Digital detection method and system for virtual pre-assembly of large bending and twisting components

Abstract

The method comprises the following steps: constructing an initial geometric model by acquiring three-dimensional coordinates of feature points on the surface of the component, introducing the initial geometric model into an intelligent pre-assembly analysis platform of a steel bridge, and then calling a curvature inversion model of the bending-torsion component to calculate the axis curvature and torsion angle parameters, so as to obtain the virtual pre-assembly of the large bending-torsion component. Correcting the spatial attitude of the component through the attitude calibration model of the special-shaped component, so that the model is consistent with the actual component; an image mark matching recognition algorithm is used for completing virtual and actual mark point matching positioning, a coordinate mapping relation is established, virtual pre-splicing is executed in a platform, splicing surface gaps, alignment tolerance and axis alignment deviation are calculated, and finally a digital detection report including curvature distribution, attitude parameters and splicing deviation is output. According to the method, through multi-model cooperation and full-process digital operation, the detection precision and efficiency are improved, and reliable technical support is provided for pre-assembly quality management and control of the large bending and twisting component.

Description

Technical Field

[0001] This invention relates to the field of pre-assembly technology for large bending and torsion components, and in particular to a digital inspection method and system for virtual pre-assembly of large bending and torsion components. Background Technology

[0002] Large bending and torsion components are core load-bearing parts of steel bridges, and their manufacturing precision and pre-assembly compatibility directly affect the overall load-bearing capacity and service safety of the bridge. As bridge engineering develops towards larger spans and more complex shapes, the requirements for curvature changes, spatial posture, and splicing precision of bending and torsion components continue to increase. Traditional inspection methods relying on manual measurement and physical pre-assembly are no longer sufficient to meet the demands of efficient and precise construction. Virtual pre-assembly technology, with its digital and visualization advantages, has become a key path to solving the pre-assembly challenges of large bending and torsion components. However, how to achieve accurate mapping between geometric models and actual components, posture calibration, and deviation quantification remains a critical technical challenge that the industry urgently needs to overcome. There is a pressing need to build a digital inspection method and system that integrates multi-model collaboration and intelligent algorithm support to provide technical assurance for the quality control of steel bridge construction.

[0003] Existing technologies have two significant shortcomings: First, traditional virtual pre-assembly inspection methods lack a precise mathematical model to support the geometric model construction and the posture calibration of actual components, relying heavily on empirical parameter adjustments. This leads to deviations between the model posture and the actual components, failing to accurately reflect the true curvature distribution and spatial positional relationships of the components, thus affecting the reliability of pre-assembly deviation calculations. Second, the intelligence level of marker point matching and coordinate mapping is insufficient, often employing single feature matching methods, which are easily affected by the component surface environment and measurement errors. This results in low accuracy of coordinate unification between virtual and actual components, and the pre-assembly deviation analysis lacks a multi-parameter coupled quantitative system, making it difficult to comprehensively and accurately identify key deviations such as splicing surface gaps, misalignment, and axis alignment, thus limiting the application value of the inspection results. Summary of the Invention

[0004] In order to overcome the shortcomings and deficiencies of existing technologies, this invention provides a digital inspection method and system for virtual pre-assembly of large bending and torsion components.

[0005] The technical solution adopted in this invention is a digital inspection method for virtual pre-assembly of large bending and torsion components, comprising the following steps: S1 Obtaining the three-dimensional coordinate data of discrete feature points on the surface of the large bending and torsion component, constructing an initial geometric model of the component based on the spatial distribution of the feature points, and extracting the component's axial direction vector, cross-sectional contour boundary point set, and relative distance parameters between feature points; S2 Importing the initial geometric model into a steel bridge intelligent pre-assembly analysis platform, calling the bending and torsion component curvature inversion model to calculate the component's axial curvature distribution, and obtaining the curvature values ​​and torsion angle parameters of each cross-section of the axial direction; S3 Using an irregular component attitude calibration model to adjust the attitude of the initial geometric model, and analyzing the deviation of the three-dimensional coordinates of the feature points. S4. Analyze and correct the spatial attitude parameters of the components to make the model attitude consistent with the actual components; S5. Use an image marker matching and recognition algorithm to match and locate the preset marker points on the component surface, obtain the precise coordinates of the marker points in the virtual pre-assembly coordinate system, and establish the coordinate mapping relationship between the virtual and actual components; S6. Based on the calibrated geometric model and coordinate mapping relationship, perform virtual pre-assembly operation in the steel bridge intelligent pre-assembly analysis platform to calculate the gap value, misalignment amount, and axis alignment deviation parameters of the component splicing surface; S7. Based on the deviation parameters generated during the virtual pre-assembly process, output the component pre-assembly adaptability test results, and form a digital test report including curvature distribution, attitude parameters, and splicing deviation.

[0006] Furthermore, the expression for the curvature inversion model of the bending-torsional member is: ,in, Let be the combined curvature at a point on the axis of the bending / torsional member. For the three-dimensional coordinates of the feature point, For the axis arc length parameter, This is the curvature correction factor. The torsional angle of the component section. The second derivative in the x-direction of the axis. The rate of change of the torsion angle along the axis. They are respectively The first derivative of the direction along the axis.

[0007] Furthermore, the expression for the attitude calibration model of the irregular component is: ,in, For attitude calibration, Adjust the direction vector components for attitude control. This represents the deviation value of the feature point coordinates. For attitude calibration weighting coefficients, The components are wound around The rotational deviation angle of the shaft.

[0008] Furthermore, the expression for the image marker matching and recognition algorithm is: ,in, To match the similarity of the marker points, These are virtual and real marker sets, respectively. Let i be the feature vector of the i-th marker point. For matching coefficients, The number of markers, The average distance between virtual and actual marker points. These are the standard spacing parameters.

[0009] Furthermore, the pre-assembly deviation calculation model of the intelligent pre-assembly analysis platform for steel bridges is as follows: ,in, For total pre-assembly deviation, This is the deviation weighting coefficient. For geometric model deviation, To correct attitude calibration deviation, This is due to coordinate mapping deviation. For the local deviation of the k-th splicing surface, This represents the number of spliced ​​surfaces.

[0010] Furthermore, the digital detection parameter coupling model for the virtual pre-assembly of the large bending and torsion component is as follows: ,in, To detect comprehensive parameters, The coupling coefficient is... For the overall curvature, For attitude calibration, To match similarity, For total pre-assembly deviation, The surface area of ​​the component. This represents the volume of the component.

