High-precision automatic adjustment and calibration system for prefabricated part production and assembly

Abstract

The invention relates to a high-precision automatic adjustment and calibration system for prefabricated part production and assembly. The system comprises a multi-sensor acquisition layer used for acquiring whole-process state data of a prefabricated part; the digital twin modeling layer is used for obtaining a digital component model according to the component data; the digital component model comprises a temperature field, a stress field, a strain tensor field and a curing state field; the deformation prediction layer is used for performing deformation analysis on component concrete shrinkage characteristics and steel bar stress distribution based on the digital component model to obtain deformation prediction parameters; the precise control execution compensation layer is used for obtaining correction execution parameters according to the deformation prediction parameters; the correction execution parameter is used for indicating the machine tool to adjust the prefabricated part according to the correction execution parameter. By means of the system, thermal deformation and shrinkage deformation in the forming process can be predicted, structural compensation and mold adjusting strategies are adopted in advance, later error accumulation is remarkably reduced, the forming precision of the prefabricated part is improved, and cost is reduced.

Description

Technical Field

[0001] The invention belongs to the field of adjustment and calibration, and in particular relates to a high-precision automatic adjustment and calibration system for production and assembly of prefabricated components. Background Art

[0002] With the development of building industrialization technology, mass production of prefabricated components for assembly construction has become the mainstream construction method. To ensure high compatibility between components and structural integrity, the production and installation accuracy of prefabricated components have become key control factors.

[0003] Traditional dimensional verification relies primarily on manual spot checks and single-point measurement, which is not only inefficient and limited in accuracy, but also makes it difficult to effectively predict and proactively intervene in deformation during the concrete curing process. This can easily lead to problems such as accumulated installation errors and poor component assembly. The maturity of multi-source measurement technologies such as laser scanning, machine vision, and contact sensors, combined with digital modeling and simulation calculations, has opened up new possibilities for precision control throughout the lifecycle of precast components.

[0004] However, most existing systems are still at the data collection or static analysis stage, and lack a predictive identification mechanism for component deformation trends, as well as the ability to fine-tune the linkage with the construction and assembly process. Summary of the Invention

[0005] Based on this, it is necessary to provide a high-precision automatic adjustment and calibration system for the production and assembly of prefabricated components that can predict and adjust the deformation of components to address the above technical problems.

[0006] In a first aspect, the present application provides a high-precision automatic adjustment and calibration system for the production and assembly of prefabricated components, comprising:

[0007] The multi-sensor acquisition layer is used to obtain the status data of the entire process of prefabricated components; the status data of the entire process includes component data and concrete material characteristic data;

[0008] The digital twin modeling layer is used to obtain a digital component model based on component data; the digital component model includes temperature field, stress field, strain tensor field and solidification state field;

[0009] The deformation prediction layer is used to perform deformation analysis on the concrete shrinkage characteristics and steel stress distribution of the component based on the digital component model to obtain deformation prediction parameters;

[0010] The precision control execution compensation layer is used to obtain correction execution parameters based on the deformation prediction parameters; the correction execution parameters are used to instruct the machine tool to adjust the prefabricated components according to the correction execution parameters.

[0011] In one embodiment, the digital twin modeling layer includes a data processing module, a physical mesh reconstruction module, an attribute fusion mapping layer, and a component status module;

[0012] The data processing module is used to pre-process the component data to obtain a standard data set; the standard data set includes point cloud data, edge feature data, curing temperature spatiotemporal data and strain distribution data;

[0013] The solid mesh reconstruction module is used to reconstruct the mesh based on the point cloud data to obtain a three-dimensional geometric mesh model of the component; the three-dimensional geometric network model of the component includes mesh nodes;

[0014] The attribute fusion mapping module is used to map the curing temperature spatiotemporal data and strain distribution data to the grid nodes through interpolation to obtain the attribute tensor field;

[0015] The component status module is used to perform component boundary and structural topology based on edge feature data, and to obtain a digital component model by combining the component's three-dimensional geometric mesh model and attribute tensor field.

[0016] In one embodiment, the deformation prediction layer includes a physical model simulation module, a data-driven prediction module, and a composite fusion output module;

[0017] The physical model simulation module is used to perform finite element simulation of the heat-stress-deformation relationship based on the digital component model using concrete material property data to obtain predicted deformation and displacement; concrete material property data includes material thermal conductivity parameters and material elastic modulus.

[0018] The data-driven prediction module is used to extract the characteristic tensor structure from the digital component model and obtain the data-driven predicted deformation field based on time series analysis;

[0019] The composite fusion output module is used to fuse the predicted deformation displacement and the data-driven predicted deformation field to obtain the deformation prediction parameters.

[0020] In one embodiment, a finite element simulation of the thermal-stress-deformation relationship is performed based on the digital component model using concrete material property data to obtain predicted deformation displacement, including:

[0021] Based on the temperature field of the digital component model and the thermal conductivity parameters of the material, a heat diffusion simulation is performed to obtain the predicted temperature field at the future moment;

[0022] Based on the stress field and strain tensor field of the digital component model, combined with the elastic modulus of the material body and the predicted temperature field at future moments, the time-varying elastic modulus is obtained;

[0023] The stress tensor is obtained by combining the thermal-stress field based on the predicted temperature field and the time-varying elastic modulus at future moments;

[0024] The deformation variable is predicted based on the stress tensor to obtain the predicted deformation displacement.

[0025] In one embodiment, obtaining the correction execution parameter according to the deformation prediction parameter includes:

[0026] Generate the reverse force compensation parameter corresponding to the mold pressurization according to the regional shrinkage of the component in the deformation prediction parameter;

[0027] Generate temperature adjustment parameters corresponding to the mold heating plate area according to the asymmetric deformation of the component in the deformation prediction parameters;

[0028] Generate corresponding fine-tuning actuator correction parameters according to component displacement in deformation prediction parameters;

[0029] The reverse compensation parameter, the temperature adjustment parameter and the fine-tuning actuator correction parameter are determined as the correction execution parameters.

