Intelligent detection and deviation correction system for steel structure construction

Abstract

The invention relates to the technical field of solid deformation measurement, particularly provides an intelligent detection and deviation correction system for steel structure construction, and solves the problems that in the prior art, solid deformation measurement means are limited, deviation can be found only through static measurement or post analysis, and the measurement accuracy is poor. And real-time monitoring and dynamic correction in the construction process cannot be realized. The system comprises a virtual model module integrated with various sensors and used for collecting steel structure component data in real time; the steel deviation prediction module constructs a steel deviation prediction model and trains the steel deviation prediction model according to the analysis result of the construction physical model; inputting the construction physical model into the steel deviation prediction model, and predicting a deviation trend; a correction suggestion generation module generates a correction suggestion according to the predicted deviation trend; and triggering an automatic adjustment instruction based on the correction suggestion. Through organic combination of the virtual model, deviation prediction and correction suggestions, intelligent monitoring and deviation correction of the steel structure construction process are realized.

Description

Technical Field

[0001] The present invention relates to the technical field of measuring the deformation of solids, and particularly to an intelligent detection and deviation correction system for steel structure construction. Background Art

[0002] Steel structure construction is widely used in modern architecture. Due to its high strength, light weight, and good seismic performance, it has become the preferred structural form for large buildings, bridges, industrial plants, and other projects. However, during the steel structure construction process, due to reasons such as complex component dimensions, variable construction environments, and human operation errors, problems such as geometric dimension deviations, weight deviations, and stress deformations are likely to occur. In particular, the measurement and control of solid deformations (such as elastic deformation, plastic deformation, or thermal deformation caused by temperature of steel) are one of the difficulties in construction quality. These deformations will directly affect the installation accuracy and overall stability of the structure. Traditional steel structure construction inspections mainly rely on manual measurement and empirical judgment, which have the defects of low detection efficiency, insufficient accuracy, and poor real-time performance. In addition, the existing technologies have limited means for measuring solid deformations. Often, deviations can only be discovered through static measurement or post-event analysis, and real-time monitoring and dynamic correction during the construction process cannot be achieved. These problems not only increase construction costs and construction periods but may also pose potential structural safety hazards. Summary of the Invention

[0003] To achieve the above object, the present invention adopts the following technical solutions: On the one hand, the present invention provides an intelligent detection and deviation correction system for steel structure construction, comprising: A virtual model module, which is used to integrate a variety of sensors to collect data such as geometric dimensions, weight deviations, and stress deformations of steel structure components in real time; construct a construction physical model based on the physical construction scenario, and compare the point cloud data with the construction physical model. If the analysis result is greater than a preset value, it is output to the steel deviation prediction module; A steel deviation prediction module, which is used to construct a steel deviation prediction model based on historical steel structure data and real-time steel structure data, and train the steel deviation prediction model according to the analysis result of the construction physical model; input the construction physical model into the steel deviation prediction model to predict the deviation trend; A correction suggestion generation module, which is used to generate correction suggestions according to the predicted deviation trend; trigger an automatic adjustment instruction based on the correction suggestions; if the deviation exceeds a preset threshold, issue a warning and locate the deviation problem area.

[0004] In an optional implementation manner, the virtual model module includes: A coordinate alignment sub-module, which is used to align the measured point cloud data with the construction physical model in the coordinate system, select the positions of the common steel structure nodes between the two, and convert the measured point cloud from the scanning coordinate system to the design coordinate system of the construction physical model; perform spatial positioning by generating point cloud lines; A construction attribute comparison sub-module, which is used to compare the geometric differences between the point cloud and the construction virtual model, calculate the deviation values between the actual cloud points and the construction physical model point by point; perform semantic segmentation on the point cloud, identify the basic components, and perform objectified comparison with the corresponding construction attributes in the construction physical model; A dataset generation sub-module, which is used to fit the point cloud data into the elements of the construction physical model for the identified key constructions and compare with the design values; generate a virtual point cloud dataset from the parameters in the construction physical model, and divide the virtual point cloud dataset into a training set, a validation set, and a test set.

[0005] In an optional implementation, the construction attribute comparison sub-module includes: A point cloud segmentation unit, which is used to perform hierarchical feature extraction on the point cloud, learn geometric features point by point; adaptively construct a local neighborhood graph, and enhance the feature expression of the steel beam nodes through the graph attention mechanism; perform plane segmentation on the point cloud to separate the building floor plan and the component point cloud; An independent cluster unit, which is used to divide the point cloud into independent clusters, each cluster corresponding to a basic mechanism, and perform refined segmentation in combination with silver information; for multi-source data, map the 2D semantic segmentation result to the 3D point cloud to supplement semantic information; A deviation calculation unit, which is used to perform pairwise comparison between the segmented component point cloud and the construction attributes in the construction physical model through feature matching to calculate the geometric deviation; if abnormal segmentation occurs, trigger a feedback optimization mechanism to adjust the clustering threshold.

[0006] In an optional implementation, the steel deviation prediction module includes: A variable distribution difference sub-module, which is used to construct a steel deviation prediction model based on historical steel structure data and real-time steel structure data, merge the historical steel structure data and the real-time steel structure data to obtain a dataset, and divide it into a training set and a test set; introduce a control dataset to compare the variable distribution deviation between the original training set and the control dataset; An optimal model sub-module, which is used to train multiple base models with historical steel structure data to generate a new training set and a test set; train a meta-model with the new dataset generated by the base models, and optimize the weights through backpropagation; determine the optimal steel deviation prediction model through cross-validation; A prediction deviation sub-module, which is used to adjust the weights if the variable distribution shift is shown in the control dataset test, and re-train the steel deviation prediction model; input the construction physical model into the steel deviation prediction model to predict the deviation trend;

[0007] In an alternative embodiment, the variable distribution difference sub-module includes: A deviation program calculation unit for calculating the distribution difference of each variable in two data sets, generating a deviation probability value; reflecting the deviation degree of the variable in the control data set from that in the original data set by comparing the deviation probability values; checking whether the frequencies of various categories in the control data set are consistent with those in the original data set; A target variable adjustment unit for readjusting the target variable when the value distribution of the target variable in the control data set differs from that in the original training set by more than a preset standard value; A deviation degree judgment unit for marking the target variables with deviation probabilities greater than a preset value, if the control data set with a probability of 1 completely lacks the variable value; visualizing the distribution overlap of the target variable in the original data set and the control data set to judge the deviation degree; generating an automated deviation report.

