Construction limited conditions under large tonnage precast beam fast assembly control method and system

Abstract

The present application relates to the technical field of precast beam assembly, and particularly relates to a large-tonnage precast beam rapid assembly control method and system under construction limited conditions. The method comprises the following steps: obtaining environmental perception data and beam body measurement data of a construction area; analyzing a dynamic obstacle boundary coefficient vector and a support ideal installation pose matrix according to the environmental perception data; extracting a beam body six-degree-of-freedom deviation vector and a hoisting system dynamic damping coefficient matrix through the beam body measurement data; fusing the dynamic obstacle boundary coefficient vector, the support ideal installation pose matrix, the beam body six-degree-of-freedom deviation vector and the hoisting system dynamic damping coefficient matrix to generate a fusion safety pose deviation vector; and generating an assembly control instruction based on the fusion safety pose deviation vector and the support ideal installation pose matrix to control hoisting equipment to complete assembly of the precast beam.

Description

Technical Field

[0001] This invention relates to the field of precast beam assembly technology, specifically to a method and system for rapid assembly control of large-tonnage precast beams under construction constraints. Background Technology

[0002] In bridge engineering, especially in complex construction scenarios such as highway expansion, the assembly of large-tonnage precast beams is a critical and challenging task. In existing technologies, a semi-automatic method combining laser scanning measurement and manual monitoring is usually used for assembly control.

[0003] Specifically, existing methods acquire static point cloud data of the construction area using fixed laser scanners to determine the support positions. Simultaneously, dynamic obstacles are avoided through operator observation or simple sensor monitoring, and the lifting equipment is controlled to lower the beam based on a pre-set lifting trajectory. However, this method has significant technical drawbacks. First, in complex environments where construction and traffic occur simultaneously, existing static point cloud processing methods struggle to effectively and in real-time perceive and distinguish dynamic obstacles (such as moving construction vehicles or personnel) within the construction area, leading to collision risks and significant safety hazards during lifting. Second, for large-tonnage precast beams, their posture changes during lifting are easily affected by wind loads, rope sway, and other factors. Existing methods lack sufficient accuracy in estimating the beam's pose, typically relying solely on sparse point cloud matching, which is susceptible to drift due to occlusion. This fails to accurately reflect the six-degree-of-freedom pose deviation between the beam and the supports in real-time, directly impacting the final assembly accuracy. Furthermore, existing methods are usually based on simplified dynamic models or fixed damping parameters for hoisting control, which fail to fully consider the dynamic swaying characteristics of the hoisting rope-slinging device-beam coupling system during hoisting. This results in control commands being unable to effectively suppress swaying, making the assembly process unstable and difficult to achieve rapid assembly while ensuring safety and accuracy. Summary of the Invention

[0004] To address the shortcomings of existing methods and the needs of practical applications, this invention provides a method for rapid assembly control of large-tonnage precast beams under construction constraints, comprising the following steps: The process involves acquiring environmental perception data and beam measurement data of the construction area; analyzing the dynamic obstacle boundary coefficient vector and the ideal installation pose matrix of the supports based on the environmental perception data; extracting the six-degree-of-freedom deviation vector of the beam and the dynamic damping coefficient matrix of the hoisting system from the beam measurement data; fusing the dynamic obstacle boundary coefficient vector, the ideal installation pose matrix of the supports, the six-degree-of-freedom deviation vector of the beam, and the dynamic damping coefficient matrix of the hoisting system to generate a fused safe pose deviation vector; and generating assembly control commands based on the fused safe pose deviation vector and the ideal installation pose matrix of the supports to control the hoisting equipment to complete the assembly of the precast beam.

[0005] Optionally, analyzing the dynamic obstacle boundary coefficient vector based on the environmental perception data includes the following steps: Based on the 3D point cloud data of the construction area, the area is constrained according to the preset hoisting operation safety space; according to the displacement of the constrained point cloud per unit time, the segmentation threshold is dynamically determined to segment the constrained point cloud into static facility point cloud and dynamic obstacle point cloud; the dynamic obstacle point cloud is contour fitted to extract the spatial boundary, movement direction and approach speed features of the dynamic obstacle, and generate the dynamic obstacle boundary coefficient vector.

[0006] Optionally, based on the environmental perception data, the ideal installation pose matrix of the support is analyzed, including the following steps: Point clouds of the bearing installation area are extracted from the 3D point cloud data of the construction area; outliers in the point cloud of the bearing installation area are removed using a random sampling consensus algorithm; based on the point cloud after removing outliers, the bearing reference plane is fitted using the least squares method, and the bearing reference plane is corrected according to the bridge alignment design parameters; based on the corrected bearing reference plane, the bearing center coordinates, normal vector, and installation tilt angle are calculated to generate the ideal installation pose matrix of the bearing.

[0007] Optionally, the six-degree-of-freedom deviation vector of the beam is extracted from the beam measurement data, including the following steps: Based on the laser point cloud data of the precast beam and the three-dimensional coordinate data of the preset target on the precast beam, the point cloud is matched using an iterative nearest point algorithm to obtain the initial pose of the beam. The three-dimensional coordinate data of the preset target is used as a rigid constraint to correct the initial pose of the beam to obtain the measured pose of the beam. The linear offset and angular deflection between the measured pose of the beam and the ideal installation pose matrix of the support are calculated to generate the six-degree-of-freedom deviation vector of the beam.

[0008] Optionally, the dynamic damping coefficient matrix of the hoisting system is extracted from the beam measurement data, including the following steps: Using the six-degree-of-freedom deviation vector of the beam as the observed value, and combining the hoisting load, hoisting rope length, and environmental wind load, a state space of the hoisting rope-hoisting equipment-beam coupled system is constructed. An unscented Kalman filter algorithm is used to recursively estimate the sway state of the coupled system. Based on the sway state data obtained from the recursive estimation, the sway modal parameters of the coupled system are identified. These sway modal parameters include the sway amplitude ratio, phase difference, and attenuation characteristics. Based on these sway modal parameters, a dynamic damping coefficient matrix of the hoisting system characterizing the overall sway characteristics of the coupled system is constructed.

[0009] Optionally, the method for rapid assembly control of large-tonnage precast beams under construction constraints further includes: Based on the observation data of a common benchmark target from multiple laser trackers and the six-degree-of-freedom deviation vector of the beam, global optimization is performed using the bundle adjustment method to calculate the system joint error coefficient. The step of extracting the six-degree-of-freedom deviation vector of the beam from the beam measurement data also includes correcting the three-dimensional coordinate data of the preset target using the system joint error coefficient.