[0011] Further, step S3 includes the following sub-steps: S31, extracting the three-dimensional coordinate data of the calibrated feature points in the initial geometric model, comparing them one by one with the measured coordinates of the corresponding feature points of the actual component, and calculating the coordinate deviation value of each feature point in the x, y, and z directions; S32, inputting the coordinate deviation value into the irregular component posture calibration model, and determining the optimal rotation angle and translation amount of the component around the x, y, and z axes through iterative calculation, so as to minimize the sum of squares of the deviations between the model feature point coordinates and the measured coordinates; S33, adjusting the spatial posture of the initial geometric model according to the calculated rotation angle and translation amount, updating the three-dimensional coordinates of all feature points in the model, and forming a preliminary calibrated geometric model; S34, verifying the deviation of the preliminary calibrated model. If the average deviation of the feature points exceeds a preset threshold, repeat steps S31 to S33 until the average deviation meets the detection accuracy requirements.

[0012] Further, step S4 includes the following sub-steps: S41, image acquisition of preset marker points on the surface of the component, extraction of shape features, grayscale features and spatial position features of each marker point, and construction of a marker point feature database; S42, generation of virtual marker points corresponding to actual marker points in the virtual pre-assembly model, and calculation of feature similarity between virtual and actual marker points using an image marker matching and recognition algorithm; S43, screening of marker point pairs with similarity higher than a set threshold, establishing a one-to-one correspondence between virtual and actual marker points, and determining the coordinate mapping matrix; S44, based on the coordinate mapping matrix, conversion of the three-dimensional coordinate data of the actual component to the virtual pre-assembly coordinate system, completing the coordinate unification between the virtual and actual components.

[0013] Further, S5 includes the following sub-steps: S51, importing the calibrated geometric model and the actual component data after coordinate unification into the intelligent pre-assembly analysis platform for steel bridges, setting pre-assembly constraints, including splicing surface fitting requirements, axis alignment standards, and allowable deviation ranges; S52, performing virtual splicing operations on the platform according to the preset pre-assembly sequence, calculating the gap value and misalignment of each splicing surface in real time, and recording the angular deviation of the axis at the splicing point; S53, monitoring the deviation data during the pre-assembly process in real time, and automatically adjusting the spatial position of the component and recalculating the splicing if the deviation of a certain splicing surface exceeds the allowable range; S54, after completing the virtual pre-assembly of all components, summarizing the deviation parameters of each splicing surface and the axis alignment status, and generating detailed pre-assembly deviation data.

[0014] A digital inspection system for virtual pre-assembly of large bending and torsion components is presented. This system, applied to the digital inspection method for virtual pre-assembly of large bending and torsion components, includes: a 3D coordinate data acquisition unit, used to acquire 3D coordinate data of surface feature points of the large bending and torsion components, establishes a data transmission connection with the intelligent pre-assembly analysis platform for steel bridges, and transmits the acquired coordinate data to the platform in real time; a geometric model construction and curvature inversion unit, which receives the coordinate data output by the 3D coordinate data acquisition unit, constructs the initial geometric model of the component, calls the curvature inversion model of the bending and torsion component to calculate the axial curvature distribution, and sends the model and curvature parameters to the attitude calibration unit; and an irregular component attitude calibration unit, which receives the model data transmitted by the geometric model construction and curvature inversion unit, and corrects the component space through the irregular component attitude calibration model. The attitude measurement unit transmits the calibrated model to the marker matching and positioning unit; the image marker matching and positioning unit collects image information of marker points on the component surface, uses an image marker matching and recognition algorithm to match and locate virtual and actual marker points, establishes a coordinate mapping relationship, and sends it to the virtual pre-assembly unit; the virtual pre-assembly and deviation calculation unit receives the attitude-calibrated model, coordinate mapping relationship, and platform preset parameters, performs virtual pre-assembly operations, calculates splicing deviation parameters, and transmits the deviation data to the inspection report generation unit; the digital inspection report generation unit receives the deviation parameters, curvature distribution, and attitude parameters output by the virtual pre-assembly and deviation calculation unit, integrates the data according to a preset format, generates a digital inspection report including various inspection indicators, and provides a visual output of the inspection results.

[0015] Compared with the prior art, the present invention has at least one of the following beneficial effects:

[0016] At the methodological level, this invention accurately obtains the component axis curvature and torsion angle parameters by acquiring the three-dimensional coordinates of feature points and constructing an initial geometric model, combined with a specialized curvature inversion model, replacing the traditional empirical parameter adjustment and solving the problem of mismatch between the model and the actual component posture.

[0017] By using an irregular component attitude calibration model, coordinate deviation analysis and iterative correction are employed to achieve accurate calibration of the component's spatial attitude, significantly improving the consistency between the model and the actual component.

[0018] An image marker matching and recognition algorithm is used to achieve efficient matching between virtual and actual marker points, establish a stable coordinate mapping relationship, avoid the drawbacks of single feature matching being easily interfered with, and ensure the accuracy of coordinate uniformity.

[0019] At the system level, through multi-unit modular design, the entire process of data acquisition, model building, attitude calibration, matching and positioning, pre-assembly calculation and report generation is automated. Combined with the deviation quantification and multi-parameter coupling analysis of the intelligent pre-assembly analysis platform for steel bridges, the gaps between splicing surfaces, misalignment, and axis alignment deviations are comprehensively and accurately identified, forming a complete detection index system. This completely changes the situation of one-sided and insufficient accuracy of deviation analysis in traditional detection, and provides more reliable and efficient technical support for the quality control of pre-assembly of large bending and torsion components. Attached Figure Description

[0020] Figure 1 This is a flowchart of the method steps of the present invention;

[0021] Figure 2 This is a diagram showing the system unit composition of the present invention. Detailed Implementation

[0022] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. The application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0023] like Figure 1 As shown, the digital inspection method for virtual pre-assembly of large bending and torsion components includes the following steps:

[0024] S1 acquires the three-dimensional coordinate data of discrete feature points on the surface of large bending and torsion components, constructs the initial geometric model of the component based on the spatial distribution of feature points, and extracts the component's axial direction vector, cross-sectional contour boundary point set, and relative distance parameters between feature points.