[0030] In one embodiment, the data processing module is used to pre-process the component data to obtain a standard data set, including:

[0031] Based on the calibration model, the three-dimensional point cloud data and contact measurement data in the component data are spatially aligned, and the external parameter matrix is used to perform unified coordinate transformation to obtain point cloud data;

[0032] Extracting geometric features from image data in component data to obtain edge feature data;

[0033] Remove noise from the curing temperature distribution and internal stress in the component data to obtain curing temperature spatiotemporal data and strain distribution data;

[0034] The point cloud data, edge feature data, curing temperature spatiotemporal data and strain distribution data are encapsulated as a standard data set.

[0035] In one embodiment, the system further comprises an assembly calibration layer;

[0036] The assembly calibration layer is used to compare the real-time measured installation position data with the preset component positioning accuracy in real time and generate calibration instructions; the calibration instructions are used to instruct the actuator to fine-tune the component installation position; the calibration instructions include assembly position parameters and assembly angle parameters.

[0037] In a second aspect, the present application also provides a high-precision automatic adjustment and calibration method for the production and assembly of prefabricated components, comprising:

[0038] Obtaining the whole process status data of prefabricated components; the whole process status data includes component data;

[0039] Obtaining a digital component model based on component data;

[0040] Based on the digital component model, deformation analysis is performed on the component concrete shrinkage characteristics and steel bar stress distribution to obtain deformation prediction parameters;

[0041] Correction execution parameters are obtained according to the deformation prediction parameters; the correction execution parameters are used to instruct the machine tool to adjust the prefabricated component according to the correction execution parameters.

[0042] In a third aspect, the present application also provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, it implements the steps of the high-precision automatic adjustment and calibration method for the production and assembly of any of the above-mentioned prefabricated components.

[0043] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the high-precision automatic adjustment and calibration method for the production and assembly of any of the above-mentioned prefabricated components.

[0044] This high-precision automatic adjustment and calibration system for prefabricated component production and assembly predicts thermal and shrinkage deformation during the molding process, proactively implementing structural compensation and mold adjustment strategies. This significantly reduces late-stage error accumulation and improves prefabricated component molding accuracy. Combining real-time positioning with predictive displacement error control, it ensures accurate component positioning, minimized seams, and high structural stability during component installation, improving component assembly alignment accuracy. By integrating data-driven and physical models, the system establishes an automated decision-making channel between design and construction, replacing traditional manual measurement and adjustment processes with algorithms and control, reducing reliance on human experience and enhancing intelligence. The deformation prediction and compensation mechanism is independent of a single component shape or operating parameter; it can be adaptively modeled and adjusted based on data. It is suitable for a variety of prefabricated scenarios, including bridges, subways, factories, and prefabricated housing, enhancing the system's environmental adaptability and versatility. By combining feedforward control with closed-loop correction, it avoids improper installation and subsequent rework caused by component errors, indirectly saving labor and material costs, and significantly reducing error-related rework and resource waste. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the following briefly introduces the drawings required for use in the embodiments or related technical descriptions. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0046] Figure 1 A structural diagram of the high-precision automatic adjustment and calibration system for the production and assembly of prefabricated components of the present invention;

[0047] Figure 2 Schematic diagram of the process of the high-precision automatic adjustment and calibration method for the production and assembly of prefabricated components of the present invention. DETAILED DESCRIPTION

[0048] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0049] In one embodiment, Figure 1 As shown, a high-precision automatic adjustment and calibration system for the production and assembly of prefabricated components is provided. This embodiment uses the system applied to a terminal as an example for illustration. It is understood that the system can also be applied to a server, or to a system including a terminal and a server, and implemented through the interaction between the terminal and the server. In this embodiment, the system includes the following steps:

[0050] The multi-sensor acquisition layer is used to obtain the full-process status data of prefabricated components; the full-process status data includes component data and concrete material characteristic data.

[0051] In principle, the full-process status data refers to a collection of various types of information related to deformation during the entire process of covering components, including raw material mold placement in the early stage of production, mid-term maintenance monitoring of forming, dynamic response during transportation, and final assembly and positioning, including component data and concrete material property data.

[0052] Component data is the result of multi-dimensional perception of the physical component itself. Specifically, a high-precision laser scanning device can be used to collect point cloud data on the surface of prefabricated components in real time. Point cloud data is a data set consisting of three-dimensional spatial coordinate points that describes the contour and size of the component surface. A high-speed laser scanner can scan the component contour in full coverage to obtain information such as the component's actual shape error, edge offset, and surface irregularities. To assist in the semantic interpretation of point cloud data, a multi-angle visual imaging system can be used to obtain image data of the component for identifying non-geometric properties such as surface defects, crack distribution, and mark location. Image data can be combined with point cloud data to form a semantically enhanced spatial model. Furthermore, to obtain the contact stress, stiffness response, and boundary load conditions of the component, a contact sensor array can be integrated, including embedded piezoelectric force sensors and boundary strain gauges, to record the real-time stress state of the component during production or lifting, thereby reflecting the stress response distribution of the component under actual use. Furthermore, infrared temperature imaging and embedded temperature sensors can be used to capture the temperature evolution of concrete during the forming and curing stages in real time, identifying internal temperature gradients and assessing deformation caused by these differences. Optionally, during transportation, the system can also track component displacement, impact loads, and vibration responses using accelerometers and position sensors, providing dynamic boundary conditions for subsequent state inference.

[0053] Among them, the concrete material property data is directly provided by the material database and the mixing station data acquisition interface, mainly including the water-cement ratio of concrete raw materials, cement type, aggregate grading, admixture type and dosage and other proportioning parameters, as well as elastic modulus, etc., to reflect the deformation characteristics of concrete at different ages.