[0010] Optionally, the process of fusing the dynamic obstacle boundary coefficient vector, the ideal installation pose matrix of the support, the six-degree-of-freedom deviation vector of the beam, and the dynamic damping coefficient matrix of the hoisting system to generate a fused safe pose deviation vector includes the following steps: A weighted extended Kalman filter is used to fuse and estimate the six-degree-of-freedom deviation vector of the beam, the dynamic damping coefficient matrix of the hoisting system, and the joint error coefficient of the system. A Bayesian network risk inference model is constructed, using the boundary coefficient vector of the dynamic obstacle as input, to infer the risk level of collision between the beam and the dynamic obstacle. Based on the risk level, the fusion weights of the six-degree-of-freedom deviation vector of the beam in the weighted extended Kalman filter are adaptively adjusted to obtain the fused safe pose deviation vector.

[0011] Optionally, before generating assembly control instructions, the following may also be included: Based on the swing mode parameters in the dynamic damping coefficient matrix of the hoisting system, the swing phase of the current beam is matched to identify abnormal swing characteristics; using Mahalanobis distance clustering analysis, the distribution of feature points in the fused safe pose deviation vector and the ideal installation pose matrix of the support is discriminated, and outlier feature points are eliminated; based on the retained effective feature points, the ideal installation pose matrix of the support is refitted and optimized to obtain the purified reference assembly matrix.

[0012] Optionally, the step of generating assembly control commands based on the fused safety pose deviation vector and the ideal installation pose matrix of the support to control the hoisting equipment to complete the assembly of the precast beam includes the following steps: A spatiotemporal graph model is constructed, where nodes represent the beam's pose state at different times, and edges represent the spatial motion relationship and swing constraints between nodes at adjacent times. The dynamic obstacle boundary coefficient vector, system joint error coefficient, fused safe pose deviation vector, and purified reference assembly matrix are input into the spatiotemporal graph network to learn and generate the optimal assembly trajectory. The pose deviation between the beam's six-degree-of-freedom deviation vector and the optimal assembly trajectory is compared in real time, and closed-loop correction is performed based on the hoisting system's dynamic damping coefficient matrix and the dynamic obstacle boundary coefficient vector. Based on the corrected trajectory, control commands are generated to drive the hoisting equipment to walk, lift, and turn.

[0013] Secondly, to efficiently execute the rapid assembly control method for large-tonnage precast beams under construction constraints provided by this invention, this invention also provides a rapid assembly control system for large-tonnage precast beams under construction constraints, comprising: an input device, an output device, a processor, and a memory, wherein the input device, output device, processor, and memory are interconnected, and the memory stores program instructions used for the rapid assembly control method for large-tonnage precast beams under construction constraints. The rapid assembly control system for large-tonnage precast beams under construction constraints provided by this invention has a compact structure and stable performance, and can stably execute the rapid assembly control method for large-tonnage precast beams under construction constraints provided by this invention, further enhancing the overall applicability and practical application capability of this invention.

[0014] This invention, through a dynamic obstacle boundary coefficient vector generated based on a dynamic segmentation threshold, can perceive the contours and movement trends of dynamic obstacles within the construction area in real time and with high accuracy. This fundamentally solves the problem of collision risks caused by dynamic interference, which is difficult to address with traditional methods, and significantly improves the safety of hoisting operations. Secondly, by fusing laser point clouds with preset target coordinates and using bridge alignment constraints for correction, the determined six-degree-of-freedom deviation vector of the beam can accurately reflect the six-degree-of-freedom deviation of the beam relative to the ideal pose of the support. This solves the problem of insufficient accuracy caused by sparse point cloud features and occlusion in traditional pose estimation methods, thus ensuring the final assembly accuracy. Furthermore, by constructing a dynamic damping coefficient matrix for the hoisting system, especially by combining the swing mode parameters identified by unscented Kalman filtering and the time-varying stiffness parameters identified by recursive least squares, the dynamic characteristics of the rope-slinging device-beam coupling system can be comprehensively and in real time characterized. This allows control commands to effectively suppress swing, solving the problem of control instability caused by neglecting the dynamic characteristics of the system in traditional methods, and achieving a stable and efficient assembly process. Finally, by fusing multi-source information such as the dynamic obstacle boundary coefficient vector, the beam six-degree-of-freedom deviation vector, and the dynamic damping coefficient matrix of the hoisting system, and generating a fused safe pose deviation vector based on Bayesian network risk inference and weighted extended Kalman filtering, it is possible to simultaneously consider assembly accuracy and dynamic safety. Combined with the purified baseline assembly matrix and the optimized trajectory generated by the spatiotemporal graph network, closed-loop precise control of the hoisting equipment is finally achieved, comprehensively solving the technical challenge of coordinating safety, accuracy, efficiency, and stability in complex construction environments. Attached Figure Description

[0015] Figure 1 A flowchart illustrating a rapid assembly control method for large-tonnage precast beams under construction constraints, provided by an embodiment of the present invention; Figure 2 This is a framework diagram of a rapid assembly control system for large-tonnage precast beams under construction constraints, provided as an embodiment of the present invention. Detailed Implementation

[0016] Specific embodiments of the present invention will now be described in detail. It should be noted that the embodiments described herein are for illustrative purposes only and are not intended to limit the invention. In the following description, numerous specific details are set forth in order to provide a thorough understanding of the invention. However, it will be apparent to those skilled in the art that these specific details are not necessary to practice the invention. In other instances, well-known circuits, software, or methods have not been specifically described to avoid obscuring the invention.

[0017] Throughout this specification, references to an embodiment, example, or illustration mean that a particular feature, structure, or characteristic described in connection with that embodiment or example is included in at least one embodiment of the invention. Therefore, phrases appearing in various places throughout the specification, such as "in one embodiment," "in an embodiment," "an example," or "an illustration," do not necessarily refer to the same embodiment or example. Furthermore, specific features, structures, or characteristics can be combined in any suitable combination and / or sub-combination in one or more embodiments or examples. Moreover, those skilled in the art will understand that the illustrations provided herein are for illustrative purposes and are not necessarily drawn to scale.

[0018] Please see Figure 1 This invention provides a method for rapid assembly control of large-tonnage precast beams under construction constraints, comprising the following steps: S1. Obtain environmental perception data and beam measurement data of the construction area.

[0019] In one embodiment, three-dimensional point cloud data of the construction area is acquired by lidar, three-dimensional coordinate data of a preset target on the precast beam is acquired by a laser tracker, and tension and acceleration data of the hoisting system are acquired by sensors (such as tension sensors and acceleration sensors).

[0020] The lidar continuously collects three-dimensional spatial information of the construction area at a certain scanning frequency to form a point cloud dataset. The laser tracker continuously tracks multiple high-reflectivity targets embedded on the precast beam. By measuring the distance and angle, the spatial three-dimensional coordinates of each target are calculated. The tension sensor and acceleration sensor of the hoisting system collect the force on the hoisting rope and the acceleration of the beam in real time.

[0021] S2. Based on the environmental perception data, analyze the dynamic obstacle boundary coefficient vector and the ideal installation pose matrix of the support.