[0025] Specifically, step S1 is the foundational data acquisition and model building stage of the entire digital inspection process, and its implementation quality directly determines the accuracy of subsequent inspection results. During implementation, a high-precision 3D laser scanner is used to perform a full-area scan of the surface of large bending and tortuous components. 3D coordinate data of discrete feature points are collected at a density of no less than 150 feature points per square meter, with scanning accuracy controlled within 0.1 millimeters, ensuring data coverage of core areas such as the splicing surfaces at both ends of the component, abrupt changes in cross-section, and key control points along the axis. Based on the collected 3D coordinate data, point cloud processing and model reconstruction are performed using reverse modeling software to construct a 1:1 scale initial geometric model of the component. The model includes basic geometric information such as the component's total length, cross-sectional dimensions, and wall thickness distribution. Subsequently, key technical parameters are automatically extracted from the model. The axis direction vector is generated by fitting the coordinates of the center points at both ends of the component and the center point of the middle section. The boundary point set of the section contour is extracted with no less than 36 points per section. The relative distance parameters between feature points include the center distance between adjacent sections and the perpendicular distance between key feature points and the axis. These parameters provide basic data support for subsequent curvature calculation, attitude calibration and pre-assembly analysis, realizing the accurate transformation from physical components to digital models and laying the data foundation for full-process digital inspection.

[0026] S2 imports the initial geometric model into the intelligent pre-assembly analysis platform for steel bridges, calls the curvature inversion model of bending and torsion members to calculate the curvature distribution of the member axis, and obtains the curvature value and torsion angle parameters of each section of the axis;

[0027] Specifically, step S2, which involves the precise calculation of the component's axial curvature and torsion angle, is a crucial step in revealing the component's bending and torsional characteristics. During implementation, the initial geometric model constructed in S1 is imported into the intelligent pre-assembly analysis platform for steel bridges in a standard data format. The platform's built-in data interface automatically reads parameters such as the three-dimensional coordinates of feature points and the axial direction vector from the model. Subsequently, the curvature inversion model for bending and torsional components is invoked. Using the axial arc length as the calculation step size, the component's axial line is segmented at 0.5-meter intervals. The model analyzes and calculates the spatial coordinate changes at each node to obtain the curvature value and section torsion angle parameters corresponding to each node. The curvature value calculation results are retained to four decimal places, and the torsion angle parameters are accurate to 0.01 degrees. During the calculation process, the platform automatically filters out abnormal data points and uses a moving average method to smooth the calculation results, ensuring the continuity and reliability of the parameters. The axial curvature distribution data obtained through this step can accurately reflect the degree of bending and the change pattern of the component, while the torsion angle parameter clearly presents the torsion state of the component along the axis. These core parameters provide a quantitative basis for subsequent attitude calibration, solving the problem of difficulty in accurately obtaining the bending and torsional characteristic parameters of the component in traditional testing, and enabling virtual pre-assembly to be carried out based on the real mechanical properties of the component.

[0028] S3 uses an irregular component attitude calibration model to adjust the attitude of the initial geometric model. Through the three-dimensional coordinate deviation analysis of feature points, it corrects the spatial attitude parameters of the component to make the model attitude consistent with the actual component.

[0029] Specifically, step S3, through attitude calibration, ensures the spatial attitude of the virtual model matches that of the actual component, guaranteeing the accuracy of pre-assembly analysis. During implementation, the three-dimensional coordinates of at least 50 key feature points are extracted from the measured data collected in S1. These feature points are evenly distributed on the component surface, including edge points of the splicing surface, axis control points, and geometric abrupt changes. The coordinates of these feature points are compared one by one with the corresponding feature points in the initial geometric model, calculating the coordinate deviation values ​​of each feature point in the x, y, and z directions. These deviation values ​​are then input into the irregular component attitude calibration model. The model iteratively calculates the rotation angles of the component around the x, y, and z axes and the translation amounts in the three directions. The iteration count is set to 10-15 times until the sum of squared deviations between the model feature point coordinates and the measured coordinates reaches its minimum. Based on the calculated rotation angles (accurate to 0.001 degrees) and translation amounts (accurate to 0.01 millimeters), the spatial attitude of the initial geometric model is adjusted, updating the three-dimensional coordinates of all feature points in the model, forming a pre-calibrated geometric model. Afterwards, the model after preliminary calibration is verified for deviation. The average deviation of all key feature points is calculated. If the average deviation exceeds the preset threshold of 0.2 mm, the above comparison, calculation and adjustment process is repeated until the average deviation meets the detection accuracy requirements. This step ensures that the virtual model can truly restore the spatial posture of the actual component and eliminates the influence of the posture deviation between the model and the entity on the subsequent pre-assembly inspection.

[0030] S4 uses an image marker matching and recognition algorithm to match and locate preset marker points on the surface of the component, obtains the precise coordinates of the marker points in the virtual pre-assembly coordinate system, and establishes a coordinate mapping relationship between the virtual and the actual component.

[0031] Specifically, step S4, establishing a coordinate mapping between the virtual and actual components through marker point matching and positioning, is a crucial step in achieving precise association between the two. Before implementation, at least 40 marker points are pre-set on the surface of the actual component. These marker points are high-contrast, high-reflectivity circular markers with a diameter of 20-30 mm, evenly distributed at both ends and the middle area of ​​the component to avoid obstruction and facilitate image acquisition. During implementation, a high-definition industrial camera is used to acquire multi-angle images of the marker points on the component surface. The camera resolution is no less than 1920×1080 pixels, and the acquisition distance is controlled at 1-3 meters to ensure clear and identifiable marker point images. Image processing technology is used to extract the shape, grayscale, and spatial location features of each marker point, constructing a database containing all marker point feature information. Simultaneously, virtual marker points corresponding to the positions of the actual marker points are generated in the virtual pre-assembly model. An image marker matching and recognition algorithm is used to compare and calculate the feature information of the virtual and actual marker points, selecting marker point pairs with a similarity higher than 95% to establish a one-to-one correspondence. Based on the successfully matched marker pairs, a coordinate mapping matrix is ​​calculated using a coordinate transformation algorithm. This matrix is ​​then used to transform all the three-dimensional coordinate data of the actual components into the virtual pre-assembly coordinate system, achieving coordinate unification between the virtual and actual components. The coordinate transformation error is controlled within 0.15 mm, providing a coordinate reference for the accurate implementation of subsequent virtual pre-assembly.