[0054] The digital twin modeling layer is used to obtain a digital component model based on component data; the digital component model includes temperature field, stress field, strain tensor field and solidification state field.

[0055] The digital twin modeling layer constructs a virtual mirror model—the digital component model—that dynamically responds to the actual component state. This allows the transformation of multimodal state data collected during the actual production process of prefabricated components into a computable and analyzable digital representation. Component data is typically derived from the multi-sensor acquisition layer, including point cloud data, image data, contact measurement data, temperature distribution information, and internal stress measurement information. The digital twin modeling layer preprocesses this component data, performing denoising, alignment, standardization, and time synchronization on the raw data to ensure consistency across the spatial and temporal dimensions of the multimodal data, facilitating subsequent fusion modeling.

[0056] During the construction of digital component models, different types of data are mapped to different physical fields. Point cloud data is primarily used to restore the spatial geometry of the component and serves as the basis for generating the component's solid model. Image data is used to identify the component's surface condition, texture features, and potential external defects, further refining the accuracy of the geometric model. Contact measurement data helps calibrate key structural boundaries of the 3D contour and improve model dimensional accuracy. Temperature distribution data and internal stress data are mapped to the temperature field and stress field, respectively, to express the dynamic changes in the component's current thermal and stress states.

[0057] The deformation prediction layer is used to perform deformation analysis on the component concrete shrinkage characteristics and steel bar stress distribution based on the digital component model to obtain deformation prediction parameters.

[0058] The deformation prediction layer extracts multi-physics information from the digital twin model. Based on the component's current and future curing state, temperature field, and stress-strain distribution, it identifies and predicts potential deformation trends during precast component production and assembly. This provides a quantifiable basis for precise control and compensation. The system also incorporates concrete material property data, including the mix ratios of cement, aggregate, and admixtures, strength grade, elastic modulus, expansion coefficient, and other parameters that influence component volume change, strength growth rate, and stress conductivity. Schematically, the strain tensor field is generated based on the material's elastic-plastic model and the temperature-humidity coupling mechanism, with local strain calculation achieved through finite element numerical analysis. The curing state field is predictively modeled based on the concrete's hydration kinetics and age-development curves. This model identifies the degree of curing of concrete at different locations at different times, reflecting the entire process of component performance development. The core task of this layer is to extract the geometric offsets and structural deformations caused by hydration shrinkage, temperature stress gradients, and reinforcement constraints from the complex spatial-temporal dynamics represented in the digital component model, thereby deriving deformation prediction parameters.

[0059] Optionally, the deformation prediction parameter is a composite variable set that can include spatial feature information such as linear dimension offset, curvature change, and angle offset. If the current deformation exceeds the allowable deviation, the system will pass the prediction parameter to the precision control execution compensation layer to adjust the component processing path.

[0060] The precision control execution compensation layer is used to obtain correction execution parameters based on the deformation prediction parameters; the correction execution parameters are used to instruct the machine tool to adjust the prefabricated components according to the correction execution parameters.

[0061] Based on the deformation prediction parameters output by the deformation prediction layer, the precision control execution compensation layer generates corrective execution parameters recognizable by the machine tool control system. These parameters are transmitted in real time to the processing and assembly terminals, enabling them to adjust the component's shape, positioning datum, or assembly interface without changing the component's core structural performance, thereby achieving closed-loop correction of geometric errors. Based on the fundamental design principle of fine-tuning instead of rework, errors are proactively identified and precisely pre-compensated during the manufacturing process, avoiding the need for entire components to be scrapped or processed on-site due to accumulated deviations.

[0062] Schematically, the precision control execution compensation layer performs semantic translation of the deformation prediction parameters, that is, converts the spatial offset, angle change or curvature change output by the finite element simulation into the corresponding process adjustment variables. Specifically, the system predicts that the free end of the beam will be shortened by 1.5mm due to concrete shrinkage, so it is necessary to pre-extend the insertion depth by 1.5mm when anchoring the steel bars, or adjust the corresponding distance outward when positioning the formwork. If an arching trend is found, it is necessary to adjust the lateral formwork support force or set a local reverse pressure device for flattening. In the actual control system, this adjustment is represented by a correction execution parameter table, which clearly defines the adjustment target, execution method, actuator and control value.

[0063] Optionally, the system can incorporate real-time feedback from multiple sensor acquisition layers during execution. For example, point cloud scan results can be used to verify that corrective actions have achieved their objectives. If residual errors still exist, the system will adaptively modify execution parameters and perform secondary compensation, forming a closed-loop control chain. To ensure that compensation actions have no negative impact on structural safety, the control system incorporates a safety boundary constraint mechanism, prohibiting adjustments that exceed the material's allowable deviation range. The entire execution path is recorded and compared with the original model, ensuring full process traceability.