[0022] In this embodiment, based on the highway bridge construction design specifications and actual on-site operation requirements, and combined with core parameters such as the limited hoisting height, lateral working width, and longitudinal safety distance, as well as the dimensions of the precast beam itself, the crane's operating radius, and the distribution of existing bridges, roadbeds, and access roads around the construction area, a hoisting safety boundary constraint frame is constructed. The selected range fits the actual construction operation space, preventing invalid point clouds (such as distant passing vehicles, vegetation in non-construction areas, etc.) from entering the subsequent processing flow, significantly reducing the amount of calculation of invalid data, improving the efficiency of point cloud segmentation, and at the same time avoiding interference from irrelevant points to the identification of dynamic obstacles.

[0023] Then, the distribution density and reflection intensity change rate of the neighborhood space of the point cloud within the constraint box are statistically analyzed in real time. Combined with the differences in motion characteristics between dynamic targets (construction machinery, temporary passing vehicles) and static facilities (precast beams, cap beams, fences) in the highway construction scenario, the displacement of the point cloud per unit time is used as the core basis for threshold adjustment. The dynamic segmentation threshold is dynamically adapted to the point cloud change patterns under different construction periods (such as peak traffic period, night construction period) and different environmental conditions to generate a dynamic segmentation threshold that is adapted to the current scenario, so as to distinguish between static construction facilities and moving obstacles, and solve the problem that fixed thresholds cannot adapt to complex construction environments and are prone to misjudgment or omission.

[0024] Next, edge contour fitting is performed on the point cloud of dynamic obstacles identified by dynamic threshold segmentation. A combination of polygon approximation and curve smoothing is used to remove local abnormal protrusions in the contour fitting process. Core features such as spatial boundary coordinates, movement direction, approximation speed, lateral offset, and contour size of the obstacle are extracted. Combined with the hoisting safety boundary constraint requirements, the extracted feature parameters are normalized and integrated into a standardized dynamic obstacle boundary coefficient vector.

[0025] In another embodiment, the bearing installation surface of the highway expansion project is easily affected by the renovation of old bridges, foundation settlement, and construction errors. The present invention uses RANSAC to remove abnormal measuring points in the bearing area, and then uses least squares fitting of the bearing reference plane to obtain the ideal installation pose matrix that meets the stress requirements of the bridge.

[0026] Specifically, precise spatial positioning of the point cloud of the bearing installation area is carried out. In combination with the construction characteristics of the old bridge renovation and new bridge connection of the highway expansion project, the bridge construction coordinate system is used as the benchmark. Combined with the position coordinates and size parameters of the bearing pad stone marked in the cap beam design drawings, as well as the point cloud data of the cap beam edge contour measured on site, the effective measurement range of the cap beam bearing pad stone is locked by combining point cloud spatial coordinate matching and area selection.

[0027] At the same time, based on the actual size of the bearing pad, a reasonable measurement redundancy range is reserved to avoid omission of effective measurement areas due to point cloud acquisition deviation, and to ensure that the point cloud data extracted subsequently all come from the actual installation surface of the bearing pad.

[0028] Then, the RANSAC (Random Sampling Consensus) algorithm was used to robustly process the point cloud of bearing pad stones within the locked range, identify and remove various outliers, including common anomalies of bearing pad stones in highway expansion construction, such as protrusions or depressions caused by damage and weathering of the concrete surface of bearing pad stones in old bridge reconstruction, false point clouds formed by cement slurry, dust and other debris falling during construction, and measurement deviations caused by light path obstruction and reflection interference during laser measurement.

[0029] An initial planar model is constructed by randomly selecting a small number of point clouds. The model fitting accuracy is verified by iteratively. A reasonable number of iterations and error thresholds are set. Valid point clouds that meet the fitting accuracy requirements and truly reflect the planar characteristics of the bearing pad are retained. All outliers that do not conform to the model rules are removed to ensure that the remaining point cloud data can truly represent the actual planar state of the bearing pad.

[0030] Furthermore, based on the effective point cloud purified by the RANSAC algorithm, the least squares method is used to fit the reference plane of the bearing pad. By minimizing the sum of the squares of the distances from all effective points to the fitted plane, a preliminary fitting of the reference plane is achieved, ensuring that the fitted plane can fit the actual surface of the bearing pad to the maximum extent. At the same time, in combination with the structural design requirements of the highway bridge, the design parameters of the bridge's cross slope and longitudinal slope (these parameters are taken from the bridge construction design drawings and verified by on-site measurements) are introduced to correct the angle of the preliminary fitted plane. The tilt angle of the fitted plane is adjusted to be consistent with the overall alignment of the bridge and to meet the bridge's stress standards. This avoids uneven stress after bearing installation due to deviations between the fitted plane and the bridge alignment, which would affect the stability of the bridge structure. This ensures that the fitted reference plane can serve as the ideal plane reference for bearing installation.

[0031] Based on the revised bearing pad reference plane, combined with the bearing design dimensions and installation requirements, the core parameters of the ideal bearing installation posture are calculated: the bearing center coordinates are calculated by fitting the geometric center coordinates of the plane to ensure that the bearing center is precisely aligned with the design position; the vertical direction reference of the bearing installation is determined by fitting the normal vector of the plane to ensure that the verticality of the bearing installation meets the engineering requirements; the bearing installation inclination angle is calculated by fitting the angle between the plane and the cross slope and longitudinal slope of the bridge to ensure that the bearing installation posture is compatible with the bridge alignment.

[0032] The calculated parameters, such as the support center coordinates, normal vector, and installation tilt angle, are standardized and integrated to construct a standardized ideal installation pose matrix for the support. This matrix can comprehensively characterize the ideal installation position and attitude of the support.

[0033] S3. Using the beam measurement data, extract the six-degree-of-freedom deviation vector of the beam and the dynamic damping coefficient matrix of the hoisting system.

[0034] In this embodiment, considering the structural characteristics and hoisting stress requirements of the large-tonnage precast beam, multiple highly reflective laser targets are pre-embedded at designated stress-symmetrical positions on the top surface of the precast beam (preferably selected in areas on both sides of the beam's center of gravity symmetry line, far from the hoisting stress point and unobstructed). The targets are made of wear-resistant, interference-resistant, and highly reflective engineering-grade laser reflective sheets. The fixing method employs pre-embedded anchoring combined with sealing protection to prevent damage from cement slurry, dust, or collisions during construction. This laser target serves as a rigid positioning reference point for pose estimation, providing stable and accurate spatial coordinate references, effectively avoiding the randomness errors caused by sparse point cloud features and local occlusion during pure point cloud matching.