[0008] In an alternative embodiment, the optimal model sub-module includes: A prediction result generation unit for sorting historical steel structure data according to features to form an initial training set; generating multiple base models based on the historical steel structure data, and using the initial training set to train the multiple independent base models respectively to generate corresponding prediction results, where the prediction results are the capture of the trends or patterns of steel material deviations in the historical steel structure data; A feature combination processing unit for combining the prediction results of each base model with the features of the initial training set to form an extended feature space; if each base model generates a prediction value, the number of features of the initial training set increases by the same number as the number of base models; A normalization processing unit for applying the prediction results of the base models to the test set to generate a new test set; performing normalization processing on the newly generated training set and test set to eliminate the dimensional differences between different features.

[0009] In an alternative embodiment, the prediction result generation unit includes: A parameter combination design sub-unit for designing multiple unique parameter combinations according to the multi-dimensional features of the initial training set, and designing a base model for capturing dynamic change trends to identify the trends or patterns of steel material deviations; A data deconstruction processing sub-unit for deconstructing the historical steel structure data from multiple representation perspectives according to the different representations of short-term deviation, long-term deviation, local deviation and global deviation of steel material deviations, obtaining the recognition results corresponding to short-term deviation, long-term deviation, local deviation and global deviation; integrating the recognition results into a comprehensive deviation prediction value using weighted average; An index adding subunit is used to add a confidence index to the deviation prediction value and evaluate the reliability and uncertainty of the prediction result using Monte Carlo simulation.

[0010] In an alternative implementation, the prediction deviation sub-module includes: An adjustment mechanism triggering unit is used to calculate the deviation probability, category deviation, and target variable deviation by comparing the variable distributions of the control data set and the original training set. If a distribution deviation greater than the preset value is detected, the weight adjustment mechanism is triggered; after the base model generates a secondary data set, the meta-model optimizes the value through backpropagation; A predicted value generating unit is used to retrain the steel deviation prediction model based on the updated weights; input the physical model parameters of the current construction into the adjusted steel deviation prediction model to generate a predicted value of the future steel deviation; A deviation judgment unit restores the standardized predicted value to the actual physical dimension; restores the long-term change law of the data by fitting the trend term; calculates the absolute deviation and relative deviation between the predicted value and the actual measured value to identify abnormal points; if the deviation exceeds the threshold, an early warning is issued.

[0011] In an alternative implementation, the correction suggestion generation module includes: A first trend analysis module is used to analyze the prediction deviation trend to obtain whether the prediction deviation shows short-term local expansion or long-term global accumulation; for short-term local expansion, a local correction plan; for long-term global accumulation, a comprehensive adjustment; A second trend analysis module is used to obtain geometric dimension deviation, stress deformation deviation, or weight deviation from the prediction deviation trend; An execution instruction conversion module is used to convert the geometric dimension deviation, stress deformation deviation, or weight deviation into executable instructions, link with construction equipment, and automatically adjust welding parameters, displacement member positions, or apply auxiliary supports according to the correction suggestions; detect the welded seams by ultrasonic waves.

[0012] In an alternative implementation, for geometric dimension deviation such as length, width, or height deviation, an adjustment plan is generated through a parametric design tool, such as suggesting local cutting and welding or overall component replacement; For stress deformation deviation, optimize the support structure or adjust the external load distribution; perform mechanical property tests on bolts, nuts, and washers in the support structure; For weight deviation, provide a counterweight adjustment or material optimization distribution plan.

[0013] The virtual model module of the present invention collects key data such as geometric dimensions, weight deviations, and stress deformations of steel structure components in real time by integrating multiple sensors, and constructs a construction physical model in combination with the physical construction scenario; through the comparative analysis of point cloud data and the construction physical model, deviations during the construction process can be quickly identified; if the deviation exceeds the preset value, the result will be output to the steel deviation prediction module. The steel deviation prediction module constructs a steel deviation prediction model based on historical steel structure data and real-time collected steel structure data, and conducts model training through the analysis results of the construction physical model; by inputting the construction physical model into the prediction model, the deviation trend of steel structure components can be effectively predicted, providing early warnings for potential problems during the construction process, thereby reducing the impact of deviations on construction quality. The correction suggestion generation module automatically generates targeted correction suggestions based on the predicted deviation trend and triggers corresponding automatic adjustment instructions; if the deviation exceeds the preset threshold, a warning signal will be issued, and the deviation problem area will be accurately located, facilitating construction personnel to take corrective measures in a timely manner to ensure that the construction quality meets the requirements. Description of the Drawings

[0014] The drawings are used to provide a further understanding of the present invention and constitute a part of the specification. They are used together with the embodiments of the present invention to explain the present invention and do not constitute a limitation to the present invention. In the drawings: Figure 1 It is a block diagram of an intelligent detection and deviation correction system for steel structure construction provided in Embodiment 1 of the present invention; Figure 2 It is a block diagram of the virtual model module provided in Embodiment 2 of the present invention; Figure 3 It is a block diagram of the steel deviation prediction module provided in Embodiment 3 of the present invention; Figure 4 It is a block diagram of the correction suggestion generation module provided in Embodiment 4 of the present invention; Figure 5 It is a block diagram of the electronic device provided by the present invention; Figure 6 It is a block diagram of the computer-readable storage medium provided by the present invention. Detailed Embodiments

[0015] Next, the technical solutions in the embodiments of the present invention will be described in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments.

[0016] Hereinafter, terms such as "first" and "second" are only used for convenience of description and should not be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, features defined with "first", "second", etc. may explicitly or implicitly include one or more of such features. In the description of the present invention, unless otherwise specified, the meaning of "a plurality" is two or more.

