automatically generates and applies the reconstruction control parameters, so that the lifting task can quickly adapt to sudden changes in the environment and continue to execute, avoiding interruption of operation and safety accidents.

[0080] In summary, the construction management method of an upward bridge-building machine provided in the embodiment of the present application has the following technical effects: 1. Through twin models, path optimization, real-time data collection and reinforcement learning optimization, this method realizes accurate planning of the lifting path and dynamic compensation of control parameters, effectively improving the efficiency, stability and safety of the lifting task of the upward bridge-building machine.

[0081] 2. The control parameters are dynamically adjusted by continuous fitting and constraint termination conditions, so that the continuity and consistency of the control parameters on the path can be guaranteed, and the stability and efficiency of the lifting operation are optimized.

[0082] 3. Combined with real-time wind speed data backtracking and forecast analysis, pre-compensation parameters are generated to proactively respond to the impact of wind speed fluctuations on lifting tasks, significantly improving the safety and reliability of the lifting process.

[0083] Embodiment 2, based on the same inventive concept as the construction management method of an upward bridge-building machine in the above embodiment, Figure 2 As shown, the embodiment of the present application provides a construction management device for an upward bridge-building machine, the device comprising: The twin model establishment module 11 is used to establish a twin model using the upward bridge-building machine and the application scenario after the equipment assembly of the upward bridge-building machine is initialized; the importance identification establishment module 12 is used to optimize the bridge hoisting path based on the twin model, configure the calibration path with the hoisting path optimization result, and perform path segmentation on the calibration path to establish the position importance identification of the path segmentation result; the first control parameter establishment module 13 is used to establish the control constraint according to the path segmentation result and the position importance identification, take the calibration path as the optimization target, perform control parameter fitting under the control constraint based on the parameter optimization network, and establish the first control parameter. control parameters; a state data set establishment module 14, which is used to use the wire rope tension sensor, gyroscope, acceleration sensor, and displacement sensor installed on the upward bridge-building machine to collect real-time data and establish a real-time state data set; a second control parameter establishment module 15, which is used to monitor the wind speed of the application scenario, establish real-time wind speed data, use the real-time state data set and the real-time wind speed data as input data, perform hoisting stability optimization through a reinforcement learning channel, and establish a second control parameter; a hoisting task execution module 16, which is used to control the upward bridge-building machine to perform a bridge hoisting task after compensating the first control parameter with the second control parameter.

[0084] Furthermore, the first control parameter establishment module 13 is also used to perform the following steps: obtain the initial position of the bridge, use the initial position as the basic position, and initialize the parameter optimization network; obtain the speed constraint and the accuracy constraint in the control constraints, configure the preview distance according to the speed constraint and the accuracy constraint, and use the preview distance to select a preview point in the calibration path, use the basic position as the starting point to perform approximate fitting of the preview point, and establish fitting control parameters; use the parameter optimization network to perform continuous fitting, and establish the first control parameters according to all the fitting control parameters.

[0085] Furthermore, the first control parameter establishment module 13 is also used to perform the following steps: configure a continuous space using the path segmentation result; establish a continuity constraint in the continuous space; use the approximate fitting result as an update starting point, perform continuous fitting using the update starting point and the continuity constraint, and establish new fitting control parameters; when any continuous fitting triggers the constraint end point of the continuity constraint, the current continuous fitting is terminated and continuous fitting is re-executed; and establish the first control parameter based on all fitting control parameters.

[0086] Furthermore, the second control parameter establishment module 15 is further used to perform the following steps: establish a reward function through the reinforcement learning channel, and the reward function is as follows: ; in, Represents the reward value, and are the weight factors for force uniformity and hoisting stability, Characterizes the standard deviation of the wire rope tension, is the total number of wire ropes, Characterize any wire rope, Characterization The tension of the wire rope, is the maximum swing angle, is the threshold of the swing angle, Characterizes position deviation, is the maximum position deviation threshold; the reward function is used to perform hoisting stability optimization and establish the second control parameter.