[0064] In this high-precision automatic adjustment and calibration system for precast component production and assembly, the multi-sensor acquisition layer integrates multiple sensing methods, including laser scanning point clouds, high-definition image acquisition, contact displacement measurement, temperature distribution monitoring, and internal stress acquisition, to capture state information throughout the entire precast component process, from mold casting and curing to demolding and assembly. This process not only achieves high coverage and precision, but also ensures data integrity from the macroscopic morphology to the microscopic strain level. By jointly collecting component data with concrete material properties, this layer provides a realistic, continuous, and semantically explicit multimodal information foundation for subsequent modeling and analysis, significantly improving data quality and modeling reliability. The digital twin modeling layer reconstructs a 3D geometric mesh model based on standardized component data and integrates spatiotemporal attributes such as temperature, stress, and strain into model nodes, creating a real-time, updatable digital component model. This model not only dynamically reflects the structural state of the precast component but also provides spatial resolution of thermodynamic parameters. This provides a comprehensive structure-state-property modeling foundation for subsequent physics-driven predictions, enabling a bidirectional mapping between physical and cyberspace. The deformation prediction layer incorporates concrete shrinkage characteristics and reinforcement stress distribution into the dynamic prediction process. Through thermal diffusion simulation, time-varying elastic modulus modeling, and coupled thermal-stress field analysis, the system predicts potential deformation trends under conditions such as temperature gradients, shrinkage variations, and constraint release, even before actual precast component deformation occurs. Furthermore, the data-driven prediction model incorporates neural networks and time series analysis mechanisms to further enhance prediction sensitivity and generalization. The resulting fused deformation prediction parameters are highly timely, reliable, and possess high structural resolution, enabling the generation of differentiated and customized correction references for each component. The precision control execution compensation layer intelligently generates correction execution parameters based on the deformation prediction results, including mold reversal force, hot zone temperature control commands, and fine-tuning actuator parameters. This layer possesses significant feedforward control capabilities, enabling preemptive adjustments to molding and assembly strategies, effectively mitigating component precision deviations caused by curing shrinkage, temperature variations, and material inhomogeneity. This layer directly interfaces with the machine tool control system or assembly robot control system, enabling millimeter-level precision in control command issuance, ensuring that each component maintains design accuracy during demolding and assembly.

[0065] In one embodiment, the digital twin modeling layer includes a data processing module, a physical mesh reconstruction module, an attribute fusion mapping layer, and a component status module;

[0066] The data processing module is used to pre-process the component data to obtain a standard data set; the standard data set includes point cloud data, edge feature data, curing temperature spatiotemporal data and strain distribution data.

[0067] The data processing module is responsible for extracting effective information from multi-source data and unifying the data structure and temporal and spatial benchmarks. Schematically, the input data sources primarily include point cloud data collected by laser scanning equipment, precise measurements such as edge holes provided by contact sensors, temperature distribution images acquired by infrared arrays, internal stresses acquired by strain gauges, and component edge images provided by structural edge contour detection cameras. Because these various data types vary significantly in acquisition accuracy, sampling frequency, spatial resolution, and noise levels, the data processing module must standardize them. Specifically, point cloud data can be processed through noise filtering algorithms such as Statistical Outlier Removal, ground removal, and coordinate alignment. Edge images require image enhancement and contour extraction algorithms such as Canny to extract edge feature curves. Temperature and strain data require temporal resampling, coordinate conversion, and unit unification. Ultimately, a structured standard dataset is formed: a fused, multidimensional data package consisting of point cloud data, edge feature data, curing temperature spatiotemporal data, and strain distribution data.

[0068] The solid mesh reconstruction module is used to reconstruct the mesh based on the point cloud data to obtain a three-dimensional geometric mesh model of the component; the three-dimensional geometric network model of the component includes mesh nodes.

[0069] The function of the solid mesh reconstruction module is to restore the three-dimensional geometric shape of the component based on the component point cloud data and construct a meshed geometric model structure. Schematically, point cloud data is essentially a collection of discrete sampling points on the component surface, which needs to be reconstructed into a continuous surface through surface fitting or spatial triangulation methods. The triangular mesh model can be generated using the Poisson Surface Reconstruction method or the surface approximation algorithm based on the Delaunay triangulation. The mesh model consists of nodes, edges and faces, where nodes are the basic units for subsequent physical property mapping. During the construction process, the system retains all micro-scale details that match the scanning accuracy to ensure that the geometric model does not lose small deformations or irregular parts. In addition, model reconstruction also includes processing links such as normal vector calculation, mesh simplification, and hole filling to form a three-dimensional geometric mesh model with a complete topological structure that can be used for subsequent finite element analysis.

[0070] The attribute fusion mapping module is used to map the curing temperature spatiotemporal data and strain distribution data to the grid nodes through interpolation to obtain the attribute tensor field.

[0071] The attribute fusion mapping module is used to map the physical field data, i.e., the spatiotemporal data of curing temperature and the strain distribution data, to the geometric mesh model constructed above, thereby forming a set of physical-geometric fusion digital component models. Specifically, by spatially projecting each temperature or strain measurement value to the nearest mesh node, or using three-dimensional interpolation methods such as Inverse Distance Weighting, a continuous tensor field is constructed in the mesh domain. For example, if the top area of a component heats up faster than the bottom area during the curing process, the mapping process will form a higher temperature tensor value at the top node; similarly, the strain data will be reflected in the tensor gradient between nodes, forming a continuous strain tensor field. This tensor field records multi-dimensional tensor attributes such as temperature distribution (T), stress state (σ), and strain distribution (ε) in units of nodes through data structures, providing a basis for predictive analysis.

[0072] The component status module is used to perform component boundary and structural topology based on edge feature data, and to obtain a digital component model by combining the component's three-dimensional geometric mesh model and attribute tensor field.

[0073] The component state module is the final output link of the entire modeling layer, and its purpose is to integrate the geometric model, edge features, and attribute tensors into a complete digital component model. In this module, the boundary lines are matched and corrected using edge feature data to ensure that the boundary contour of the mesh model is consistent with the real component. Subsequently, the structural topology analysis algorithm is used to identify possible connection relationships, cavity structures, and internal nested features in the model, which can be used to verify key parameters such as the accuracy of the assembly interface position and the symmetry of the steel bar layout. Furthermore, the aforementioned three-dimensional geometric mesh and tensor attributes are bound through a unified data structure to form a digital component model with geometric, thermal, and topological integrity. This model can be used for static observation and visualization, as well as as a direct input to the deformation prediction layer and the control compensation layer, and plays an intermediate role in the entire system of "state expression-prediction analysis-control output".

[0074] In one embodiment, the deformation prediction layer includes a physical model simulation module, a data-driven prediction module, and a composite fusion output module;

[0075] The physical model simulation module is used to perform finite element simulation of the heat-stress-deformation relationship based on the digital component model using concrete material property data to obtain predicted deformation displacement; concrete material property data includes material thermal conductivity parameters and material body elastic modulus.