[0032] Based on the calibrated geometric model and coordinate mapping relationship, S5 performs virtual pre-assembly operations in the intelligent pre-assembly analysis platform for steel bridges, and calculates the gap value, misalignment amount and axis alignment deviation parameters of the component splicing surface;

[0033] Specifically, step S5 is the core step in virtual pre-assembly implementation and deviation quantification calculation, directly outputting key data on component splicing compatibility. During implementation, the geometric model calibrated in S3 and the actual component data after coordinate unification in S4 are imported into the intelligent pre-assembly analysis platform for steel bridges. First, pre-assembly constraints are set, including allowable gap values ​​for splicing surfaces, misalignment control standards, and axis alignment deviation ranges. The allowable gap value for splicing surfaces is set to no more than 0.5 mm, the misalignment control standard is no more than 0.3 mm, and the axis alignment deviation range is no more than 0.2 degrees. Then, according to the pre-assembly sequence specified in the steel bridge construction design, virtual component splicing is performed on the platform. The platform calculates the gap value and misalignment of each splicing surface by real-time calculation of the spatial coordinate difference between corresponding points on each splicing surface. Simultaneously, by fitting the axial direction vectors of the components before and after splicing, the angular deviation of the axis at the splicing point is calculated. During pre-assembly, the platform monitors deviation data in real time. If the gap value, misalignment, or axis alignment deviation of a certain splicing surface exceeds the preset allowable range, the system automatically adjusts the spatial position of the component and recalculates the splicing, adjusting in increments of 0.05 mm until the deviation meets the requirements. After completing the virtual pre-assembly of all components, the platform summarizes the deviation parameters and axis alignment of each splicing surface, generating detailed data including the specific deviation values ​​and distribution locations of each splicing surface, providing a quantitative basis for subsequent test result analysis.

[0034] Based on the deviation parameters generated during the virtual pre-assembly process, S6 outputs the component pre-assembly adaptability test results, forming a digital test report including curvature distribution, attitude parameters, and splicing deviation.

[0035] Specifically, step S6 is the output and summary of test results, providing direct technical guidance for the pre-assembly construction of components. During implementation, the pre-assembly deviation details generated in S5 are first categorized and organized into three main types: curvature distribution, attitude parameters, and splicing deviations. Curvature distribution data includes the curvature values ​​and torsion angles of each calculated node along the axis; attitude parameters include the calibrated component rotation angles, translation amounts, and average deviations of feature points; and splicing deviations include the gap values, misalignment amounts, and axis alignment deviations of each splicing surface. Subsequently, compliance checks are performed on all types of data, and deviation parameters exceeding the allowable range and their corresponding locations are marked according to the steel bridge construction quality acceptance standards. Based on this, a digital test report is constructed according to a preset format. The report includes modules such as basic component information, test equipment parameters, test process descriptions, details of various technical parameters, and deviation analysis results. The technical parameter details are presented in tabular form, and the deviation analysis results are presented in a combination of textual descriptions and graphical displays, clearly and intuitively reflecting the pre-assembly compatibility of the components. After the inspection report is generated, it can be exported to common document formats and visualized on the intelligent pre-assembly analysis platform for steel bridges, facilitating technical personnel to view, analyze, and archive it. This step transforms the entire process of inspection data into practically valuable results, providing precise data support for deviation adjustments and process optimization during component pre-assembly construction, thus ensuring the quality of steel bridge construction.

[0036] Preferably, the expression for the curvature inversion model of the bending-torsional member is: ,in, Let be the combined curvature at a point on the axis of the bending / torsional member. For the three-dimensional coordinates of the feature point, For the axis arc length parameter, This is the curvature correction factor. The torsional angle of the component section. The second derivative in the x-direction of the axis. The rate of change of the torsion angle along the axis. They are respectively The first derivative of the direction along the axis.

[0037] Specifically, the curvature inversion model for bending and torsion components is a tool for accurately acquiring the axial bending and torsion characteristics of large bending and torsion components. Its implementation directly serves the curvature and torsion angle calculation in step S2. This model integrates key parameters such as the three-dimensional coordinates of component feature points, axial arc length, and cross-sectional torsion angle to construct a multi-dimensional coupled calculation logic. The correction coefficients introduced are pre-calibrated based on the component material, cross-sectional type, and manufacturing process to ensure the model is adaptable to bending and torsion components of different specifications. During implementation, the model first extracts the three-dimensional coordinate data of feature points obtained in step S1, divides the calculation nodes according to an axial arc length step of 0.5 meters, and comprehensively analyzes the second derivative of the coordinates, the rate of change of the torsion angle, and the change in the cross-sectional profile at each node. During the calculation process, the second derivative of the coordinates reflects the local change in the degree of axial curvature, the rate of change of the torsion angle along the axial direction captures the torsional trend of the component, and the square root of the sum of the squares of the directional derivatives of the cross-sectional profile quantifies the change in the spatial attitude of the cross-section. Finally, the correction coefficients are used to weight and compensate for errors in the calculation results of each sub-item, making the calculation results more consistent with the actual characteristics of the component. The application of this model solves the problem that traditional curvature calculation relies on only a single parameter and lacks accuracy. The calculated comprehensive curvature value is accurate to four decimal places, and the torsion angle parameter is accurate to 0.01 degrees. This provides a precise quantitative basis for bending and torsion characteristics for subsequent attitude calibration, ensuring that the virtual model can realistically reproduce the spatial bending and torsion state of the component, and laying a core foundation for the accuracy of the entire digital inspection process.

[0038] Preferably, the expression for the attitude calibration model of the irregular component is: ,in, For attitude calibration, Adjust the direction vector components for attitude control. This represents the deviation value of the feature point coordinates. For attitude calibration weighting coefficients, The components are wound around The rotational deviation angle of the shaft.

[0039] Specifically, the irregular component attitude calibration model is a key technology for achieving consistency between the virtual model and the actual component's spatial attitude, and is specifically applied to the attitude adjustment stage in step S3. This model uses feature point coordinate deviation, attitude adjustment direction vector, and rotation deviation angle as core input parameters. Weighting coefficients balance the influence of each parameter on attitude calibration, with these coefficients dynamically adjusted based on detection accuracy requirements and component complexity, typically set within a range of 0.3-0.7. During implementation, the model first receives the coordinate deviation data between the measured feature points and the initial model feature points from step S1, including deviations in the x, y, and z directions. Simultaneously, it acquires the component's rotation deviation angle around the three-dimensional coordinate axes, calculated from the vector angle between the measured coordinates and the model coordinates. The model performs vector operations on the coordinate deviation and attitude adjustment direction vector, combined with the comprehensive influence of the rotation deviation angle, to calculate the required attitude calibration amount for the component. This calibration amount includes both rotation angle and translation amount, with the rotation angle accurate to 0.001 degrees and the translation amount accurate to 0.01 millimeters. During the calculation process, the model continuously adjusts the weight coefficients through an iterative optimization algorithm until the sum of squared deviations of feature point coordinates reaches its minimum value, ensuring that the calibrated model's posture is highly consistent with the actual component. The application of this model effectively avoids the drawbacks of traditional posture adjustment relying on empirical judgment. Through quantitative calculation, it achieves precise correction of posture deviations, with the average calibration deviation controlled within 0.2 mm, eliminating interference caused by posture deviations for the subsequent accurate virtual pre-assembly.