[0017] In the present invention, unless otherwise clearly specified and defined, the term "connection" should be understood in a broad sense. For example, "connection" can be a fixed mechanical connection, a detachable mechanical connection, or integrated; or, "connection" can be a direct connection or an indirect connection through an intermediate medium. In addition, unless otherwise clearly specified and defined, the term "coupling" should be understood in a broad sense. For example, "coupling" can be a direct electrical connection. For example, physical contact and electrical conduction occur between two components, and it can also be understood that different components in a circuit structure are electrically connected through physical lines such as copper foils or wires of a printed circuit board (PCB) that can transmit electrical signals for electrical signal transmission; or, "coupling" can be an indirect electrical connection between two components through an intermediate medium; or, "coupling" can be an electrical connection between two components in an air-spaced / non-contact manner. For example, two components are electrically connected by means of capacitive coupling for electrical signal transmission.

[0018] In the embodiments of the present invention, orientation terms such as "upper", "lower", "left", "right", etc. can include, but are not limited to, being defined relative to the schematic placement of components in the drawings. It should be understood that these directional terms can be relative concepts, which are used for relative description and clarification, and they can change accordingly with the change of the orientation of the components in the drawings.

[0019] Embodiment 1: As Figure 1 shown, the embodiments of the present invention provide an intelligent detection and deviation correction system for steel structure construction, including: A virtual model module, which is used to integrate a variety of sensors to collect data such as geometric dimensions, weight deviations, and stress deformations of steel structure components in real time; construct a construction physical model based on the physical construction scenario, and compare the point cloud data with the construction physical model. If the analysis result is greater than the preset value, it is output to the steel deviation prediction module; A steel deviation prediction module, which is used to construct a steel deviation prediction model based on historical steel structure data and real-time steel structure data, and train the steel deviation prediction model according to the analysis result of the construction physical model; input the construction physical model into the steel deviation prediction model to predict the deviation trend; A correction suggestion generation module is used to generate correction suggestions according to the predicted deviation trend, trigger an automatic adjustment instruction based on the correction suggestions, and issue a warning and locate the deviation problem area when the deviation exceeds a preset threshold.

[0020] In the above embodiment, in this embodiment, the virtual model module collects key data such as geometric dimensions, weight deviation, and stress deformation of steel structure components in real time by integrating multiple sensors, and constructs a construction physical model in combination with the physical construction scenario. Through the comparative analysis of point cloud data and the construction physical model, the deviation in the construction process can be quickly identified. If the deviation exceeds the preset value, the result will be output to the steel deviation prediction module. The steel deviation prediction module constructs a steel deviation prediction model based on historical steel structure data and real-time collected steel structure data, and conducts model training through the analysis results of the construction physical model. By inputting the construction physical model into the prediction model, the deviation trend of steel structure components can be effectively predicted, providing early warnings for potential problems in the construction process, thereby reducing the impact of deviations on construction quality. The correction suggestion generation module automatically generates targeted correction suggestions according to the predicted deviation trend and triggers corresponding automatic adjustment instructions. If the deviation exceeds the preset threshold, a warning signal will be issued, and the deviation problem area will be accurately located, facilitating construction personnel to take corrective measures in a timely manner to ensure that the construction quality meets the requirements.

[0021] In summary, this embodiment realizes the intelligent monitoring and deviation correction of the steel structure construction process through the organic combination of virtual models, deviation prediction, and correction suggestions. Its core value lies in real-time, accuracy, and automation: through the combination of sensors and models, deviations in the construction process can be detected in a timely manner; through the prediction model, the deviation trend can be predicted in advance; through the correction suggestion module, solutions can be automatically generated and adjustment instructions can be triggered, significantly improving construction efficiency and quality, and reducing construction risks and costs.

[0022] Embodiment 2: As Figure 2 shown, on the basis of Embodiment 1, the virtual model module provided by the embodiment of the present invention specifically includes: A coordinate alignment sub-module is used to align the measured point cloud data with the construction physical model in the coordinate system, select the positions of common steel structure nodes between the two, and convert the measured point cloud from the scanning coordinate system to the design coordinate system of the construction physical model; perform spatial positioning by generating point cloud lines; A construction attribute comparison sub-module is used to compare the geometric differences between the point cloud and the construction virtual model, calculate the deviation values of the actual cloud points and the construction physical model point by point, perform semantic segmentation on the point cloud to identify basic components, and perform object-based comparison with the corresponding construction attributes in the construction physical model; Among them, the construction attributes include design dimensions and material parameters, etc.; A dataset generation sub-module, which is used to fit the elements of the construction physical model with point cloud data for the identified key constructions and compare them with the design values; generate a virtual point cloud dataset from the parameters in the construction physical model, and divide the virtual point cloud dataset into a training set, a validation set, and a test set.

[0023] In the above embodiment, in this embodiment, the coordinate alignment sub-module unifies the coordinate systems of the measured point cloud data and the construction physical model to ensure that the data is compared and analyzed in the same coordinate system; by identifying common steel structure nodes, the point cloud is converted from the scanning coordinate system to the design coordinate system, thereby achieving accurate spatial positioning; coordinate alignment is the basis for subsequent analysis, ensuring the accuracy and reliability of geometric and attribute comparisons. The construction attribute comparison sub-module is mainly used for the geometric difference analysis between the point cloud data and the construction virtual model. By calculating the deviation value between the cloud point and the model point by point, the difference between the actual construction and the design is quantified; at the same time, semantic segmentation technology is used to identify the basic components in the point cloud and make an object-based comparison with the construction attributes in the construction physical model; the construction attributes include design dimensions, material parameters, etc., and the comparison results provide data support for the construction quality assessment. The dataset generation sub-module fits the elements of the construction physical model with the point cloud data for the identified key constructions and compares them with the design values; at the same time, generates a virtual point cloud dataset from the parameters in the construction physical model and divides it into a training set, a test set, and a test set, providing a standardized data basis for model training, algorithm verification, and quality control.

[0024] In summary, this embodiment realizes the efficient comparison and fusion of the construction physical model and the measured point cloud data through coordinate alignment, geometric difference analysis, and data generation; provides standardized and data-based technical support for construction precision detection, quality assessment, and optimization of the construction process, and is an important part of modern building information management.