[0035] Then, the ICP algorithm is used for initial matching of the laser point cloud of the beam. The initial matching posture is set based on the geometric contour features of the precast beam (such as the straight line features and chamfer features of the web and flanges) to shorten the iteration convergence time. During the matching process, the transverse slope and longitudinal slope linear constraint parameters of the highway bridge are introduced (these parameters are taken from the bridge construction design drawings and on-site measured data). By constructing linear constraint equations, the geometric rationality of each pair of matching points is checked, and erroneous matching point pairs that do not conform to the bridge engineering alignment features or exceed the reasonable deviation range are eliminated, reducing the deviation of the point cloud matching and improving the accuracy and stability of the initial matching.

[0036] Furthermore, the three-dimensional coordinate data of the pre-embedded laser target is acquired in real time using a laser tracker. After system error correction, the measured coordinates of the laser target are used as rigid constraints and substituted into the beam pose calculation process. To address the translational and rotational deviations that are prone to occur during the ICP algorithm iteration process, a deviation correction model is established based on the target coordinates. This model performs real-time verification and correction on the results of each step of the ICP iteration, suppressing iteration drift and gradually reducing pose calculation deviations. Ultimately, this achieves precise locking of the beam's spatial pose, ensuring that the pose estimation results meet the accuracy requirements of bridge assembly.

[0037] Finally, the measured pose data of the beam after target constraint correction is compared and analyzed dimension by dimension with the reference data of the bearing design and installation pose. Using the bridge construction coordinate system as a unified reference, the linear offset of the beam along the X-axis (longitudinal of the bridge), Y-axis (lateral of the bridge), and Z-axis (vertical) is calculated respectively. At the same time, the angular deflection of the beam around the X-axis, Y-axis, and Z-axis (i.e., pitch angle, roll angle, and yaw angle) is calculated. The deviation parameters of these six dimensions are normalized to form a standardized six-degree-of-freedom deviation vector of the beam.

[0038] In another embodiment, the six-degree-of-freedom deviation vector of the beam is used as the core observation value. This vector includes the linear offset of the beam along the three axes of X, Y, and Z and the angular deflection around the three axes, which can reflect the attitude change of the beam in real time during the hoisting process. Based on this, the coupling state space of the hoisting rope-hoisting tool-beam is constructed.

[0039] Considering that during highway expansion construction, the lifting load of large-tonnage precast beams is large and the stress on the lifting ropes is complex, and that factors such as wind load and changes in the length of the lifting ropes in the construction environment will dynamically affect the characteristics of the coupled system, the lifting load weight (taken from the design parameters of the precast beams and the on-site weighing data), the length of the lifting ropes (collected in real time by the crane extension sensor), and the construction wind load (monitored in real time by the on-site meteorological sensors, including wind speed and wind direction parameters) are taken as the core state influencing factors. These three factors are incorporated into the construction process of the coupled state space to ensure that the state space can truly reflect the dynamic coupling relationship between the lifting ropes, lifting equipment, and beams in the construction scenario.

[0040] The unscented Kalman filter algorithm is applied to recursively estimate the swing state of the coupled system of sling-lifting device-beam in real time. During the recursive estimation process, the initial parameters of the filter are set reasonably, and the process noise covariance and observation noise covariance are dynamically adjusted according to the characteristics of the construction scenario. This ensures that the filter algorithm can quickly track the dynamic changes of the swing state, effectively suppress observation noise and system interference, and improve the accuracy and stability of the swing state estimation.

[0041] Furthermore, based on the precise swing state data output by the unscented Kalman filter, modal parameter identification of different swing patterns was carried out, comprehensively covering common swing types during the hoisting of large-tonnage precast beams, including vertical vibration (caused by the extension and contraction of the hoisting rope and the lifting and lowering of the crane, affecting the vertical positioning accuracy of the beam), lateral swing (caused by wind load and crane movement, which can easily lead to collisions between the beam and supports and obstacles), and torsional swing (caused by uneven force on the lifting equipment and the offset of the beam's center of gravity, affecting the installation posture of the beam).

[0042] For each swing pattern, its core modal parameters are identified, including the amplitude ratio of each swing pattern (reflecting the intensity proportion of different swing types, providing a priority basis for subsequent swing suppression), phase difference (reflecting the synchronization relationship between different swing patterns, avoiding the superposition and amplification of swings), and attenuation characteristics (reflecting the attenuation rate of the swing, providing a reference for the adjustment of dynamic damping parameters), to ensure that the modal parameters can characterize the swing law of the hoisting system.

[0043] The identified swing mode parameters (amplitude ratio, phase difference, attenuation characteristics, etc. of each swing mode) are systematically integrated according to spatial motion relationships. Combined with the spatial reference of the bridge construction coordinate system, a standardized swing mode matrix is ​​constructed. This matrix clearly represents the spatial distribution law of parameter correlation of different swing modes in the form of matrix elements, including the mutual influence relationship between each swing mode, ensuring that the matrix can comprehensively reflect the overall swing characteristics of the rope-lifting device-beam coupling system.

[0044] In the embodiment, during the hoisting, lifting, lateral movement, and lowering of the large-tonnage precast beam, the elongation and stiffness of the wire rope change in real time with the load and attitude. By recursively least squares to identify the time-varying stiffness in real time, a dynamic damping coefficient matrix of the hoisting system is constructed to provide dynamic parameter support for attitude stability control.

[0045] Specifically, using real-time data on the tension of the hoisting rope and the acceleration data of the beam collected during the hoisting process as core input parameters, and combining the scenario characteristics of hoisting large-tonnage precast beams in highway expansion projects, a coupled dynamic characteristic model of wire rope-beam is constructed.

[0046] Among them, the lifting rope tension data is collected in real time by the tension sensor on the crane itself, and the sampling frequency is adapted to the dynamic working conditions of the lifting, while abnormal tension values ​​caused by sensor vibration and construction interference are eliminated simultaneously; the beam acceleration data is collected by acceleration sensors embedded in key stress parts of the beam, covering the acceleration changes of the beam along the X, Y, and Z axes, and comprehensively reflecting the attitude fluctuations and stress state of the beam during the lifting process.

[0047] During the model construction process, the elastic deformation characteristics of the lifting rope, the stiffness of the beam structure, and the force transmission law of the lifting equipment were incorporated. Dynamic factors such as load changes, rope length adjustments, and wind load disturbances during the lifting process were also fully considered to ensure that the model can accurately represent the force and motion coupling relationship between the wire rope and the beam, providing a dynamic basis that fits the actual engineering for the subsequent identification of time-varying stiffness parameters.

[0048] Next, a recursive least squares algorithm is used to identify and update the time-varying stiffness parameters of the wire rope-beam coupling system in real time, adapting to the stiffness drift problem caused by attitude changes during the hoisting of large-tonnage precast beams.