[0040] Preferably, the expression for the image marker matching and recognition algorithm is: ,in, To match the similarity of the marker points, These are virtual and real marker sets, respectively. Let i be the feature vector of the i-th marker point. For matching coefficients, The number of markers, The average distance between virtual and actual marker points. These are the standard spacing parameters.

[0041] Specifically, the image marker matching and recognition algorithm is the core technology for establishing the coordinate mapping between virtual and actual components, and it is applied to the marker point matching and positioning stage in step S4. This algorithm constructs a multi-dimensional matching logic by fusing marker point feature vectors, matching quantity, and spacing parameters. The matching coefficient is pre-set according to the image acquisition environment and marker point type to ensure the algorithm's stability in different scenarios. During implementation, the algorithm first processes the actual marker point images acquired by a high-definition industrial camera, extracting the shape, grayscale, and spatial location features of each marker point to form a feature vector library. The feature vector dimension is set to 128 dimensions to ensure comprehensive representation of marker point characteristics. Simultaneously, feature vectors of corresponding virtual marker points are extracted from the virtual pre-stitching model. The similarity between the virtual and actual marker point feature vectors is quantified by calculating the sum of squared differences. The algorithm introduces the average spacing parameter of marker points as an auxiliary judgment criterion, filtering out false matches caused by image noise or occlusion by comparing the average spacing difference between virtual and actual marker points. During implementation, the algorithm sets a similarity threshold to select qualified marker point pairs and establish a one-to-one correspondence. The similarity threshold is set to 95% based on the detection accuracy requirements to ensure matching accuracy. The coordinate mapping matrix calculated by this algorithm can accurately transform the coordinates of actual components to the virtual pre-assembled coordinate system with the coordinate transformation error controlled within 0.15 mm. This solves the problems of traditional single feature matching being easily interfered with and lacking accuracy, and provides reliable technical support for the accurate association between virtual and actual components.

[0042] Preferably, the pre-assembly deviation calculation model of the intelligent pre-assembly analysis platform for steel bridges is as follows: ,in, For total pre-assembly deviation, This is the deviation weighting coefficient. For geometric model deviation, To correct attitude calibration deviation, This is due to coordinate mapping deviation. For the local deviation of the k-th splicing surface, This represents the number of spliced ​​surfaces.

[0043] Specifically, the pre-assembly deviation calculation model of the intelligent pre-assembly analysis platform for steel bridges is the core tool for quantifying virtual pre-assembly deviations and is applied to the deviation calculation stage in step S5. This model integrates multiple parameters, including geometric model deviation, attitude calibration deviation, coordinate mapping deviation, and local splicing deviation. It balances the influence of each deviation factor through weighting coefficients, which are determined based on the pre-assembly accuracy requirements and component splicing type. Geometric model deviation accounts for 30%-40% of the weight, attitude calibration and coordinate mapping deviations each account for 20%-25%, and local splicing deviation accounts for 10%-20%. During implementation, the model first extracts the attitude calibration deviation data output in step S3, including rotation and translation deviations; simultaneously, it acquires the coordinate mapping deviation data from step S4, calculated through statistical analysis of the difference between virtual and actual marker point coordinates; then, it combines this with the basic deviation between the geometric model constructed in step S1 and the actual component. The model obtains the comprehensive deviation component by performing square root operations on the sum of squares of various deviations, and simultaneously calculates the average value of all local deviations on the splicing surfaces as the local deviation component. The total pre-assembly deviation is obtained by weighting and summing the two types of components using weighting coefficients, with the total deviation calculation accurate to 0.01 mm. This model achieves comprehensive quantification of multi-dimensional deviations, overcoming the shortcomings of traditional deviation calculations that only focus on a single dimension and provide only partial analysis. It can comprehensively reflect key deviations such as splicing surface gaps, misalignment, and axis alignment, providing accurate quantitative basis for judging the pre-assembly suitability of components.

[0044] Preferably, the digital detection parameter coupling model for the virtual pre-assembly of the large bending and torsion component is as follows: ,in, To detect comprehensive parameters, The coupling coefficient is... For the overall curvature, For attitude calibration, To match similarity, For total pre-assembly deviation, The surface area of ​​the component. This represents the volume of the component.

[0045] Specifically, the virtual pre-assembly digital inspection parameter coupling model for large bending and torsion components is a core tool for integrating full-process inspection data and forming comprehensive inspection results. Its implementation runs through the entire inspection process and ultimately serves the generation of the inspection report in step S6. This model constructs a comprehensive inspection parameter evaluation system by coupling curvature distribution, attitude calibration, marker point matching similarity, total pre-assembly deviation, and component geometric parameters. The coupling coefficient is pre-calibrated according to the component's importance level and inspection standards to ensure the model highlights the impact of key inspection indicators. During implementation, the model first collects and calculates the comprehensive curvature value, outputs the attitude calibration, obtains the matching similarity, and calculates the total pre-assembly deviation. Simultaneously, it extracts the component's surface area, volume, and other geometric parameters obtained in step S1. The model multiplies the curvature, attitude calibration, and matching similarity to obtain the basic inspection parameter components. Then, it calculates the deviation influence component using the ratio of the total pre-assembly deviation to the component's surface area and volume. Finally, it weights and sums the two types of components using the coupling coefficient to obtain the comprehensive inspection parameters. This comprehensive parameter fully reflects the bending and torsional characteristics, posture accuracy, matching quality, and pre-assembly adaptability of the component. The numerical range is divided into three intervals according to the testing standards: qualified, needing adjustment, and unqualified. The application of this model solves the problem of scattered and difficult-to-comprehensive evaluation of traditional testing parameters. Through multi-parameter coupling, it achieves a comprehensive quantitative assessment of the pre-assembly quality of the component, provides core evaluation indicators for the test report, and makes the test results more authoritative and valuable.