[0087] Furthermore, the device is also used to perform the following steps: perform data backtracking on the real-time wind speed data to establish a backtracked wind speed data set; perform a wind speed steady-state analysis using the backtracked wind speed data set to establish a first steady-state analysis result; obtain a wind speed prediction result, the wind speed prediction result has a prediction confidence level, perform a wind speed steady-state analysis based on the wind speed prediction result and the prediction confidence level, and establish a second steady-state analysis result; establish pre-compensation parameters based on the first steady-state analysis result and the second steady-state analysis result; and perform control compensation of the upward bridge-building machine using the pre-compensation parameters.

[0088] Furthermore, the device is also used to perform the following steps: obtain compensation results and establish a trigger warning threshold mapped to the compensation results; when using the compensation results to control the upward bridge-building machine to perform a bridge lifting task, collect responses through the wire rope tension sensor, gyroscope, acceleration sensor, and displacement sensor to establish a response data set; authenticate the response data set through the trigger warning threshold and establish a response warning.

[0089] Furthermore, the device is also used to perform the following steps: synchronously activate the visual sensing device, and use the visual sensing device to perform scene attention monitoring of the application scenario, the scene attention monitoring including dynamic scene attention monitoring and dynamic scene attention monitoring; establish scene anomalies according to the scene attention monitoring results, generate reconstruction control parameters through the scene anomalies, and use the reconstruction control parameters to control the upward bridge-building machine to continue to perform the bridge lifting task.

[0090] Any step of the method described above can be stored as a computer instruction or program in an unlimited computer memory, and can be called and recognized by an unlimited computer processor to implement any method in the embodiments of the present application, without any unnecessary restrictions.

[0091] Furthermore, the first or second mentioned above not only represents an order relationship, but also represents a specific concept, and / or refers to the selection of multiple elements individually or in whole. Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalent technologies, the present application intends to include these modifications and variations.

Claims

1. A construction management method for an upward bridge-building machine, characterized in that: The method comprises: After the upward bridge-building machine is initialized for equipment assembly, a twin model is established using the upward bridge-building machine and the application scenario; Optimizing the bridge hoisting path based on the twin model, configuring a calibration path based on the optimization result of the hoisting path, segmenting the calibration path, and establishing a position importance identifier of the path segmentation result; After establishing control constraints according to the path segmentation result and the position importance identifier, taking the calibrated path as the optimization target, performing control parameter fitting under the control constraints based on the parameter optimization network, and establishing first control parameters; Using the wire rope tension sensor, gyroscope, acceleration sensor and displacement sensor installed on the upward bridge-building machine to collect real-time data and establish a real-time status data set; Monitor the wind speed of the application scenario, establish real-time wind speed data, use the real-time state data set and the real-time wind speed data as input data, perform hoisting stability optimization through a reinforcement learning channel, and establish a second control parameter; After compensating the first control parameter with the second control parameter, the upward bridge-building machine is controlled to perform the bridge hoisting task.

2. A construction management method for an upward bridge-building machine as claimed in claim 1, characterized in that: The step of fitting the control parameters under the control constraints based on the parameter optimization network to establish the first control parameters includes: Obtaining an initial position of the bridge, using the initial position as a base position, and initializing a parameter optimization network; Acquire the speed constraint and the accuracy constraint in the control constraint, configure the preview distance according to the speed constraint and the accuracy constraint, select the preview point in the calibration path by using the preview distance, perform approximate fitting of the preview point with the basic position as the starting point, and establish fitting control parameters; The parameter optimization network is used to perform continuous fitting, and the first control parameter is established according to all the fitting control parameters.