[0049] During the initialization phase, the initial stiffness value is set by combining the design stiffness parameters of the precast beam and the material properties of the wire rope (elastic modulus, cross-sectional area). Subsequently, for each set of data on the tension of the suspension rope and the acceleration of the beam, a recursive calculation is performed to gradually correct the stiffness parameters and achieve online adaptive updating of the parameters.

[0050] Meanwhile, the forgetting factor is introduced to optimize the recursive process, and historical data is given reasonable weights. This retains the dominant influence of recent data on stiffness parameters while taking into account the reference value of early data. It effectively suppresses the impact of construction interference and sensor noise on parameter identification, ensuring that stiffness parameters can track changes in hoisting posture in real time (such as beam tilting, rope extension and contraction, load center of gravity shift, etc.), thus solving the shortcomings of traditional fixed stiffness parameters that cannot adapt to dynamic hoisting conditions.

[0051] Then, using the core information represented by the oscillation mode matrix, such as the amplitude ratio, phase difference, and attenuation characteristics of vertical vibration, lateral oscillation, and torsional oscillation, the optimal damping ratio of the hoisting system under different beam postures is calculated. Considering the strength differences and impact priorities of different oscillation types, damping control weights are assigned to lateral oscillation (which easily leads to collision risks), torsional oscillation (which affects installation accuracy), and vertical vibration (which affects beam drop stability). Combined with real-time stiffness parameters obtained through recursive least squares identification, a damping ratio optimization model is constructed. Through model calculation, the optimal damping parameters are determined for different beam postures (such as pitch, roll, and yaw) and different oscillation amplitudes, ensuring that the damping system can accurately suppress various oscillations while avoiding excessive damping that reduces hoisting efficiency. This achieves a balance between stability and efficiency, providing dynamic damping support for the attitude stability control of the hoisting system.

[0052] Finally, the calculated multi-dimensional damping parameters are systematically integrated to form a standardized dynamic damping coefficient matrix for the hoisting system. During the integration process, using the bridge construction coordinate system as a reference, the matrix includes parameters such as the optimal damping parameters under different postures, the damping weights corresponding to each swing type, and the damping attenuation coefficient. This matrix can comprehensively reflect the damping characteristics of the hoisting system under different dynamic working conditions, achieving real-time matching of damping parameters with the beam posture and swing state.

[0053] S4. The dynamic obstacle boundary coefficient vector, the ideal installation posture matrix of the support, the six-degree-of-freedom deviation vector of the beam, and the dynamic damping coefficient matrix of the hoisting system are fused to generate a fused safe posture deviation vector.

[0054] Multiple laser trackers are prone to installation tilt angle errors, optical path refraction errors, and collaborative positioning errors in the narrow construction space of highways. This invention uses the beam adjustment method to perform global optimization constraints on the measurement beams of the entire station, realize joint self-calibration between devices, and obtain the system comprehensive error coefficient for pose compensation.

[0055] Specifically, considering the limited space and the constraints of multi-equipment collaborative operation during highway expansion construction, cross-station public laser targets are deployed within the construction area to serve as a reference for unified observation by multiple laser trackers. High-stability, interference-resistant engineering-grade laser reflectors are selected for the targets. Their deployment locations must meet the requirements of unobstructed view, coverage of the entire construction area, and ease of simultaneous observation by multiple devices. Priority is given to unobstructed areas at higher elevations such as the top of the cap beam and the top of the construction enclosure. Furthermore, the target deployment points must be precisely correlated with the bridge construction coordinate system. The three-dimensional coordinates of the targets are determined in advance through static measurements and calibration records are maintained. Simultaneously, the target surfaces are sealed to prevent construction dust and cement slurry from affecting laser reflection. The number of targets is determined reasonably based on the length of the construction area and the number of laser trackers deployed, generally no less than three, forming a triangular reference network to ensure that clear and stable target signals are obtained by multiple devices during observation.

[0056] Based on the observation principle of multi-station laser trackers and the measurement requirements of highway construction, a beam adjustment optimization model is constructed. This model integrates the observation beams of all laser trackers, the calibration coordinates of cross-station common reference targets, and the measured pose data of the beam into a single global constraint equation. During model construction, considering the narrow construction space of highways, key conditions such as equipment installation attitude constraints, optical path propagation constraints, and observation accuracy constraints are introduced to determine the correlation between various observation parameters. The installation tilt angle, internal optical parameters, and optical path refraction errors of the laser trackers are all considered as variables to be optimized. Simultaneously, the measured pose of the beam is used as a verification constraint to ensure that the model can comprehensively reflect the actual situation of multi-device collaborative observation, achieving global unified optimization of multi-source observation data and avoiding the impact of single-device observation deviations on overall calibration accuracy.

[0057] Furthermore, with minimizing the observation residuals of each laser tracker as the core optimization objective, an iterative optimization strategy was formulated to gradually correct the installation attitude, internal parameters, and optical path deviations of each laser tracker. During the iteration process, the parameters of each device were first initialized, and the initial observation residuals were calculated by substituting them into the beam adjustment optimization model. It was then determined whether the residuals met the accuracy requirements for construction surveying (referencing the millimeter-level accuracy standards for highway bridge assembly). If the residuals did not meet the requirements, the installation tilt angle, internal focal length, and optical path compensation parameters of the laser trackers were adjusted successively, and the observation residuals were recalculated. This process was repeated iteratively until the residuals reached their minimum and stabilized.

[0058] Meanwhile, by taking into account the dynamic characteristics of the construction scenario, a reasonable upper limit for the number of iterations and a residual convergence threshold are set to avoid parameter distortion caused by excessive iteration, ensuring that the optimized equipment parameters can adapt to the observation needs of narrow construction spaces, effectively suppressing accuracy drift in the multi-device collaborative positioning process, and improving the stability of the overall measurement system.

[0059] After iterative optimization, based on the optimized equipment parameters and observation data, the comprehensive error of the multi-station laser tracker system is calculated, including: system comprehensive positioning deviation (reflecting the overall position deviation when multiple devices are working together to position, which directly affects the accuracy of beam and support posture measurement), angle deviation (reflecting the deviation of the laser tracker's observation angle, which is related to the equipment installation tilt angle and optical path refraction), and distance deviation (reflecting the system error in the laser ranging process, which is related to the equipment's internal parameters and ambient light interference).

[0060] The various deviation parameters obtained from the calculation are standardized, abnormal deviation values ​​are eliminated, and all deviation parameters are integrated into a standardized system joint error coefficient in combination with the accuracy requirements of the construction scenario. The weight ratio of each error component is set to ensure that the measurement data in the entire assembly control process is accurate and reliable.

[0061] Furthermore, using weighted extended Kalman filtering as the core fusion framework, the three core data types—the six-degree-of-freedom deviation vector of the beam, the dynamic damping coefficient matrix of the hoisting system, and the joint error coefficient of the multi-station laser tracker system—are fused and estimated to ensure that the fusion result takes into account both pose accuracy and meets the requirements of hoisting dynamic stability.