[0025] Embodiment 3: Based on Embodiment 2, the construction attribute comparison sub-module provided by the embodiment of the present invention specifically includes: A point cloud segmentation unit, which is used to extract hierarchical features from the point cloud, learn geometric features point by point; adaptively construct a local neighborhood graph to enhance the feature expression of the steel beam nodes through the graph attention mechanism; perform plane segmentation on the point cloud to separate the building floor plan and the component point cloud; An independent cluster unit, which is used to divide the point cloud into independent clusters, each cluster corresponding to a basic mechanism, and perform refined segmentation in combination with silver information; for multi-source data, map the 2D semantic segmentation result to the 3D point cloud to supplement semantic information; A deviation calculation unit, which is used to compare the segmented construction point cloud with the construction attributes in the construction physical model object by object through feature matching and calculate the geometric deviation; if an abnormal segmentation occurs, trigger a feedback optimization mechanism to adjust the clustering threshold.

[0026] In the above embodiments, in this embodiment, the point cloud segmentation unit performs structured processing on the point cloud through hierarchical feature extraction and geometric feature learning; adaptively constructs a local neighborhood graph and introduces a graph attention mechanism to enhance the feature expression of key parts such as steel beam nodes; at the same time, separates the building floor plan and component point cloud to provide a basis for clustering and comparative analysis. The independent cluster unit divides the point cloud into independent clusters, and each cluster corresponds to a basic component; refines the clusters by combining shape information (such as silver information) to ensure the accuracy of component recognition; for multi-source data, this unit maps the 2D semantic segmentation result to the 3D point cloud to supplement semantic information and further improve the segmentation accuracy. The deviation calculation unit compares the segmented building point cloud with the building attributes in the construction physical model object by object through feature matching technology to calculate the geometric deviation; in addition, for abnormal segmentation situations, this unit triggers a feedback optimization mechanism to dynamically adjust the clustering threshold to ensure the reliability of the segmentation result.

[0027] In summary, the construction attribute comparison sub-module of this embodiment realizes the refined comparative analysis of the construction physical model and the measured point cloud data through point cloud segmentation, clustering, and deviation calculation; provides a standardized and automated technical means for construction quality inspection, geometric accuracy evaluation, and anomaly recognition, and is an important support for improving the efficiency and accuracy of construction management; helps to timely discover construction errors, optimize the construction process, and ensure that the project quality meets the design requirements.

[0028] Embodiment 4: As Figure 3 shown, on the basis of Embodiment 1, the steel deviation prediction module provided by the embodiment of the present invention specifically includes: The variable distribution difference sub-module is used to construct a steel deviation prediction model based on historical steel structure data and real-time steel structure data, merge the historical steel structure data and real-time steel structure data to obtain a data set, and divide it into a training set and a test set; introduce a control data set to compare the variable distribution deviation between the original training set and the control data set; The optimal model sub-module is used to train multiple base models with historical steel structure data to generate a new training set and a test set; train a meta-model with the new data set generated by the base model, and optimize the weights through backpropagation; determine the optimal steel deviation prediction model through cross-validation; The prediction deviation sub-module is used to adjust the weights and retrain the steel deviation prediction model if the variable distribution shows an offset during the test of the control data set; input the construction physical model into the steel deviation prediction model to predict the deviation trend.

[0029] In the above embodiments, in this embodiment, the variable distribution difference sub-module constructs a steel deviation prediction model based on historical steel structure data and real-time steel structure data; generates a data set by combining historical and real-time data, and divides it into a training set and a test set; introduces a control data set, compares the variable distribution deviation between the original training set and the control data set, and provides data support for model optimization. The optimal model sub-module trains multiple base models using historical steel structure data, generates a new training set and a test set; trains the new data set through a meta-model, and optimizes the model weights in combination with the backpropagation algorithm; determines the optimal steel deviation prediction model through cross-validation to improve the prediction accuracy. When the prediction deviation sub-module shows variable distribution offset during the test of the control data set, it adjusts the weights and retrains the steel deviation prediction model; inputs the construction physical model into the optimized prediction model and outputs the prediction result of the steel deviation trend.

[0030] In summary, the steel deviation prediction module of this embodiment realizes the accurate prediction of steel deviation through the fusion of historical data and real-time data, model training and optimization; provides data-driven technical support for quality control, error warning and decision-making optimization during the construction process, helps to improve construction accuracy and safety; can effectively reduce material waste and ensure that the project quality meets the design expectations.

[0031] Embodiment 5: Based on Embodiment 4, the variable distribution difference sub-module provided by the embodiment of the present invention includes: A deviation program calculation unit, configured to calculate the distribution difference of each variable in the two data sets, generate a deviation probability value; reflect the deviation degree of the variable in the control data set from the original data set by comparing the deviation probability value; check whether the frequency of each category in the control data set is consistent with the original data set; A target variable adjustment unit, configured to re-adjust the target variable when the value distribution of the target variable in the control data set is different from that of the original training set by more than a preset standard value; An offset degree judgment unit, configured to mark the target variable with a deviation probability greater than a preset value. If the control data set with a probability of 1 completely lacks the variable value; visualize the distribution overlap of the target variable in the original data set and the control data set, and judge the offset degree; generate an offset automation report.