[0076] The physical model simulation module constructs a finite element simulation model based on physical mechanisms to simulate the coupled thermal-mechanical-deformation processes within concrete components during various stages, such as curing and cooling. Its input is the digital component model output by the digital twin modeling layer and concrete material property data. The digital component model provides the component's 3D geometry, node structure, and initial temperature / strain tensor distribution. The concrete material property data includes essential thermophysical and mechanical parameters such as thermal conductivity, specific heat capacity, volumetric heat source rate, elastic modulus, and Poisson's ratio. The module employs the thermo-elastic-plastic finite element method to construct a set of thermal stress equations based on the component's curing heat source, environmental heat exchange boundary conditions, and constraint boundaries. The model iteratively solves the thermal expansion, contraction, and stress responses of each mesh node under the influence of the temperature field, and then derives the corresponding strain and displacement responses, forming a time-series predicted deformation and displacement field. This module is particularly suitable for identifying nonuniform deformation caused by thermal gradients or internal force concentrations induced by reinforcement constraints.

[0077] The data-driven prediction module is used to extract the characteristic tensor structure from the digital component model and obtain the data-driven predicted deformation field based on time series analysis.

[0078] As a supplementary path, the data-driven prediction module learns the temporal variation patterns from the curing behavior of existing historical components, explores the statistical relationship between high-dimensional tensor field data and actual deformation results, and realizes the prediction of future states. Specifically, the module extracts characteristic tensor structures from the digital component model, including the node temperature change rate tensor, initial strain tensor, geometric grid density tensor, boundary constraint tensor, etc. In the data preparation stage, the module fits the tensor characteristics of historical components in different states with the actual measured displacement / stress response, and dynamically encodes and decodes the input tensor flow by constructing a recursive neural network structure based on GRU or a time series prediction network with Transformer temporal attention mechanism to predict the deformation evolution field at key moments in the future. Since the module relies on a large amount of historical component construction and measurement data for training, it has stronger generalization ability when there are a large number of prior component samples, and is particularly suitable for processing situations where there are many types of components and high heterogeneity.

[0079] The composite fusion output module is used to fuse the predicted deformation displacement and the data-driven predicted deformation field to obtain the deformation prediction parameters.

[0080] Schematically, the predicted deformation displacement field output by the physical simulation module and the predicted deformation tensor field output by the data-driven module are spatially aligned and dimensionally unified, and then a weighted fusion algorithm is used to achieve a fused prediction output. Specifically, the system statistically analyzes the difference residuals between the two prediction results at each node to determine their credibility in a specific area. Then, based on the residual weights, the fusion proportions of different prediction sources are assigned to generate a unified set of deformation prediction parameters. The data format output after fusion includes but is not limited to: deformation variables of key parts of the component, deformation direction, time point prediction offset, and stress concentration area grade assessment.

[0081] In one embodiment, a finite element simulation of the thermal-stress-deformation relationship is performed based on the digital component model using concrete material property data to obtain predicted deformation displacement, including:

[0082] S11. Based on the temperature field of the digital component model and the thermal conductivity parameters of the material, a heat diffusion simulation is performed to obtain the predicted temperature field at the future moment.

[0083] S12. Based on the stress field and strain tensor field of the digital component model, combined with the elastic modulus of the material itself and the predicted temperature field at future moments, the time-varying elastic modulus is obtained.

[0084] S13. Based on the predicted temperature field and time-varying elastic modulus at future moments, a thermal-stress field is combined to obtain a stress tensor.

[0085] S14. Predicting the deformation variable based on the stress tensor to obtain the predicted deformation displacement.

[0086] Schematically, the digital component model includes distribution information for fundamental physical quantities such as temperature, stress, and strain tensor fields. Based on current state information, finite element simulation predicts the component's deformation trends due to the coupling of thermal diffusion and stress over future periods of time, ultimately outputting a predicted displacement field to facilitate subsequent precise control and compensation.

[0087] Specifically, heat diffusion simulation is performed based on the heat conduction equation (a combination of the Fourier Law and energy conservation):

[0088]

[0089] Where ρ is the material density, c is the specific heat capacity, k is the thermal conductivity, Q is the heat source intensity per unit volume, and T(x,y,z) is the time-varying temperature field. By simulating the current heat source conditions and the material's thermal conductivity, the temperature distribution at future time points can be deduced using mesh discretization and explicit / implicit time stepping algorithms.

[0090] On the basis of thermal diffusion, since concrete is a temperature-sensitive material, its elastic modulus decreases as the temperature increases:

[0091] E(T)=E0·(1-α E (T-T0))

[0092] Among them, E0 is the initial elastic modulus, α E is the temperature sensitivity coefficient, and T0 is the reference temperature.

[0093] Based on the elastic modulus, the stress tensor is predicted under thermal-stress coupling:

[0094]

[0095] Among them, σ xx To predict the stress component, v is the Poisson's ratio, ∈ xx is the strain component. This relationship can be generalized to a six-dimensional constitutive matrix expression in the form of Voigt vectors in the three-dimensional space tensor.

[0096] Furthermore, the predicted stress components derived from the known strain tensor distribution and constitutive relationship are combined with the relationship between nodes and boundary conditions to perform iterative deduction of finite element deformation variables, and finally the displacement prediction value of each grid node is obtained to form a three-dimensional predicted deformation field, and the predicted deformation displacement is obtained.

[0097] In one embodiment, obtaining the correction execution parameter according to the deformation prediction parameter includes:

[0098] S21. Generate a reverse force compensation parameter corresponding to mold pressurization according to the regional shrinkage of the component in the deformation prediction parameter.

[0099] S22. Generate temperature adjustment parameters corresponding to the mold heating plate area according to the asymmetric deformation of the component in the deformation prediction parameters.