[0062] Considering the constrained characteristics and safety management standards of highway expansion construction, the construction safety level is divided into three levels: high, medium, and low (high level corresponds to the critical stage of beam hoisting and lowering, and scenarios involving close-range interference from obstacles; low level corresponds to non-critical stages such as lateral movement and lifting). The fusion weights of each data point are adaptively allocated according to different safety levels: the weight of the system joint error coefficient increases with the safety level to ensure high-precision error compensation; the weight of the dynamic damping coefficient matrix is ​​positively correlated with the swing amplitude, with higher weights for more severe swings, prioritizing attitude stability; the beam's six-degree-of-freedom deviation vector, as the core observation data, maintains its basic weight to ensure accurate representation of pose deviations. Through weight allocation, the advantages of multi-source data are complemented, avoiding the impact of single data deviations on the fusion results.

[0063] Secondly, a Bayesian network is introduced to construct a construction safety risk inference model, with the dynamic obstacle boundary coefficient vector as the core inference condition. The features contained in this vector, such as the obstacle spatial boundary, movement direction, approach speed, and lateral offset, are all used as input nodes of the Bayesian network.

[0064] Considering the characteristics of highway construction and traffic flow, the prior probabilities corresponding to different obstacle types (construction machinery, temporary vehicles), different approach distances, and different movement speeds are preset. Through the probabilistic reasoning process of Bayesian network, the posterior probability of the beam colliding with dynamic obstacles in different pose states is characterized one by one. Based on the collision probability, three safety risk levels are divided: high risk corresponds to the scenario where the obstacle approaches quickly and is close to the hoisting area; low risk corresponds to the scenario where the obstacle is far from the hoisting area and moves slowly.

[0065] Then, based on the safety risk level derived from Bayesian network inference, the fusion weights of the six-degree-of-freedom deviation vector of the beam are adaptively adjusted to achieve a dynamic balance between accuracy and safety.

[0066] For high-risk scenarios, the focus is on reducing the allowable pose deviation in dangerous directions (such as the direction of obstacle approach) while increasing the deviation weight in that direction to ensure that the system can prioritize attitude adjustments in dangerous directions and minimize collision risks. For medium-risk scenarios, the deviation weight is adjusted appropriately to balance attitude accuracy and safety control. For low-risk scenarios, the focus is on ensuring attitude accuracy while appropriately reducing safety constraint weights to improve lifting efficiency.

[0067] At the same time, combined with the swing state reflected by the dynamic damping coefficient matrix, if the swing amplitude of the beam exceeds the safety threshold, the weight of the damping characteristic related data is increased simultaneously to help suppress the swing and further reduce the risk of collision, ensuring that the weight correction is both in line with the real-time risk state and adapted to the dynamic working conditions of hoisting.

[0068] After completing multi-source data fusion, risk reasoning, and weight correction, a fused safe pose deviation vector with fused safety constraints is obtained. This vector is based on the six-degree-of-freedom deviation vector of the beam and integrates the core requirements of system error compensation, dynamic damping constraints, and safety risk management. The deviation parameters of each dimension are verified and corrected for safety, ensuring that the deviation data can accurately reflect the difference between the actual and ideal attitude of the beam, and limiting the deviation range in dangerous directions through safety constraints to avoid collision accidents caused by attitude deviation.

[0069] S5. Based on the fused safety pose deviation vector and the ideal installation pose matrix of the support, generate assembly control commands to control the hoisting equipment to complete the assembly of the precast beam.

[0070] In the embodiments, the hoisting process is prone to sudden impacts and wind disturbances that can lead to abnormal posture data. The present invention uses swing phase matching to ensure the consistency of swing patterns, and uses Mahalanobis distance clustering to eliminate abnormal assembly features, purify the benchmark assembly matrix, and avoid erroneous instructions from affecting the beam dropping accuracy.

[0071] Specifically, using the oscillation mode matrix as the core reference, oscillation phase matching is performed by utilizing core parameters such as the amplitude ratio, phase difference, and attenuation characteristics of vertical vibration, lateral oscillation, and torsional oscillation contained in the matrix. Combined with the dynamic lifting conditions during highway expansion construction (such as instantaneous wind disturbance, crane start-up and shutdown impact, and elastic deformation of the lifting rope), the reference phase pattern of the beam oscillation is extracted in real time. The currently collected real-time oscillation phase of the beam is compared frame by frame with the reference phase to identify instantaneous abnormal features inconsistent with the overall oscillation pattern, including sudden changes in oscillation amplitude, phase shift, and disordered oscillation types. These anomalies are mostly caused by sudden factors such as instantaneous strong winds, crane operation disturbances, and obstacle collision warnings. Phase matching can accurately pinpoint the occurrence time and specific characteristics of such anomalies.

[0072] Then, using Mahalanobis distance clustering analysis, combined with relevant feature points in the fusion safety pose deviation vector and the ideal installation pose matrix of the support, a set of assembly pose feature points is constructed to determine the spatial distribution pattern and normal fluctuation range of the feature points.

[0073] The Mahalanobis distance between each feature point and the cluster center of the feature point set is calculated. This distance accurately reflects the deviation of a single feature point from the overall feature distribution and is unaffected by differences in the dimensions of each dimension. A reasonable distance threshold is set based on construction accuracy requirements (millimeter-level assembly standards). Feature points exceeding the threshold or far from the cluster center are identified as outliers. False feature points and abnormal posture points caused by instantaneous wind disturbance, sensor measurement deviations, or instantaneous swaying of the suspension ropes are eliminated. This ensures that the retained feature points accurately reflect the actual position and posture of the beam assembly, providing reliable data support for subsequent optimization of the baseline assembly matrix.

[0074] Next, valid assembly pose feature points retained after screening by Mahalanobis distance clustering analysis were collected. These points, combined with the ideal installation pose matrix of the bearings and the swing mode matrix, were then refitted and optimized to establish the assembly datum. During the fitting process, key parameters such as bridge construction alignment constraints and bearing installation accuracy requirements were introduced to calibrate the spatial coordinates of the valid feature points. This corrected for potential datum offset issues that might arise after outlier removal. Simultaneously, dynamic interference factors in the construction scenario (such as temperature changes and foundation settlement) were considered to dynamically fine-tune the assembly datum, ensuring that the fitted assembly datum both met design standards and adapted to actual on-site construction conditions.

[0075] By refitting and optimizing, the interference of abnormal data in the early stage on the assembly benchmark is eliminated, so that the assembly benchmark can accurately match the ideal installation posture of the support, providing a stable and reliable benchmark reference for the subsequent beam assembly, and ensuring that the assembly accuracy meets the construction specifications of highway bridges.