[0032] Among them, the calculation formula of the deviation program calculation unit is as follows:

[0033] In the formula, represents the variable The distribution deviation probability value of represents the variable In the original data set at thek Frequency of a category; Represents a variable In the control dataset, the k frequency of a category; Represents the smoothing coefficient to avoid a zero denominator; Represents a variable In the k standard deviation in a category; n Represents the total number of categories of a variable; Calculation formula of the target variable adjustment unit:

[0034] In the formula, Represents the target variable distribution difference index; Represents the target variable of the j n-th sample in the original training set value; Represents the target variable of the j n-th sample in the control dataset value; Represents the average value of the target variable in the original training set ; Represents the average value of the target variable in the control dataset ; m Represents the total number of samples; Calculation formula of the deviation degree judgment unit:

[0035] In the formula, Represents the target variable deviation degree index; Represents the target variable probability density function in the original dataset; Represents the target variable probability density function in the control dataset; Represents the mean value in the original dataset; Represents the target variable mean value in the control dataset; Represents the target variable standard deviation in the original dataset; Represents the target variable Standard deviation in the control dataset. The deviation program calculation unit formula is used to measure the distribution deviation of variables between the original dataset and the control dataset; the core idea is to compare the frequency differences of variables in the two datasets, while adding adjustments of standard deviation and natural logarithm to ensure robustness to outliers; by calculating the distribution deviation probability value, the system can quantify the distribution differences of each variable in the two datasets, so as to identify those variables that show significant fluctuations in the control dataset. The target variable adjustment unit is used to evaluate the distribution differences of the target variable in the two datasets; it combines the concepts of mean square error and standard deviation, and calculates the distribution difference index between them by comparing the specific values of the target variable in the original dataset and the control dataset; through the distribution difference index, the system can clarify the difference degree of the target variable in the two datasets, help adjust the model parameters, and ensure that the prediction results are closer to the actual values. It is particularly important in multivariate adjustment. The deviation degree judgment unit calculates the difference in the probability density function of variables between the original dataset and the control dataset through integration, and combines the mean and standard deviation to quantify the deviation degree of variables; it provides a comprehensive judgment of variable deviation, helps the system identify those variables with significant differences in probability distribution, and thus provides a decision-making basis for the adjustment mechanism.

[0036] In the above embodiment, the variable distribution difference sub-module combines and uses the deviation program calculation unit, the target variable adjustment unit, and the deviation degree judgment unit; by calculating the distribution differences of variables in the control dataset and the original dataset, it generates a deviation probability value to accurately quantify the deviation degree of variable distribution; it checks whether the frequencies of various categories in the control dataset are consistent with the original dataset to ensure the consistency of data distribution; when the distribution of the target variable in the control dataset differs from the original training set by more than a preset standard, it automatically re-adjusts the target variable to repair the distribution deviation problem; it marks the variables with deviation probability values exceeding the preset threshold, and compares the distribution overlap of variables in the original dataset and the control dataset through visualization means, and the system evaluates the deviation degree; it generates a deviation automation report to provide a clear basis for the evaluation of data quality and subsequent optimization. This embodiment realizes the precise control and optimization of the distribution differences of the dataset through systematic calculation, adjustment, and evaluation functions, ensures the consistency and reliability of the dataset, and provides technical support for data quality control and analysis.

[0037] Embodiment 6: Based on Embodiment 4, the optimal model sub-module provided by the embodiment of the present invention includes: The prediction result generation unit is used to organize the historical steel structure data according to features to form an initial training set; generate multiple base models based on the historical steel structure data, and use the initial training set to train multiple independent base models respectively to generate corresponding prediction results, and the prediction results are the capture of the trend or law of steel deviation in the historical steel structure data by the base models. A feature combination processing unit, configured to combine the prediction results of each base model with the features of the initial training set to form an extended feature space; for each prediction value generated by a base model, the number of features in the initial training set increases by the same number as the number of base models. A normalization processing unit, configured to apply the prediction results of the base models to the test set to generate a new test set; and perform normalization processing on the newly generated training set and test set to eliminate the dimensional differences between different features.

[0038] In the above embodiment, the optimal model sub-module can efficiently construct and optimize the steel deviation prediction model, thereby improving the accuracy and generalization ability of the model. Specifically: The prediction result generation unit sorts out the historical steel structure data to form an initial training set, and then generates multiple independent base models based on this data; each base model captures the trend or pattern of steel deviation from different perspectives and generates corresponding prediction results; through the training of the base models, a preliminary understanding of the diversity and complexity of steel deviation is established. The feature combination processing unit combines the prediction results of each base model with the original features in the initial training set to form an extended feature space; not only retains the core information of the original data, but also incorporates the capture results of the base models on the deviation trend, making the feature space more rich and comprehensive, which helps to improve the expression ability of the model. The normalization processing unit applies the prediction results of the base models to the test set to generate a new test set, and performs normalization processing on the newly generated training set and test set; normalization eliminates the dimensional differences between different features, enabling the model to process each feature more fairly during training and prediction, thereby improving the stability and prediction accuracy of the model.

[0039] In summary, the optimal model sub-module of this embodiment can extract more comprehensive feature information from historical data, combine the prediction results of multiple base models to form a more robust and accurate steel deviation prediction model. It not only enhances the model's ability to handle data complexity, but also ensures the stability and reliability of the model through normalization, and finally provides a scientific and efficient solution for the prediction of steel deviation.

[0040] Embodiment 7: Based on Embodiment 6, the prediction result generation unit provided by the embodiment of the present invention includes: A parameter combination design sub-unit, configured to design multiple unique parameter combinations according to the multi-dimensional features of the initial training set, and design a base model for capturing dynamic change trends to identify the trends or patterns of steel deviation. A data deconstruction processing subunit, which is used to deconstruct historical steel structure data from multiple representation perspectives according to different representation forms of short-term deviation, long-term deviation, local deviation and global deviation of steel deviation, so as to obtain the recognition results corresponding to short-term deviation, long-term deviation, local deviation and global deviation; and use weighted average to integrate the recognition results into a comprehensive deviation prediction value; An index adding subunit, which is used to add a confidence index to the deviation prediction value and use Monte Carlo simulation to evaluate the reliability and uncertainty of the prediction result.

[0041] In the above embodiments, through the organic combination of the parameter combination design subunit, the data deconstruction processing subunit and the index adding subunit; capturing the dynamic change trend, through designing a unique parameter combination, the base model can effectively identify the trend and law of steel deviation, providing a reliable basis for prediction; comprehensively deconstructing the deviation representation form, deconstructing historical data from four dimensions of short-term, long-term, local and global, can grasp the characteristics of deviation more comprehensively and meticulously, ensuring the comprehensiveness and accuracy of the prediction result; integrating and optimizing the prediction result, integrating the recognition results of multi-dimensional deviation into a comprehensive deviation prediction value through weighted average, ensuring the comprehensiveness and practicability of the prediction result. Evaluating the prediction reliability, introducing the confidence index and Monte Carlo simulation, can quantify the reliability and uncertainty of the prediction result, helping users better understand the applicable range and potential risks of the prediction result. It can provide comprehensive, accurate and reliable prediction results in complex steel deviation scenarios, providing a scientific basis for decision-making.