[0100] S23. Generate corresponding fine-tuning actuator correction parameters according to the component displacement in the deformation prediction parameters.

[0101] S24: Determine the reverse compensation parameter, the temperature adjustment parameter, and the fine-tuning actuator correction parameter as correction execution parameters.

[0102] Schematically, a reverse-force control mechanism is developed by providing targeted responses and physical compensation for multiple potential component errors or deformations. This ensures that the dimensions, morphology, and mechanical state of prefabricated components meet high-precision manufacturing requirements during production, curing, and assembly. The specific process involves generating reverse force compensation parameters, mold heating plate temperature adjustment parameters, and fine-tuning actuator calibration parameters. These parameters are then aggregated into corrective execution parameters and applied to the control module of the machine tool or mold system.

[0103] For example, based on the values of the deformation prediction parameters regarding the regional shrinkage of the component, the system identifies local areas where cohesive shrinkage or directional shrinkage may occur. Specifically, when the temperature in the center of the component drops faster than the periphery, the material undergoes tensile stress concentration due to inward contraction, which can easily cause the center to sink or the edge to bulge. To cope with this deformation trend, the system needs to design a reverse force for the mold structure to achieve active compensation. The reverse force compensation parameter is generated by the following formula:

[0104] F comp (x,y,z)=-k mold ·Δ∈(x,y,z)·A

[0105] Among them, F comp is the reverse force per unit area that the mold needs to apply, Δ∈(x,y,z) is the predicted local shrinkage strain, k mold is the mold stiffness factor, A is the pressure surface area, and the force is mapped to the hydraulic unit or servo pressurizing system of the mold. The mold clamping force is controlled according to the spatial distribution lattice to perform mechanical reverse adjustment.

[0106] For example, in the case of asymmetric heat-induced deformation in the predicted component deformation, the system adjusts the local temperature distribution of the mold heating plate to perform equivalent thermal field correction. Specifically, if a component edge solidifies prematurely due to rapid heat dissipation, forming a tendency to warp, the local heating plate can be heated to delay the thermal field curing and coordinate the synchronous stress of the structure. The adjustment parameters are generated based on the following formula:

[0107]

[0108] Where ΔT heat is the target temperature rise value of the heating plate, Δσ asym is the differential stress caused by the uneven stress prediction in the region, β is the material thermal stress coupling coefficient, and E is the regional elastic modulus. This temperature adjustment value can be loaded into an embedded temperature-controlled resistor or heat pump system, and the PLC module drives the heating plate area to achieve dynamic control of the local temperature field.

[0109] For example, based on the predicted displacement parameters, correction instructions are generated for the corresponding fine-tuning actuators. Fine-tuning actuators are generally distributed in the mold micro-motion platform or component positioning fixture to achieve millimeter-level assembly correction. If there is an offset in the predicted displacement field, whether global or local, the system will extract the corresponding displacement vector and convert it into an execution instruction:

[0110] δ(x,y,z)=u pred (x,y,z)-u ref (x,y,z)

[0111] Among them, δ is the displacement value that the correction actuator needs to push, upred To predict the component node displacement, u ref It is a standard reference position and can be used to drive micro-motion mechanisms such as electric push rods, spiral pressure plates or precision guide rails, so as to fine-tune the position of components before solidification or in the middle of cooling to improve assembly alignment accuracy.

[0112] Ultimately, the system aggregates the reverse force compensation parameters, temperature adjustment parameters, and fine-tuning displacement correction parameters into a unified "correction execution parameter" data package. This data package is sent via an internal bus to the industrial control system, which coordinates and controls the mold's pressurization, heating, and micro-calibration subsystems, ensuring synchronized responses and dynamic collaboration. This achieves closed-loop compensation for component curing and positioning errors. In practice, the system can be integrated into industrial automation platforms such as the Siemens S7-1500 or Schneider M580 control systems, controlling various execution units through PLC logic blocks and analog / digital output modules. During system operation, sensor feedback can be combined to form a feedback control closed loop, updating execution parameters in real time to ensure accurate and effective compensation during dynamic adjustment. This complete mechanism significantly improves the dimensional consistency of precast components and their subsequent assembly alignment, making it suitable for industrial intelligent manufacturing tasks in areas such as subway segments, bridge sections, and high-precision prefabricated floor slabs.

[0113] In one embodiment, the data processing module is used to pre-process the component data to obtain a standard data set, including:

[0114] S31. Based on the calibration model, the three-dimensional point cloud data and contact measurement data in the component data are spatially aligned, and the external parameter matrix is used to perform unified coordinate transformation to obtain point cloud data.

[0115] S32. Extract geometric features from the image data in the component data to obtain edge feature data.

[0116] S33. De-noise the solidification temperature distribution and internal stress in the component data to obtain solidification temperature spatiotemporal data and strain distribution data.

[0117] S34. Encapsulate the point cloud data, edge feature data, curing temperature spatiotemporal data, and strain distribution data into a standard data set.

[0118] Schematically, the module spatially aligns the three-dimensional point cloud data and contact measurement data in the component data. Three-dimensional point cloud data usually comes from laser scanning equipment, which is dense, unstructured, and high-resolution, but its coordinate system is based on the installation posture of the laser rather than the component design reference system. Contact measurement data comes from devices such as electric micrometers and coordinate measuring arms. Although they are highly accurate, they are limited in quantity and are usually used for key dimension review. In order to unify these spatial data, the system pre-builds a calibration model that includes the relative position relationship between the laser scanning equipment, the contact measurement probe, and the component positioning fixture. Under this model, the system performs coordinate transformation through the external parameter matrix, which is as follows:

[0119] P world =R·P local +T

[0120] Among them, P local is the point cloud or contact point in the original acquisition coordinates, R is the rotation matrix, and T is the translation vector. After batch conversion of all points, the system uniformly maps the laser point cloud and contact points to the component design reference system, generating spatially aligned point cloud data with real-world geometric reference significance, serving as the skeleton data source for subsequent 3D modeling.

