A curtain wall keel structure intelligent optimization design method

Abstract

The application relates to the technical field of architectural curtain wall keel structure design, and discloses an intelligent optimization design method for a curtain wall keel structure, which comprises the following steps: firstly, geometric parameters, load distribution data and material performance indexes are acquired and verified; secondly, multi-dimensional structure characteristic parameters are generated by analyzing the verified data, and a structure optimization benchmark model is constructed; thirdly, the keel structure data are verified in real time according to the model, and abnormal matching of node displacement and stress distribution is quantitatively evaluated; fourthly, structure abnormality indexes are generated according to the verification results to judge design defects; and finally, when defects exist, optimization design schemes containing keel section adjustment, node reinforcement and the like are automatically generated and sorted. The method comprehensively considers multiple factors, realizes multi-target optimization, can improve the design quality and efficiency of the curtain wall keel structure, guarantees the safety of the structure and reduces the cost.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of architectural curtain wall keel structure design, in particular to an intelligent optimization design method for curtain wall keel structure. BACKGROUND

[0002] With the rapid development of the construction industry, architectural curtain walls are widely used in modern buildings due to their aesthetic appearance, transparency and other characteristics. As the key support system of architectural curtain walls, the rationality and reliability of the design of curtain wall keel structure directly affect the safety, stability and service life of the curtain wall.

[0003] Traditional curtain wall keel structure design mainly relies on the experience of engineers and conventional mechanical calculation methods. In the design process, it is often difficult to accurately consider the influence of actual factors such as panel size deviation and connector positioning error on the structural performance when dealing with geometric parameters, resulting in some deviations between the design and actual construction. In terms of load distribution data processing, complex load factors such as wind pressure gradient changes and temperature stress distribution are usually simplified, which cannot fully and accurately reflect the load conditions that the curtain wall bears in actual use. For material performance indicators, only the average performance parameters of the material are considered, and the actual changes in material performance such as elastic modulus dispersion value and yield strength fluctuation range are ignored, which may cause performance deficiency or material waste of the designed keel structure when facing actual working conditions.

[0004] In the design verification link, the traditional method mainly uses static and local analysis methods, which are difficult to monitor the mechanical performance changes of the keel structure under different working conditions in real time. The correlation analysis of node displacement and stress distribution is not deep enough, and the abnormal matching of node displacement and stress distribution cannot be found in time, so that potential design defects cannot be found and corrected in time. Moreover, the traditional design method lacks systematicness and scientificalness in optimizing the design scheme, and can only optimize for a single target, such as simply pursuing the minimization of structure weight or cost minimization, which is difficult to balance between structure weight, stiffness and cost, and cannot meet the comprehensive requirements of modern architecture for high performance and low cost of curtain wall keel structure. With the continuous increase of building height, the increasing complexity of modeling and the diversification of use environment, the traditional curtain wall keel structure design method has gradually been difficult to meet the actual engineering needs, and an intelligent, efficient and scientific design method is urgently needed to improve the design quality and efficiency of curtain wall keel structure. SUMMARY

[0005] The purpose of the present application is to provide an intelligent optimization design method for curtain wall keel structure to solve the problems raised in the background.

[0006] To achieve the above purpose, the present application provides the following technical scheme: an intelligent optimization design method for curtain wall keel structure, the method comprising:

[0007] S1: Obtain geometric parameters, load distribution data and material performance indicators of the building curtain wall, and perform data verification on the geometric parameters, load distribution data and material performance indicators;

[0008] S2: Analyze the verified data to generate multi-dimensional structure characteristic parameters, select features according to the multi-dimensional structure characteristic parameters to form a feature vector set, and construct a structure optimization benchmark model according to the feature vector set;

[0009] S3: Real-time verification of the keel structure data according to the structure optimization benchmark model, identification of the topological correlation of node displacement and stress distribution by using multi-level data correlation technology, and quantitative evaluation of abnormal matching of node displacement and stress distribution;

[0010] S4: Generating a structure abnormality index according to the verification result of the structure optimization benchmark model, and determining whether there is a design defect according to the structure abnormality index.

[0011] Preferably, the geometric parameters include panel size deviation and connector positioning error; the load distribution data includes wind pressure gradient change and temperature stress distribution; the material performance indicators include elastic modulus discrete value and yield strength fluctuation range; the data verification on the geometric parameters, load distribution data and material performance indicators includes data normalization, outlier correction, dimension unification and integrity test; the geometric parameter verification generates a geometric feature sequence, and the load distribution data verification generates a load feature sequence; and the geometric feature sequence and the load feature sequence of different dimensions are spatially associated to form a multi-source structure feature set.