[0046] Preferably, step S3 includes the following sub-steps: S31, extracting the three-dimensional coordinate data of the calibrated feature points in the initial geometric model, comparing them one by one with the measured coordinates of the corresponding feature points of the actual component, and calculating the coordinate deviation value of each feature point in the x, y, and z directions; S32, inputting the coordinate deviation value into the irregular component posture calibration model, and determining the optimal rotation angle and translation amount of the component around the x, y, and z axes through iterative calculation, so as to minimize the sum of squares of the deviations between the model feature point coordinates and the measured coordinates; S33, adjusting the spatial posture of the initial geometric model according to the calculated rotation angle and translation amount, updating the three-dimensional coordinates of all feature points in the model, and forming a pre-calibrated geometric model; S34, verifying the deviation of the pre-calibrated model. If the average deviation of the feature points exceeds a preset threshold, repeat steps S31 to S33 until the average deviation meets the detection accuracy requirements.

[0047] Specifically, the irregular component attitude calibration process in step S3 achieves precise attitude matching between the virtual model and the actual component through four sub-steps. Each sub-step is progressive and focuses on the core technical objective. S31 first selects no fewer than 50 key feature points from the initial geometric model constructed in step S1. These feature points cover the edges of the component's splicing surfaces, axis control points, and geometric abrupt changes, ensuring uniform distribution and representativeness. Then, the three-dimensional coordinate data of these feature points is extracted and compared one by one with the measured coordinates of the corresponding feature points on the actual component. The specific deviation values ​​of each feature point in the x, y, and z directions are calculated through coordinate differences, providing basic data for subsequent calibration. S32 inputs the coordinate deviation values ​​of all feature points into the irregular component attitude calibration model. The model starts an iterative calculation program, adjusting the rotation angles of the component around the x, y, and z axes and the translation amounts in the three directions. The objective function is to minimize the sum of squared feature point coordinate deviations. The number of iterations is set to 10-15 to ensure the stability and accuracy of the calculation results. S33 adjusts the spatial attitude of the initial geometric model based on the optimal rotation angle (accurate to 0.001 degrees) and translation amount (accurate to 0.01 mm) calculated in S32. It updates the 3D coordinates of all feature points in the model using a coordinate transformation algorithm, forming a pre-calibrated geometric model that approximates the actual component's attitude. S34 verifies the deviation of the pre-calibrated model by calculating the average deviation of all key feature points. If the average deviation exceeds a preset threshold of 0.2 mm, it returns to S31 to re-perform coordinate comparison, model calculation, and attitude adjustment until the average deviation meets the detection accuracy requirements. This step-by-step process completely eliminates the attitude deviation between the initial model and the actual component, providing a precise attitude reference for subsequent virtual pre-assembly.

[0048] Preferably, step S4 includes the following sub-steps: S41, image acquisition of preset marker points on the surface of the component, extraction of shape features, grayscale features and spatial location features of each marker point, and construction of a marker point feature database; S42, generation of virtual marker points corresponding to actual marker points in the virtual pre-assembly model, and calculation of feature similarity between virtual and actual marker points using an image marker matching and recognition algorithm; S43, screening of marker point pairs with similarity higher than a set threshold, establishing a one-to-one correspondence between virtual and actual marker points, and determining the coordinate mapping matrix; S44, based on the coordinate mapping matrix, conversion of the three-dimensional coordinate data of the actual component to the virtual pre-assembly coordinate system, completing the coordinate unification between the virtual and actual components.

[0049] Specifically, the image marker matching and positioning process in step S4 involves four sub-steps to construct a coordinate mapping relationship between the virtual and actual components, ensuring the accuracy and stability of coordinate consistency. S41 first involves acquiring images of at least 40 pre-set marker points on the surface of the actual component using a high-definition industrial camera with a resolution of at least 1920×1080 pixels. The acquisition distance is controlled between 1-3 meters to avoid image blurring or distortion. Subsequently, image processing techniques are used to extract the shape features, grayscale features, and spatial location features of each marker point, constructing a 128-dimensional marker point feature database to ensure comprehensive and unique feature information. S42, in the virtual pre-assembly model constructed in step S1, corresponding virtual marker points are generated according to the distribution of the actual marker points. An image marker matching and recognition algorithm is then called to compare and calculate the feature databases of the virtual and actual marker points. The similarity between the two is quantified through the difference in feature vectors, providing a basis for matching and selection. S43 sets a 95% similarity threshold, filters out marker point pairs with similarity higher than this threshold, eliminates false matches caused by image noise and occlusion, and establishes a one-to-one correspondence between virtual and actual marker points. Based on these successfully matched marker point pairs, a coordinate mapping matrix is ​​generated through matrix operations, clarifying the transformation rules between the two coordinate systems. S44 uses the coordinate mapping matrix obtained in S43 to transform the 3D coordinate data of all feature points of the actual component into the virtual pre-assembled coordinate system, completing the coordinate unification between the virtual and actual components. The coordinate transformation error is controlled within 0.15 mm. Through this step-by-step operation, a precise association between the virtual model and the actual component is achieved, providing a unified coordinate benchmark for subsequent deviation calculations.

[0050] Preferably, step S5 includes the following sub-steps: S51, importing the calibrated geometric model and the actual component data after coordinate unification into the intelligent pre-assembly analysis platform for steel bridges, setting pre-assembly constraints, including splicing surface fitting requirements, axis alignment standards, and allowable deviation ranges; S52, performing virtual splicing operations on the platform according to the preset pre-assembly sequence, calculating the gap value and misalignment of each splicing surface in real time, and recording the angular deviation of the axis at the splicing point; S53, monitoring the deviation data during the pre-assembly process in real time, and automatically adjusting the spatial position of the component and recalculating the splicing if the deviation of a certain splicing surface exceeds the allowable range; S54, after completing the virtual pre-assembly of all components, summarizing the deviation parameters and axis alignment of each splicing surface, and generating detailed pre-assembly deviation data.