[0121] Furthermore, the module extracts geometric features from the image data within the component data. This image data typically comes from industrial camera systems and is used to record component surface texture and edge information. Because images can be subject to lighting interference, occlusion, and distortion, the system can employ image enhancement techniques such as median filtering and gamma correction for preprocessing. Edge recognition is then performed on the image using the Canny operator, Sobel filtering, or a deep learning-based edge detection model to extract the component's geometric feature lines, including key geometric information such as the outer contour, internal grooves, and reserved hole locations. Ultimately, the edge features are encoded as vectors, with each boundary represented by a set of connected points, forming edge feature data that can be used for topological modeling.

[0122] The module performs targeted noise removal and spatiotemporal reconstruction on the curing temperature distribution and internal stress data in the component data. The temperature distribution data comes from thermocouples or fiber optic temperature sensors placed inside the component. Their signals may be affected by factors such as environmental fluctuations and electromagnetic interference, resulting in anomalies such as mutations and drift. The system uses sliding window median filtering, Kalman filtering, and heat conduction equation correction to smooth and interpolate the original temperature data, obtaining the temperature evolution trajectory of each key point inside the component on a continuous time scale, forming the "curing temperature spatiotemporal data." Similarly, stress and strain data comes from strain gauges or fiber optic layout systems, and is also preprocessed to eliminate occasional errors and reconstruct the true strain tensor distribution as "strain distribution data."

[0123] The point cloud data, edge feature data, curing temperature spatiotemporal data, and strain distribution data are encapsulated as a standard data set.

[0124] In one embodiment, the system further comprises an assembly calibration layer;

[0125] The assembly calibration layer is used to compare the real-time measured installation position data with the preset component positioning accuracy in real time and generate calibration instructions; the calibration instructions are used to instruct the actuator to fine-tune the component installation position; the calibration instructions include assembly position parameters and assembly angle parameters.

[0126] Schematically, the assembly calibration layer is used to achieve high-precision, automated alignment during component installation. This layer compares real-time measurement data with preset component positioning accuracy and generates calibration instructions with millimeter-level control accuracy. This guides the mechanical actuators to fine-tune the components, achieving intelligent alignment and precise assembly.

[0127] The system deploys high-precision spatial measurement devices, including optical measurement systems, in the component pre-assembly area. During the installation process, the device obtains the component's position information in three-dimensional space in real time, namely the position parameters and attitude angles, which constitute the real-time installation position data of the current component. For example, this data is formatted as a six-dimensional state vector P actual =(x, y, z, α, β, γ). Correspondingly, the system defines the target positioning parameters of each component in the structure according to the architectural design drawings, construction specifications and construction working conditions, that is, the preset component positioning accuracy. These accuracy parameters include both position tolerance and angle tolerance, and are loaded as the target reference state vector P when the system is initialized. target =(x0, y0, z0, α0, β0, γ0), the deviation recognition module of the assembly calibration layer calculates the assembly error of the current component in the form of vector difference ΔP = P actual -P target =(Δx, Δy, Δz, Δα, Δβ, Δγ), where each item represents position offset and attitude deflection, respectively. If any offset exceeds a set threshold, the system identifies it as an assembly deviation and automatically activates the calibration instruction generation module, which translates the offset vector into control instructions recognizable by the actuator. These include assembly position parameters and assembly angle parameters. The assembly position parameters guide linear displacement actuators (such as hydraulic cylinders and electric linear actuators) to perform micron-level displacement or lifting of the component to compensate for Δx, Δy, and Δz. The assembly angle parameters are used by a servo turntable or universal platform to precisely rotate the component about its three principal axes, thereby offsetting the attitude errors Δα, Δβ, and Δγ, generating a six-degree-of-freedom calibration instruction. The instruction is then sent to assembly actuators, such as multi-axis robots, three-dimensional translation platforms, and hoisting slides with feedback mechanisms. These actuators perform fine adjustments to the component based on the received calibration instructions.

[0128] In actual construction, this assembly calibration layer can be applied to a variety of construction scenarios, including bridge segment assembly, prefabricated floor slab installation, and the placement of special-shaped components. The introduction of the assembly calibration layer not only effectively reduces reliance on construction personnel experience but also significantly improves assembly efficiency and component joint accuracy. It is particularly suitable for assembling special-shaped, high-precision, or large-volume prefabricated components.

[0129] It should be understood that, although the various steps in the flowcharts involved in the various embodiments described above are displayed in sequence according to the instructions of the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the flowcharts involved in the various embodiments described above can include multiple steps or multiple stages, and these steps or stages are not necessarily executed and completed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of steps or stages in other steps.

[0130] Based on the same inventive concept, the embodiments of the present application also provide a method for high-precision automatic adjustment and calibration of prefabricated component production and assembly for implementing the high-precision automatic adjustment and calibration system for prefabricated component production and assembly. The solution to the problem provided by this device is similar to the solution described in the above-mentioned system. Therefore, the specific limitations of one or more embodiments of the method for high-precision automatic adjustment and calibration of prefabricated component production and assembly provided below can be found in the above-mentioned limitations of the high-precision automatic adjustment and calibration system for prefabricated component production and assembly, and will not be repeated here.

[0131] In an exemplary embodiment, Figure 2 As shown, a high-precision automatic adjustment and calibration method for the production and assembly of prefabricated components is provided, comprising:

[0132] S201. Acquire the whole-process status data of the prefabricated component; the whole-process status data includes component data.

[0133] S202. Obtain a digital component model according to the component data.

[0134] S203. Perform deformation analysis on the component concrete shrinkage characteristics and steel bar stress distribution based on the digital component model to obtain deformation prediction parameters.