[0076] Based on the refitted and optimized assembly datum, and combined with the purified pose feature points, the ideal installation pose matrix of the supports, and the swing mode constraint parameters, a standardized purified datum assembly matrix is ​​constructed. During the matrix construction process, the bridge construction coordinate system is used as a unified datum, and the physical meaning of each element in the matrix is ​​defined, covering core information such as the ideal assembly pose of the beam, the support installation datum, and the swing suppression constraint.

[0077] Furthermore, the construction space constraints, historical motion trajectories, real-time deviations, system errors, and oscillation characteristics are constructed into a spatiotemporal correlation graph structure. By learning the assembly motion law through the spatiotemporal graph network, continuous and smooth hoisting control commands are output to achieve high-precision, fast, and stable precast beam assembly.

[0078] Specifically, considering the core characteristics of highway expansion construction sites—narrow spaces, limited hoisting space, and multiple constraint couplings—a spatiotemporal graph model adapted for the assembly of large-tonnage precast beams is constructed. In the model, nodes are defined as key beam poses at different times (covering the beam's six degrees of freedom parameters and swing state parameters). Each node encapsulates core information such as beam pose data at the corresponding time, dynamic damping parameters of the hoisting system, and system joint error coefficients, representing the spatial position and motion state of the beam at a given moment. Edges represent the spatial motion relationships (such as beam lateral displacement distance, lifting height, and turning angle) and swing constraints between nodes at adjacent times. The swing patterns and phase constraints in the swing mode matrix are integrated into the edge definition, including motion constraint boundaries between different poses and swing suppression requirements. Simultaneously, dynamic obstacle boundary coefficient vector-related constraints are embedded to ensure that the spatiotemporal graph model can comprehensively reflect the assembly constraints under the limited construction environment, providing a foundational framework that fits the actual working conditions for subsequent trajectory learning.

[0079] All preceding feature parameters are input into the spatiotemporal graph network to achieve collaborative empowerment of multi-source data, ensuring the comprehensiveness and relevance of network learning. The input data includes dynamic obstacle boundary coefficient vector (for avoiding collision risks), beam six-degree-of-freedom deviation vector (for accurately locating attitude deviations), swing mode matrix (for constraining swing patterns), support ideal installation pose matrix (for locking assembly datum), system joint error coefficient (for error compensation), hoisting system dynamic damping coefficient matrix (for ensuring attitude stability), fused safe pose deviation vector (for balancing accuracy and safety), and purified datum assembly matrix (for locking ideal assembly pose).

[0080] By conducting deep learning on these multi-source data, the spatiotemporal graph network can uncover the spatiotemporal correlations and constraint logic between different parameters, autonomously learn the optimal motion trajectory rules for precast beam assembly under the limited space of highway expansion, adapt to the core requirements of swing suppression, obstacle avoidance, and precision control during hoisting, automatically avoid unreasonable motion trajectories (such as trajectories that exceed the hoisting space, are prone to swing superposition, or have collision risks), and generate an initial optimal trajectory that fits the on-site working conditions.

[0081] Using the fused safe pose deviation vector as the core reference, the system compares the current actual pose of the beam with the optimal trajectory pose planned by the spatiotemporal graph network in real time, accurately calculates the deviation between the two, and determines whether the deviation exceeds the millimeter-level precision threshold for highway bridge assembly. If the deviation is within the allowable range, the current trajectory remains unchanged; if the deviation exceeds the threshold, a closed-loop correction process is immediately initiated, and the trajectory parameters are adjusted specifically by combining the swing mode matrix and the dynamic damping coefficient matrix. To address deviations caused by swaying, damping parameters are adjusted synchronously to suppress swaying and reduce the amplification of deviations. To address deviations caused by obstacle avoidance, the trajectory direction and velocity are fine-tuned by combining the dynamic obstacle boundary coefficient vector. To address deviations caused by system errors, real-time compensation is performed using the system joint error coefficient to ensure that trajectory correction can respond quickly and adapt accurately, achieving closed-loop control of planning-comparison-correction-execution. This simultaneously considers sway suppression and dynamic obstacle avoidance, ensuring that the beam's posture always conforms to the optimal trajectory and preventing the accumulation of deviations from affecting assembly accuracy.

[0082] Based on the optimal trajectory output by the spatiotemporal graph network and the pose parameters after closed-loop correction, standardized control commands that can directly drive the crane actuator are generated. The command output is strictly adapted to the operating characteristics of large-tonnage cranes and the high-efficiency requirements of highway expansion construction, covering the three core operating dimensions of crane travel, lifting, and steering.

[0083] The travel control commands include the crane's travel speed (adaptively adjusted based on obstacle distance and trajectory complexity, with low and stable speed during close-range beam lowering and moderate speed increase during long-range lateral movement) and travel direction, precisely matching the path requirements of the trajectory plan; the lifting control commands combine the distance between the beam and the support, and the swing state, including lifting speed and stroke, to avoid aggravated beam swaying caused by high-speed lifting, and adopt a low-speed slow descent mode during beam lowering to ensure precise docking between the beam and the support; the steering control commands precisely adjust the crane's steering angle to adapt to the operational limitations of narrow construction spaces and avoid collisions between the crane and existing bridges, construction machinery, and dynamic obstacles.

[0084] All control commands undergo system error compensation and safety verification to ensure accuracy and reliability, driving the crane to accurately lower and assemble the beams, ultimately achieving rapid, safe, and high-precision assembly of large-tonnage precast beams, meeting the construction schedule and quality requirements of the highway expansion project.

[0085] It should be noted that the specific implementation methods described above, such as image processing, numerical simulation, and the construction and training of machine learning models, can all be accomplished by the processor by calling the corresponding computer program instructions stored in memory. Those skilled in the art can implement the above functions using algorithms and tools known in the prior art, according to actual needs.

[0086] Please see Figure 2 In an embodiment, to efficiently execute the rapid assembly control method for large-tonnage precast beams under construction constraints provided by the present invention, the present invention also provides a rapid assembly control system for large-tonnage precast beams under construction constraints, comprising: an input device 1, an output device 2, a processor 3, and a memory 4. The input device 1, output device 2, processor 3, and memory 4 are interconnected. The memory 4 stores program instructions used to execute the steps of the rapid assembly control method for large-tonnage precast beams under construction constraints. The rapid assembly control system for large-tonnage precast beams under construction constraints of the present invention has a compact structure and stable performance, and can stably execute the rapid assembly control method for large-tonnage precast beams under construction constraints provided by the present invention, further improving the overall applicability and practical application capability of the present invention.

[0087] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered within the scope of the present invention.