[0051] Specifically, step S5, the virtual pre-assembly and deviation calculation process, includes four sub-steps to ensure the standardization of pre-assembly operations and the comprehensiveness of deviation quantification. S51 first imports the calibrated geometric model and the actual component data after coordinate unification into the intelligent pre-assembly analysis platform for steel bridges. Then, according to the steel bridge construction design requirements, pre-assembly constraints are set, specifying that the allowable gap between splicing surfaces is no greater than 0.5 mm, the misalignment control standard is no greater than 0.3 mm, and the axis alignment deviation range is no greater than 0.2 degrees. Simultaneously, the component pre-assembly sequence is determined, defining the execution standards for the pre-assembly operation. S52, following the pre-assembly sequence set in S51, the virtual splicing program is started in the platform. The platform calculates the spatial coordinate difference between corresponding feature points of each splicing surface in real time to obtain the specific gap value and misalignment of each splicing surface. Simultaneously, by fitting the axial direction vectors of the components before and after splicing, the angular deviation of the axis at the splicing point is calculated. All deviation data are recorded and stored in real time. S53 continuously monitors deviation data during pre-assembly. If the gap value, misalignment, or axis alignment deviation of a certain splicing surface exceeds the allowable range set in S51, the system automatically triggers an adjustment mechanism, changing the spatial position of the component in 0.05 mm increments and re-executing the splicing calculation until all deviation parameters meet the constraints. After completing the virtual pre-assembly of all components in S54, the platform summarizes the deviation parameters of each splicing surface, including the specific values ​​and distribution locations of the gap value, misalignment, and axis alignment deviation for each splicing surface, generating detailed pre-assembly deviation data. This provides comprehensive and accurate quantitative basis for generating the inspection report in step S6, ensuring that the inspection results truly reflect the pre-assembly adaptability of the components.

[0052] like Figure 2As shown, a digital inspection system for virtual pre-assembly of large bending and torsion components is described. This system, applied to the digital inspection method for virtual pre-assembly of large bending and torsion components, includes: a three-dimensional coordinate data acquisition unit, used to acquire three-dimensional coordinate data of surface feature points of the large bending and torsion components, establish a data transmission connection with the intelligent pre-assembly analysis platform for steel bridges, and transmit the acquired coordinate data to the platform in real time; a geometric model construction and curvature inversion unit, which receives the coordinate data output by the three-dimensional coordinate data acquisition unit, constructs the initial geometric model of the component, calls the curvature inversion model of the bending and torsion component to calculate the axial curvature distribution, and sends the model and curvature parameters to the attitude calibration unit; and an irregular component attitude calibration unit, which receives the model data transmitted by the geometric model construction and curvature inversion unit, and corrects the component's attitude using the irregular component attitude calibration model. The spatial attitude measurement unit transmits the calibrated model to the marker matching and positioning unit. The image marker matching and positioning unit collects image information of marker points on the component surface, uses an image marker matching and recognition algorithm to match and locate virtual and actual marker points, establishes a coordinate mapping relationship, and sends it to the virtual pre-assembly unit. The virtual pre-assembly and deviation calculation unit receives the calibrated model, coordinate mapping relationship, and platform preset parameters, performs virtual pre-assembly operations, calculates splicing deviation parameters, and transmits the deviation data to the inspection report generation unit. The digital inspection report generation unit receives the deviation parameters, curvature distribution, and attitude parameters output by the virtual pre-assembly and deviation calculation unit, integrates the data according to a preset format, generates a digital inspection report including various inspection indicators, and provides a visual output of the inspection results.

[0053] A digital inspection method and system for virtual pre-assembly of large bending and torsion components is proposed. This method constructs an initial geometric model by collecting the 3D coordinates of feature points on the component surface. A specialized curvature inversion model accurately obtains the axis curvature and torsion angle parameters, replacing traditional empirical parameter estimation. Simultaneously, an irregular component posture calibration model is used to achieve precise matching between the model and the actual component posture through coordinate deviation analysis and iterative correction. An image marker matching and recognition algorithm improves the stability of virtual and actual marker point positioning through multi-feature fusion matching. Combined with a steel bridge intelligent pre-assembly analysis platform, deviation quantification calculation is completed, forming a complete inspection index covering curvature, posture, and splicing deviation, significantly improving the reliability of the inspection data. At the system level, the modular design achieves fully automated integration of data acquisition, model construction, posture calibration, matching and positioning, pre-assembly calculation, and report generation, avoiding the problems of excessive manual intervention and fragmented processes in traditional inspection, and significantly improving inspection efficiency.

[0054] This method and system, through the synergistic application of a specialized curvature inversion model and attitude calibration model, iterates and corrects based on measured coordinate deviations to ensure that the model can accurately reflect the spatial morphology and curvature distribution of the components. Addressing the issues of easily interfered marker point matching, low coordinate uniformity accuracy, and one-sided deviation analysis, a multi-feature fusion marker point matching algorithm is employed to improve positioning stability and establish precise coordinate mapping relationships. Simultaneously, a multi-parameter coupled analysis of splicing surface gaps, misalignment, and axis alignment deviations is achieved through a steel bridge intelligent pre-assembly analysis platform, forming a comprehensive deviation quantification system. This completely solves the shortcomings of insufficient accuracy and poor reliability of traditional detection methods, providing a systematic technical solution for the quality control of pre-assembly of large bending and torsion components.

[0055] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," "link," and "fix" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal communication between two components. Those skilled in the art will understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0056] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various equivalent changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A digital inspection method for virtual pre-assembly of large bending and torsion components, characterized in that, Includes the following steps: S1. Obtain the three-dimensional coordinate data of discrete feature points on the surface of large bending and torsion components. Construct the initial geometric model of the component based on the spatial distribution of feature points, and extract the component axis direction vector, cross-sectional contour boundary point set and relative distance parameters between feature points. S2. Import the initial geometric model into the intelligent pre-assembly analysis platform for steel bridges, call the curvature inversion model of bending and torsion members to calculate the curvature distribution of the member axis, and obtain the curvature value and torsion angle parameter of each section of the axis; S3, the attitude of the initial geometric model is adjusted by using the attitude calibration model of the irregular component. The spatial attitude parameters of the component are corrected by the three-dimensional coordinate deviation analysis of the feature points, so that the attitude of the model is consistent with that of the actual component. S4. Use an image marker matching and recognition algorithm to match and locate the preset marker points on the surface of the component, obtain the precise coordinates of the marker points in the virtual pre-assembly coordinate system, and establish the coordinate mapping relationship between the virtual and the actual component. S5, based on the calibrated geometric model and coordinate mapping relationship, performs virtual pre-assembly operation in the steel bridge intelligent pre-assembly analysis platform to calculate the gap value, misalignment amount and axis alignment deviation parameter of the component splicing surface; S6 outputs the component pre-assembly adaptability test results based on the deviation parameters generated during the virtual pre-assembly process, forming a digital test report including curvature distribution, attitude parameters, and splicing deviation.