[0135] S204 , obtaining correction execution parameters according to the deformation prediction parameters; the correction execution parameters are used to instruct the machine tool to adjust the prefabricated component according to the correction execution parameters.

[0136] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.

[0137] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0138] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to the partial description of the method embodiments. The device embodiments described above are merely illustrative, wherein the components described as separate parts may or may not be physically separated, and the parts displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the disclosed solution. A person of ordinary skill in the art can understand and implement it without expending creative work.

[0139] The above-described embodiments merely represent several implementation methods of the embodiments of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the patent application. It should be noted that a person skilled in the art may make various modifications and improvements without departing from the concept of the embodiments of the present application, and these modifications and improvements fall within the scope of protection of the embodiments of the present application.

Claims

1. A high-precision automatic adjustment and calibration system for the production and assembly of prefabricated components, characterized in that: The system comprises: The multi-sensor acquisition layer is used to obtain the status data of the entire process of prefabricated components; the entire process status data includes component data and concrete material characteristic data; A digital twin modeling layer, configured to obtain a digital component model based on the component data; the digital component model includes a temperature field, a stress field, a strain tensor field, and a solidification state field; A deformation prediction layer is used to perform deformation analysis on the component concrete shrinkage characteristics and steel bar stress distribution based on the digital component model to obtain deformation prediction parameters; The precise control execution compensation layer is used to obtain correction execution parameters according to the deformation prediction parameters; the correction execution parameters are used to instruct the machine tool to adjust the prefabricated component according to the correction execution parameters.

2. The system according to claim 1, wherein: The digital twin modeling layer includes a data processing module, a physical grid reconstruction module, an attribute fusion mapping layer and a component status module; The data processing module is used to pre-process the component data to obtain a standard data set; the standard data set includes point cloud data, edge feature data, curing temperature spatiotemporal data and strain distribution data; The reconstructed entity mesh module is used to reconstruct the mesh according to the point cloud data to obtain a three-dimensional geometric mesh model of the component; the three-dimensional geometric network model of the component includes mesh nodes; The attribute fusion mapping module is used to map the curing temperature spatiotemporal data and the strain distribution data to the grid nodes through interpolation to obtain an attribute tensor field; The component state module is used to perform component boundary and structural topology according to the edge feature data, and obtain the digital component model by combining the component three-dimensional geometric grid model and the attribute tensor field.

3. The system according to claim 1, wherein: The deformation prediction layer includes a physical model simulation module, a data-driven prediction module and a composite fusion output module; The physical model simulation module is used to perform a finite element simulation of the heat-stress-deformation relationship based on the digital component model using the concrete material property data to obtain a predicted deformation displacement; the concrete material property data includes material thermal conductivity parameters and material body elastic modulus; The data-driven prediction module is used to extract characteristic tensor structures from the digital component model and obtain a data-driven predicted deformation field based on time series analysis; The composite fusion output module is used to fuse the predicted deformation displacement and the data-driven predicted deformation field to obtain deformation prediction parameters.

4. The system according to claim 3, characterized in that The method of performing a finite element simulation of a heat-stress-deformation relationship based on the digital component model using the concrete material property data to obtain a predicted deformation displacement includes: Performing a heat diffusion simulation based on the temperature field of the digital component model and the material thermal conductivity characteristic parameters to obtain a predicted temperature field at a future moment; Based on the stress field and the strain tensor field of the digital component model, combined with the elastic modulus of the material body and the temperature field predicted at the future moment, a time-varying elastic modulus is obtained; Performing a thermal-stress field combination based on the predicted temperature field at the future moment and the time-varying elastic modulus to obtain a stress tensor; The deformation amount is predicted according to the stress tensor to obtain the predicted deformation displacement.

5. The system according to claim 1, wherein: Obtaining correction execution parameters according to the deformation prediction parameters includes: generating a reverse force compensation parameter corresponding to mold pressurization according to the regional shrinkage of the component in the deformation prediction parameter; generating temperature adjustment parameters corresponding to the mold heating plate area according to the asymmetric deformation of the component in the deformation prediction parameters; generating corresponding fine-tuning actuator correction parameters according to the component displacement in the deformation prediction parameters; The reverse compensation parameter, the temperature adjustment parameter and the fine-tuning actuator correction parameter are determined as the correction execution parameter.

6. The system according to claim 2, wherein: The data processing module is used to pre-process the component data to obtain a standard data set, including: Performing spatial alignment on the three-dimensional point cloud data and the contact measurement data in the component data based on the calibration model, and performing unified coordinate transformation using an external parameter matrix to obtain the point cloud data; Extracting geometric features from the image data in the component data to obtain edge feature data; De-noising the curing temperature distribution and internal stress in the component data to obtain the curing temperature spatiotemporal data and the strain distribution data; The point cloud data, the edge feature data, the curing temperature spatiotemporal data, and the strain distribution data are packaged into the standard data set.

7. The system according to claim 6, characterized in that The system further comprises an assembly calibration layer; The assembly calibration layer is used to compare the real-time measured installation position data with the preset component positioning accuracy in real time and generate calibration instructions; the calibration instructions are used to instruct the actuator to fine-tune the component installation position; the calibration instructions include assembly position parameters and assembly angle parameters.

8. A high-precision automatic adjustment and calibration method for the production and assembly of prefabricated components, characterized in that: The method comprises: Acquiring full-process status data of prefabricated components; the full-process status data includes component data; obtaining a digital component model according to the component data; Performing deformation analysis on the component concrete shrinkage characteristics and steel bar stress distribution based on the digital component model to obtain deformation prediction parameters; Correction execution parameters are obtained according to the deformation prediction parameters; the correction execution parameters are used to instruct the machine tool to adjust the prefabricated component according to the correction execution parameters.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to claim 8 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to claim 8 are implemented.

Images