Claims

1. A method for rapid assembly control of large-tonnage precast beams under construction constraints, characterized in that, Includes the following steps: Acquire environmental perception data and beam measurement data of the construction area; Based on the environmental perception data, analyze the dynamic obstacle boundary coefficient vector and the ideal installation pose matrix of the support; Based on the beam measurement data, the six-degree-of-freedom deviation vector of the beam and the dynamic damping coefficient matrix of the hoisting system are extracted; By fusing the dynamic obstacle boundary coefficient vector, the ideal installation pose matrix of the support, the six-degree-of-freedom deviation vector of the beam, and the dynamic damping coefficient matrix of the hoisting system, a fused safe pose deviation vector is generated. Based on the fused safe pose deviation vector and the ideal installation pose matrix of the support, assembly control commands are generated to control the hoisting equipment to complete the assembly of the precast beam.

2. The method for rapid assembly control of large-tonnage precast beams under construction constraints as described in claim 1, characterized in that, Based on the environmental perception data, the dynamic obstacle boundary coefficient vector is analyzed, including the following steps: Based on the three-dimensional point cloud data of the construction area, the area is constrained according to the preset hoisting operation safety space; Based on the displacement of the constrained point cloud per unit time, a segmentation threshold is dynamically determined to segment the constrained point cloud into a static facility point cloud and a dynamic obstacle point cloud. Contour fitting is performed on the point cloud of the dynamic obstacle to extract the spatial boundary, movement direction and approximation velocity features of the dynamic obstacle, and generate the boundary coefficient vector of the dynamic obstacle.

3. The rapid assembly control method for large-tonnage precast beams under construction constraints as described in claim 1, characterized in that, Based on the environmental perception data, the ideal installation pose matrix of the support is analyzed, including the following steps: Extract the point cloud of the support installation area from the 3D point cloud data of the construction area; Outliers in the point cloud of the support installation area are removed using a random sampling consensus algorithm. Based on the point cloud after removing outliers, the least squares method is used to fit the bearing reference plane, and the bearing reference plane is corrected according to the bridge alignment design parameters. Based on the corrected support reference plane, the center coordinates, normal vector, and installation tilt angle of the support are calculated to generate the ideal installation pose matrix of the support.

4. The method for rapid assembly control of large-tonnage precast beams under construction constraints as described in claim 1, characterized in that, The six-degree-of-freedom deviation vector of the beam is extracted using the beam measurement data, including the following steps: Based on the laser point cloud data of the precast beam and the three-dimensional coordinate data of the preset target on the precast beam, the point cloud is matched by the iterative nearest point algorithm to obtain the initial pose of the beam. Using the three-dimensional coordinate data of the preset target as a rigid constraint, the initial pose of the beam is corrected to obtain the measured pose of the beam. Calculate the linear offset and angular deflection between the measured pose of the beam and the ideal installation pose matrix of the support, and generate the six-degree-of-freedom deviation vector of the beam.

5. The rapid assembly control method for large-tonnage precast beams under construction constraints as described in claim 1, characterized in that, The dynamic damping coefficient matrix of the hoisting system is extracted using the beam measurement data, including the following steps: Using the six-degree-of-freedom deviation vector of the beam as the observed value, and combining the hoisting load, hoisting rope length and environmental wind load, the state space of the coupled system of hoisting rope-hoisting equipment-beam is constructed. The oscillation state of the coupled system is recursively estimated using an unscented Kalman filter algorithm. Based on the oscillation state data obtained by recursive estimation, the oscillation mode parameters of the coupled system are identified. The oscillation mode parameters include the oscillation amplitude ratio, phase difference, and attenuation characteristics. Based on the swing mode parameters, a dynamic damping coefficient matrix of the hoisting system is constructed to characterize the overall swing characteristics of the coupled system.

6. The method for rapid assembly control of large-tonnage precast beams under construction constraints as described in claim 1, characterized in that, Also includes: Based on the observation data of the common benchmark target from multiple laser trackers and the six-degree-of-freedom deviation vector of the beam, the bundle adjustment method is used for global optimization to calculate the joint error coefficient of the system. The step of extracting the six-degree-of-freedom deviation vector of the beam from the beam measurement data also includes correcting the three-dimensional coordinate data of the preset target using the system joint error coefficient.

7. The method for rapid assembly control of large-tonnage precast beams under construction constraints as described in claim 1, characterized in that, The process of fusing the dynamic obstacle boundary coefficient vector, the ideal installation pose matrix of the support, the six-degree-of-freedom deviation vector of the beam, and the dynamic damping coefficient matrix of the hoisting system to generate a fused safe pose deviation vector includes the following steps: A weighted extended Kalman filter is used to fuse and estimate the six-degree-of-freedom deviation vector of the beam, the dynamic damping coefficient matrix of the hoisting system, and the joint error coefficient of the system. A Bayesian network risk reasoning model is constructed, and the risk level of a collision between the beam and the dynamic obstacle is inferred by using the boundary coefficient vector of the dynamic obstacle as input. Based on the risk level, the fusion weights of the six-degree-of-freedom deviation vector of the beam are adaptively adjusted in the weighted extended Kalman filter to obtain the fused safe pose deviation vector.

8. The method for rapid assembly control of large-tonnage precast beams under construction constraints as described in claim 1, characterized in that, Before generating assembly control instructions, the following is also included: Based on the swing mode parameters in the dynamic damping coefficient matrix of the hoisting system, the current swing phase of the beam is matched to identify abnormal swing characteristics; The distribution of feature points in the fused safe pose deviation vector and the ideal installation pose matrix of the support is determined by Mahalanobis distance clustering analysis, and outlier feature points are removed. Based on the retained valid feature points, the ideal installation pose matrix of the support is refitted and optimized to obtain the purified reference assembly matrix.

9. The method for rapid assembly control of large-tonnage precast beams under construction constraints as described in claim 1, characterized in that, The process of generating assembly control commands based on the fused safety pose deviation vector and the ideal installation pose matrix of the support to control the hoisting equipment to complete the assembly of the precast beam includes the following steps: Construct a spatiotemporal graph model, where nodes represent the beam's pose state at different times, and edges represent the spatial motion relationship and swing constraints between nodes at adjacent times; The dynamic obstacle boundary coefficient vector, the system joint error coefficient, the fused safe pose deviation vector, and the purified reference assembly matrix are input into the spatiotemporal graph network to learn and generate the optimal assembly trajectory. The pose deviation of the six-degree-of-freedom deviation vector of the beam is compared with that of the optimal assembly trajectory in real time, and closed-loop correction is performed based on the dynamic damping coefficient matrix of the hoisting system and the dynamic obstacle boundary coefficient vector. Based on the corrected trajectory, control commands are generated to drive the hoisting equipment to move, lift, and turn.

10. A rapid assembly control system for large-tonnage precast beams under construction constraints, characterized in that, The rapid assembly control system for large-tonnage precast beams under construction constraints includes: an input device, an output device, a processor, and a memory. The input device, output device, processor, and memory are interconnected. The memory stores program instructions, which are used to execute the rapid assembly control method for large-tonnage precast beams under construction constraints as described in any one of claims 1-9.

Images