[0042] Embodiment 8: Based on Embodiment 4, the prediction deviation sub-module provided by the embodiment of the present invention includes: An adjustment mechanism triggering unit, which is used to calculate the deviation probability, category deviation and target variable deviation by comparing the variable distributions of the control data set and the original training set. If a distribution deviation greater than the preset value is detected, the weight adjustment mechanism is triggered; after the base model generates the secondary data set, the meta-model optimizes the value-taking process through backpropagation; A predicted value generating unit, which is used to re-train the steel deviation prediction model based on the updated weights; input the physical model parameters of the current construction into the adjusted steel deviation prediction model to generate the predicted value of future steel deviation; A deviation judgment unit, which restores the standardized predicted value to the actual physical dimension; restores the long-term change law of the data by fitting the trend term; calculates the absolute deviation and relative deviation between the predicted value and the actual measured value, and identifies the abnormal points; if the deviation exceeds the threshold, an early warning is issued.

[0043] Among them, the calculation formula of the adjustment mechanism triggering unit is as follows:

[0044] In the formula, represents the trigger threshold of the weight adjustment mechanism; represents the variable 's distribution deviation probability value (same as the deviation program calculation unit); represents the variable 's deviation degree index (same as the deviation degree judgment unit); represents the target variable 's distribution difference index (same as the target variable adjustment unit); N represents the total number of variables; m represents the total number of target variables; The calculation formula of the predicted value generation unit:

[0045] In the formula, represents the prediction deviation value; represents the weight of the l th base model in the meta-model; represents the activation function; represents the r th construction physical model parameter; represents the r th model coefficient of the parameter; represents the s th standardized parameter; represents the s th weight of the standardized parameter; represents the model error term; L, R, and S respectively represent the number of base models, the number of construction parameters, and the number of standardized parameters; The calculation formula of the deviation judgment unit:

[0046] In the formula, represents the deviation degree between the predicted value and the actual value; represents the t th predicted value at the time point; represents the t th actual measured value at the time point; represents the standard deviation of the predicted value; represents the standard deviation of the actual measured value; represents the mean value of the predicted value; represents the mean value of the actual measured value; T represents the total number of time points; The adjustment mechanism trigger unit formula synthesizes the deviation degree and the offset degree, and combines the distribution difference of the target variable to calculate the trigger threshold of the weight adjustment mechanism; it can judge whether the model needs to be adjusted; when the trigger threshold exceeds a certain threshold, the adjustment mechanism will be activated to ensure that the prediction result of the model is more accurate. The predicted value generation unit generates the final predicted deviation value by combining the weights of multiple base models and construction parameters, and using the activation function and exponential weighting method; it synthesizes the influences of multiple models and parameters and provides the basic data for deviation judgment. The deviation judgment unit calculates the deviation degree between the predicted value and the actual value by comparing the predicted value and the actual measured value, and using the proportional relationship between the exponential function and the standard deviation; it provides a quantitative evaluation of the prediction accuracy, helps the system understand the gap between the prediction result and the actual value, and further optimizes the model parameters and the adjustment mechanism.

[0047] In the above embodiment, the prediction deviation sub-module realizes the following overall significance through the combined use of multiple functional units: by comparing the variable distributions of the control data set and the original training set, it calculates the deviation probability, category deviation, and target variable deviation. If the detected distribution deviation is greater than the preset value, the weight adjustment mechanism is triggered, and the meta-model optimizes the values through backpropagation to ensure that the model parameters adapt to the data changes; based on the updated weights, the steel deviation prediction model is retrained, and the physical model parameters of the current construction are input into the adjusted prediction model to generate the predicted value of the future steel deviation, providing data support for decision-making; the standardized predicted value is restored to the actual physical dimension, and the long-term change law of the data is restored by fitting the trend term; the absolute deviation and relative deviation between the predicted value and the actual measured value are calculated to identify abnormal points; if the deviation exceeds the threshold, a warning is issued to ensure the reliability of the prediction result. This embodiment realizes the dynamic optimization and reliability guarantee of the steel deviation prediction model through a systematic deviation detection, weight adjustment, predicted value generation, and deviation judgment and warning mechanism, providing technical support for construction quality control and prediction.

[0048] Embodiment 9: As Figure 4 shown, on the basis of Embodiment 1, the correction suggestion generation module provided by the embodiment of the present invention includes: The first trend analysis module is used to analyze the prediction deviation trend to obtain whether the prediction deviation shows short-term local expansion or long-term global accumulation; for short-term local expansion, a local correction plan is made; for long-term global accumulation, a comprehensive adjustment is carried out; The second trend analysis module is used to obtain geometric dimension deviation, stress deformation deviation or weight deviation from the prediction deviation trend; Among them, for geometric dimension deviation such as length, width or height deviation, an adjustment plan is generated through a parametric design tool, such as suggesting local cutting and welding or overall component replacement; Optimize the support structure or adjust the external load distribution based on stress and deformation deviation; perform mechanical property tests on bolts, nuts, washers, etc. in the support structure; Weight deviation provides solutions for weight adjustment or material optimization distribution; The execution instruction conversion module is used to convert geometric dimension deviation, stress deformation deviation or weight deviation into executable instructions, and work in conjunction with construction equipment to automatically adjust welding parameters, displacement component position or apply auxiliary support according to correction suggestions; and detect welds through ultrasonic waves.