3. A construction management method for an upward bridge-building machine as claimed in claim 2, characterized in that: The method of using the parameter optimization network to perform continuous fitting and establishing the first control parameter according to all fitting control parameters includes: configuring a continuous space using the path segmentation result; establishing continuity constraints within the continuous space; Taking the approximate fitting result as the update starting point, performing continuous fitting using the update starting point and the continuity constraint, and establishing a newly added fitting control parameter; When any continuous fitting triggers the constraint end point of the continuity constraint, the current continuous fitting is terminated and the continuous fitting is executed again; The first control parameter is established according to all the fitting control parameters.

4. A construction management method for an upward bridge-building machine as claimed in claim 1, characterized in that: The method of performing hoisting stability optimization through a reinforcement learning channel and establishing a second control parameter includes: A reward function is established through the reinforcement learning channel, and the reward function is as follows: ； in, Represents the reward value, and are the weight factors for force uniformity and hoisting stability, Characterizes the standard deviation of the wire rope tension, is the total number of wire ropes, Characterize any wire rope, Characterization The tension of the wire rope, is the maximum swing angle, is the threshold of the swing angle, Characterizes position deviation, is the maximum position deviation threshold; The reward function is used to perform hoisting stability optimization and establish the second control parameters.

5. The construction management method of an upward bridge-building machine according to claim 1, characterized in that: The method further comprises: Performing data backtracking on the real-time wind speed data to establish a backtracking wind speed data set; Performing a wind speed steady-state analysis using the retrospective wind speed data set to establish a first steady-state analysis result; Obtaining a wind speed prediction result, the wind speed prediction result having a prediction confidence, performing a wind speed steady-state analysis according to the wind speed prediction result and the prediction confidence, and establishing a second steady-state analysis result; Establishing pre-compensation parameters according to the first steady-state analysis result and the second steady-state analysis result; Control compensation of the upward bridge-building machine is performed using the pre-compensation parameters.

6. A construction management method for an upward bridge-building machine as claimed in claim 1, characterized in that: The method further comprises: Obtaining compensation results, and establishing a trigger warning threshold mapped to the compensation results; When the compensation result is used to control the upward bridge-building machine to perform the bridge hoisting task, the response is collected through the wire rope tension sensor, gyroscope, acceleration sensor, and displacement sensor to establish a response data set; The response data set is subjected to response authentication through the trigger warning threshold, and a response warning is established.

7. A construction management method for an upward bridge-building machine as claimed in claim 1, characterized in that: The method further comprises: Synchronously activating a visual sensing device, and using the visual sensing device to perform scene attention monitoring of the application scene, wherein the scene attention monitoring includes dynamic scene attention monitoring and dynamic scene attention monitoring; A scene anomaly is established according to the scene attention monitoring result, a reconstruction control parameter is generated through the scene anomaly, and the reconstruction control parameter is used to control the upward bridge-building machine to continue to perform the bridge lifting task.

8. A construction management device for an upward bridge-building machine, characterized in that: A construction management method for an upward bridge-building machine according to any one of claims 1 to 7, the device comprising: A twin model building module, used to build a twin model using the upward bridge-building machine and the application scenario after the upward bridge-building machine is initialized for equipment assembly; An importance identification establishment module is used to optimize the bridge hoisting path based on the twin model, configure a calibration path based on the hoisting path optimization result, segment the calibration path, and establish a position importance identification of the path segmentation result; A first control parameter establishment module is used to establish control constraints according to the path segmentation result and the position importance identifier, take the calibrated path as the optimization target, perform control parameter fitting under the control constraints based on the parameter optimization network, and establish first control parameters; A status data set establishment module is used to collect real-time data using a wire rope tension sensor, a gyroscope, an acceleration sensor, and a displacement sensor installed on the upward bridge-building machine to establish a real-time status data set; The second control parameter establishment module is used to monitor the wind speed of the application scenario, establish real-time wind speed data, use the real-time state data set and the real-time wind speed data as input data, perform hoisting stability optimization through a reinforcement learning channel, and establish the second control parameter; A lifting task execution module is used to control the upward bridge-building machine to perform a bridge lifting task after compensating the first control parameter with the second control parameter.

Images