2. The digital inspection method for virtual pre-assembly of large bending and torsion components according to claim 1, characterized in that, The expression for the curvature inversion model of the bending and torsional member is: ， in, Let be the combined curvature at a point on the axis of the bending / torsional member. For the three-dimensional coordinates of the feature point, For the axis arc length parameter, This is the curvature correction factor. The torsional angle of the component section. The second derivative in the x-direction of the axis. The rate of change of the torsion angle along the axis. They are respectively The first derivative of the direction along the axis.

3. The digital inspection method for virtual pre-assembly of large bending and torsion components according to claim 1, characterized in that, The expression for the attitude calibration model of the irregular component is: ， in, For attitude calibration, Adjust the direction vector components for attitude control. This represents the deviation value of the feature point coordinates. For attitude calibration weighting coefficients, The components are wound around Rotational deviation angle of the shaft.

4. The digital inspection method for virtual pre-assembly of large bending and torsion components according to claim 1, characterized in that, The expression for the image marker matching and recognition algorithm is: ， in, For marker point matching similarity, These are virtual and real marker sets, respectively. Let i be the feature vector of the i-th marker point. For matching coefficients, The number of markers, The average distance between virtual and actual marker points. These are the standard spacing parameters.

5. The digital inspection method for virtual pre-assembly of large bending and torsion components according to claim 1, characterized in that, The pre-assembly deviation calculation model of the intelligent pre-assembly analysis platform for steel bridges is as follows: ， in, For total pre-assembly deviation, This is the deviation weighting coefficient. For geometric model deviation, To calibrate attitude deviation, This is due to coordinate mapping deviation. For the local deviation of the k-th splicing surface, This represents the number of spliced ​​surfaces.

6. The digital inspection method for virtual pre-assembly of large bending and torsion components according to claim 1, characterized in that, The digital detection parameter coupling model for the virtual pre-assembly of the large bending and torsion component is as follows: ， in, To detect comprehensive parameters, The coupling coefficient is... For the overall curvature, For attitude calibration, To match similarity, For total pre-assembly deviation, For the surface area of ​​the component, The volume of the component.

7. The digital inspection method for virtual pre-assembly of large bending and torsion components according to claim 1, characterized in that, S3 includes the following steps: S31, extract the three-dimensional coordinate data of the calibrated feature points in the initial geometric model, compare them one by one with the measured coordinates of the corresponding feature points of the actual component, and calculate the coordinate deviation value of each feature point in the x, y, and z directions. S32, input the coordinate deviation value into the irregular component attitude calibration model, and determine the optimal rotation angle and translation amount of the component around the x, y, and z axes through iterative calculation, so as to minimize the sum of squares of the deviations between the model feature point coordinates and the measured coordinates; S33, Based on the calculated rotation angle and translation amount, adjust the spatial attitude of the initial geometric model, update the three-dimensional coordinates of all feature points in the model, and form a preliminary calibrated geometric model; S34. Verify the deviation of the model after preliminary calibration. If the average deviation of feature points exceeds the preset threshold, repeat S31 to S33 until the average deviation meets the detection accuracy requirements.

8. The digital inspection method for virtual pre-assembly of large bending and torsion components according to claim 1, characterized in that, S4 includes the following steps: S41, acquire images of preset marker points on the surface of the component, extract the shape features, grayscale features and spatial position features of each marker point, and construct a marker point feature database; S42, In the virtual pre-scraping model, virtual markers corresponding to the actual markers are generated, and the feature similarity between the virtual and actual markers is calculated using an image marker matching and recognition algorithm; S43, filter out marker point pairs with similarity higher than a set threshold, establish a one-to-one correspondence between virtual and actual marker points, and determine the coordinate mapping matrix; S44, based on the coordinate mapping matrix, transforms the three-dimensional coordinate data of the actual component into the virtual pre-assembly coordinate system, thus achieving coordinate unification between the virtual and actual components.

9. The digital inspection method for virtual pre-assembly of large bending and torsion components according to claim 1, characterized in that, S5 includes the following steps: S51. Import the calibrated geometric model and the actual component data after coordinate unification into the intelligent pre-assembly analysis platform for steel bridges, and set pre-assembly constraints, including splicing surface fitting requirements, axis alignment standards and allowable deviation ranges. S52, according to the preset pre-assembly sequence, performs virtual splicing operation of components in the platform, calculates the gap value and misalignment of each splicing surface in real time, and records the angular deviation of the axis at the splicing point; S53 monitors deviation data in real time during the pre-assembly process. If the deviation of a certain splicing surface exceeds the allowable range, the spatial position of the component is automatically adjusted and the splicing calculation is recalculated. S54. After completing the virtual pre-assembly of all components, summarize the deviation parameters and axis alignment of each splicing surface, and generate detailed pre-assembly deviation data.

10. A digital inspection system for virtual pre-assembly of large bending and torsion components, characterized in that, This system is applied to the digital inspection method for virtual pre-assembly of large bending and torsion components as described in claim 1, comprising: The three-dimensional coordinate data acquisition unit is used to acquire the three-dimensional coordinate data of the surface feature points of large bending and torsion components, establish a data transmission connection with the intelligent pre-assembly analysis platform for steel bridges, and transmit the acquired coordinate data to the platform in real time. The geometric model construction and curvature inversion unit receives coordinate data output from the three-dimensional coordinate data acquisition unit, constructs the initial geometric model of the component, calls the curvature inversion model of the bending and torsion component to calculate the axis curvature distribution, and sends the model and curvature parameters to the attitude calibration unit. The irregular component attitude calibration unit receives model data transmitted by the geometric model construction and curvature inversion unit, corrects the spatial attitude of the component through the irregular component attitude calibration model, and transmits the calibrated model to the marker point matching and positioning unit. The image marker matching and positioning unit collects image information of marker points on the surface of the component, uses an image marker matching and recognition algorithm to match and locate virtual and actual marker points, establishes a coordinate mapping relationship, and sends it to the virtual pre-assembly unit. The virtual pre-assembly and deviation calculation unit receives the model after attitude calibration, coordinate mapping relationship and platform preset parameters, performs virtual pre-assembly operation, calculates splicing deviation parameters, and transmits the deviation data to the test report generation unit; The digital test report generation unit receives the deviation parameters, curvature distribution, and attitude parameters output by the virtual pre-assembly and deviation calculation unit, integrates the data according to a preset format, generates a digital test report including various test indicators, and outputs the test results in a visual format.

Images