[0049] In the above embodiment, the first trend analysis module distinguishes between the deviation trends of short-term local expansion and long-term global accumulation, and provides local correction or comprehensive adjustment plans respectively, thereby ensuring the pertinence and efficiency of the correction measures; the second trend analysis module starts from the three dimensions of geometric dimension deviation, stress deformation deviation and weight deviation, and proposes scientific and reasonable correction plans respectively, such as parametric design tools to generate adjustment plans, optimize support structures or adjust load distribution, etc., to ensure the coverage and operability of the correction strategy; automated execution and linkage, the execution instruction conversion module converts the correction suggestions into executable instructions, and links with the construction equipment to realize automatic adjustment of welding parameters, component position displacement or application of auxiliary supports, reducing the need for manual intervention and improving construction efficiency and consistency. Through the collaborative work of various modules, the system can comprehensively analyze, scientifically design and automatically execute deviation problems, realize closed-loop management from prediction to correction, improve construction quality, reduce costs, and provide strong technical support for the intelligent development of steel structure construction.

[0050] Figure 5 A block diagram is shown of an exemplary electronic device suitable for implementing embodiments of the present invention.

[0051] The electronic device may include a central processing unit / microprocessor / main control chip, etc.; a storage medium, coupled to the central processing unit / microprocessor / main control chip, etc., and storing computer-executable instructions therein, for performing the steps of each method of an embodiment of the present invention when executed by the processor.

[0052] The central processing unit / microprocessor / main control chip etc. may include but is not limited to, for example, one or more processors or microprocessors etc.

[0053] The storage medium may include, but is not limited to, for example, random access memory (RAM), read-only memory (ROM), flash memory, EPROM memory, EEPROM memory, registers, computer storage media (such as hard disk, floppy disk, solid-state drive, removable disk, CDROM, DVDROM, Blu-ray disc, etc.).

[0054] In addition, the electronic device may further include (but is not limited to) a data bus, an input / output bus / external bus / device bus, etc., a display, and input / output devices (such as a keyboard, a mouse, a speaker, etc.).

[0055] The central processing unit / microprocessor / master control chip, etc. can communicate with external devices via an I / O bus through a wired or wireless network (not shown).

[0056] The storage medium can also store at least one computer-executable instruction for performing the various functions and / or method steps in the embodiments described in the present technology when run by the central processing unit / microprocessor / master control chip, etc.

[0057] In one embodiment, the at least one computer-executable instruction can also be compiled into or form a software product, and when one or more computer-executable instructions are run by a processor, the various functions and / or method steps in the embodiments described in the present technology are performed.

[0058] Figure 6 FIG. shows a schematic diagram of a computer-readable storage medium according to an embodiment of the present invention.

[0059] As Figure 6 shown, instructions are stored on a non-transitory computer-readable storage medium, and the instructions are, for example, computer-readable instructions. When the computer-readable instructions are run by a processor, the various methods described above can be executed. The non-transitory computer-readable storage medium includes but is not limited to, for example, volatile memory and / or non-volatile memory. Volatile memory may, for example, include random access memory (RAM) and / or cache memory, etc. Non-transitory non-volatile memory may, for example, include read-only memory (ROM), a hard disk, a flash memory, etc. For example, the non-transitory computer-readable storage medium can be connected to a computing device such as a computer, and then, when the computing device runs the computer-readable instructions stored on the non-transitory computer-readable storage medium, the various methods described above can be performed.

[0060] In several embodiments provided by the present invention, it should be understood that the disclosed apparatus and method can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical function division, and there may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the couplings or direct couplings or communication connections shown or discussed with each other can be through some interfaces, and the indirect couplings or communication connections of the devices or units can be in electrical, mechanical or other forms.

[0061] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed over multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0062] In addition, each functional unit in various embodiments of the present invention may be integrated in a processing unit, may exist separately as individual physical units, or two or more units may be integrated in one unit. The above-mentioned integrated units can be implemented in the form of hardware or in the form of software functional units.

[0063] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that makes a contribution to the prior art, or all or part of this technical solution, can be embodied in the form of a software product. This computer software product stored in a storage medium specifically includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods in various embodiments of the present invention. The aforementioned storage medium includes: USB flash drives, mobile hard disks, read-only memories (English full name: Read Only Memory, English abbreviation: ROM), random access memories (English full name: Random Access Memory, English abbreviation: RAM), magnetic disks or optical discs and other various media that can store program codes.

[0064] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of various embodiments of the present invention.

Claims

1. An intelligent detection and deviation correction system for steel structure construction, characterized in that, Including: A virtual model module for integrating multiple sensors to collect geometric dimensions, weight deviations, and stress deformation data of steel structure components in real time; constructing a construction physical model based on the physical construction scenario, comparing the point cloud data with the construction physical model, and if the analysis result is greater than the preset value, outputting it to the steel deviation prediction module; A steel deviation prediction module for constructing a steel deviation prediction model based on historical and real-time steel structure data, training the steel deviation prediction model according to the analysis result of the construction physical model; inputting the construction physical model into the steel deviation prediction model to predict the deviation trend; A correction suggestion generation module for generating correction suggestions according to the predicted deviation trend; Triggering an automatic adjustment instruction based on the correction suggestion; If the deviation exceeds the preset threshold, issuing a warning and locating the deviation problem area; The steel deviation prediction module includes: A variable distribution difference sub-module for constructing a steel deviation prediction model based on historical and real-time steel structure data, merging the historical and real-time steel structure data to obtain a data set, and dividing it into a training set and a test set; introducing a control data set to compare the variable distribution deviation between the original training set and the control data set; An optimal model sub-module for training multiple base models using historical steel structure data to generate a new training set and a test set; training a meta-model with the new data set generated by the base model, optimizing the weights through backpropagation; determining the optimal steel deviation prediction model through cross-validation; A predicted deviation sub-module for adjusting the weights and retraining the steel deviation prediction model if the variable distribution shift is shown in the control data set test; inputting the construction physical model into the steel deviation prediction model to predict the deviation trend.

2. The intelligent detection and deviation correction system for steel structure construction according to claim 1, characterized in that The virtual model module includes: A coordinate alignment sub-module for aligning the measured point cloud data with the construction physical model in the coordinate system, selecting the positions of the common steel structure nodes between the two, and converting the measured point cloud from the scanning coordinate system to the design coordinate system of the construction physical model; performing spatial positioning by generating point cloud lines; A construction attribute comparison sub-module for comparing the geometric differences between the point cloud and the construction virtual model, calculating the deviation values between the actual cloud points and the construction physical model point by point; performing semantic segmentation on the point cloud to identify basic components and making object-based comparisons with the corresponding construction attributes in the construction physical model; A data set generation sub-module for fitting the point cloud data into the elements of the construction physical model for the identified key components and comparing with the design values; generating a virtual point cloud data set from the parameters in the construction physical model and dividing the virtual point cloud data set into a training set, a validation set, and a test set.

3. The intelligent detection and deviation correction system for steel structure construction according to claim 2, wherein The construction attribute comparison sub-module includes: A point cloud segmentation unit for performing hierarchical feature extraction on the point cloud, learning geometric features point by point; adaptively constructing a local neighborhood graph to enhance the feature expression of the steel beam nodes through the graph attention mechanism; performing plane segmentation on the point cloud to separate the building floor plan and the component point cloud; Independent cluster unit, used to divide the point cloud into independent clusters, each cluster corresponding to a basic mechanism, and refined segmentation is carried out in combination with silver information; for multi-source data, the 2D semantic segmentation result is mapped to the 3D point cloud to supplement semantic information; Deviation calculation unit, used to compare the segmented construction point cloud object by object with the construction attributes in the construction physical model through feature matching to calculate the geometric deviation; if abnormal segmentation occurs, a feedback optimization mechanism is triggered to adjust the clustering threshold.

4. The intelligent detection and deviation correction system for steel structure construction according to claim 1, wherein, Variable distribution difference sub-module, including: Deviation program calculation unit, used to calculate the distribution difference of each variable in two datasets, generate a deviation probability value; the deviation degree of the variable in the control dataset from that in the original dataset is reflected by comparing the deviation probability values; check whether the frequency of each category appearing in the control dataset is consistent with the original dataset; Target variable adjustment unit, used to re-adjust the target variable when the value distribution of the target variable in the control dataset is different from that in the original training set by more than a preset standard value; Deviation degree judgment unit, used to mark the target variable with a deviation probability greater than the preset value, if the control dataset with a probability of 1 completely lacks the variable value; visualize the distribution overlap of the target variable in the original dataset and the control dataset to judge the deviation degree; generate an automated deviation report.

5. The intelligent detection and deviation correction system for steel structure construction according to claim 1, characterized in that, Optimal model sub-module, including: Prediction result generation unit, used to organize the historical steel structure data according to features to form an initial training set; generate multiple base models based on the historical steel structure data, and use the initial training set to train multiple independent base models respectively to generate corresponding prediction results, and the prediction results are the capture of the trend or law of steel deviation in the historical steel structure data by the base models; Feature combination processing unit, used to combine the prediction results of each base model with the features of the initial training set to form an extended feature space; if each base model generates a prediction value, the number of features of the initial training set increases by the same number as the number of base models; Standardization processing unit, used to apply the prediction results of the base models to the test set to generate a new test set; perform standardization processing on the newly generated training set and test set to eliminate the dimensional difference between different features.

6. The intelligent detection and deviation correction system for steel structure construction according to claim 5, characterized in that, Prediction result generation unit, including: Parameter combination design sub-unit, used to design multiple unique parameter combinations according to the multi-dimensional features of the initial training set, and design a base model that captures the dynamic change trend to identify the trend or law of steel deviation; Data deconstruction processing sub-unit, used to deconstruct the historical steel structure data from the perspectives of multiple representation forms of short-term deviation, long-term deviation, local deviation and global deviation of steel deviation, and obtain the recognition results corresponding to short-term deviation, long-term deviation, local deviation and global deviation; use weighted average to integrate the recognition results into a comprehensive deviation prediction value; Index addition sub-unit, used to add a confidence index to the deviation prediction value and evaluate the reliability and uncertainty of the prediction results using Monte Carlo simulation.

7. The intelligent detection and deviation correction system for steel structure construction according to claim 1, characterized in that, Prediction deviation sub-module, including: Adjustment mechanism trigger unit, which is used to calculate the deviation probability, category deviation and target variable deviation by comparing the variable distributions of the control data set and the original training set. If a distribution deviation greater than the preset value is detected, the weight adjustment mechanism is triggered; after the base model generates the secondary data set, the meta-model optimizes the value through backpropagation; Predicted value generation unit, which is used to retrain the steel deviation prediction model based on the updated weights; input the physical model parameters of the current construction into the adjusted steel deviation prediction model to generate the predicted value of the future steel deviation; Deviation judgment unit, which restores the standardized predicted value to the actual physical dimension; restores the long-term change law of the data by fitting the trend term; calculates the absolute deviation and relative deviation between the predicted value and the actual measured value to identify abnormal points; if the deviation exceeds the threshold, a warning is issued.

8. The intelligent detection and deviation correction system for steel structure construction according to claim 1, characterized in that, Correction suggestion generation module, including: The first trend analysis module, which is used to analyze the trend of the prediction deviation to obtain whether the prediction deviation shows short-term local expansion or long-term global accumulation; for short-term local expansion, a local correction plan; for long-term global accumulation, a comprehensive adjustment; The second trend analysis module, which is used to obtain the geometric dimension deviation, stress deformation deviation or weight deviation from the trend of the prediction deviation; Execution instruction conversion module, which is used to convert the geometric dimension deviation, stress deformation deviation or weight deviation into executable instructions, link with the construction equipment, and automatically adjust the welding parameters, the position of the displacement member or apply auxiliary support according to the correction suggestion; detect the welded seam by ultrasonic wave.

9. The intelligent detection and deviation correction system for steel structure construction according to claim 8, wherein, Among them, For the geometric dimension deviation of length, width or height, an adjustment plan is generated through a parametric design tool, and it is recommended to perform local cutting and welding or replace the overall component; For the stress deformation deviation, optimize the support structure or adjust the external load distribution; perform mechanical property tests on bolts, nuts and washers in the support structure; For the weight deviation, provide a plan for counterweight adjustment or optimized material distribution.

Images