[0012] Preferably, in S2, the following steps are included:

[0013] S201: Extracting spatial feature parameters from the verified geometric feature sequence, load feature sequence and material performance indicators, respectively, the spatial feature parameters including stiffness variation coefficient, connection point offset and stress concentration degree, and combining each spatial feature parameter according to a preset weight to generate multi-dimensional structure characteristic parameters;

[0014] S202: Setting a dynamic threshold for the multi-dimensional structure characteristic parameters in the multi-source structure feature set, selecting feature parameters within the threshold range to form an initial feature vector set, and excluding feature parameters outside the threshold range;

[0015] S203: Cross-condition correlation analysis of each feature parameter in the initial feature vector set, extracting the deviation degree of the feature parameters in different conditions under the same load condition, calculating the variation coefficient of the deviation degree and marking it as a condition difference parameter;

[0016] S204: Compare the inter-condition difference parameters with the preset tolerance threshold, filter out the characteristic parameters exceeding the tolerance threshold to join the structure optimization benchmark model, and supplement the yield strength mutation characteristics to the structure optimization benchmark model according to the material performance index.

[0017] Preferably, in S3, the quantitative evaluation on the abnormal matching of node displacement and stress distribution includes the following steps:

[0018] S301: Real-time monitor the displacement change amount of the keel node and the stress change amount at the corresponding position, calculate the difference value of the change rates of the two, and if the difference value exceeds the preset reasonable interval, it is determined as an abnormal mechanical event;

[0019] S302: Count the number of abnormal mechanical events in the preset analysis period as M, and simultaneously obtain the elastic modulus fluctuation amplitude E in the material performance index;

[0020] S303: According to the correlation between the number of abnormal mechanical events M and the elastic modulus fluctuation amplitude E, a nonlinear function mapping is used to generate a mechanical abnormality index, wherein the product component and the ratio component of M and E are generated by superposition operation to generate a comprehensive evaluation value;

[0021] If the mechanical abnormality index exceeds the preset safety threshold, a design defect flag is triggered.

[0022] Preferably, in S4, the verification results of the structure optimization benchmark model in the same building curtain wall historical design data are extracted, the deviation values of each benchmark model are weighted and accumulated to generate a structure abnormality index, and if the structure abnormality index continuously exceeds the preset trigger number, it is determined that there is a design defect.

[0023] Preferably, in S202, the setting method of the dynamic threshold includes: based on the historical engineering database, the distribution law of each characteristic parameter is counted, the confidence interval of the characteristic parameter is calculated by using the probability density function, the upper limit and the lower limit of the confidence interval are taken as the initial value of the dynamic threshold, and the threshold range is dynamically adjusted according to the characteristic parameter distribution of the current project.

[0024] Preferably, in S303, the specific implementation mode of the nonlinear function mapping is: a coupling function is established with the number of abnormal mechanical events M as the independent variable and the elastic modulus fluctuation amplitude E as the dependent variable, a prediction model of the mechanical abnormality index is obtained by machine learning algorithm training, the prediction model adopts a neural network structure, the input layer includes the normalized values of M and E, and the output layer is the mechanical abnormality index.

[0025] Preferably, the method further comprises S5: when it is determined that there is a design defect, automatically generating an optimized design scheme, the optimized design scheme comprising keel section size adjustment suggestions, connection node reinforcement measures and material replacement schemes, and sorting the priority of the optimized design scheme according to the size of the structure abnormality index.

[0026] Preferably, in S5, the generation method of the optimized design scheme comprises: constructing a multi-objective optimization function, taking the minimization of the structure weight, the maximization of the stiffness and the minimization of the cost as the optimization objectives, using a genetic algorithm for multi-scheme iterative solution, and outputting a Pareto optimal solution set as a candidate optimized design scheme.

[0027] Compared with the prior art, the beneficial effects of the present application are:

[0028] The intelligent optimization design method for curtain wall keel structure provided by the present application realizes the improvement of comprehensiveness and accuracy in the data processing link. By obtaining the geometric parameters, load distribution data and material performance indicators of the building curtain wall, and performing strict data verification such as data normalization, outlier correction, dimension unification and integrity test, the actual factors such as panel size deviation, wind pressure gradient change and elastic modulus discrete value are fully considered, the data of different dimensions are spatially correlated to form a multi-source structure feature set, so that the subsequent design is based on a data basis that is more in line with the actual working conditions, and the design deviation caused by inaccurate or incomplete data is avoided.

[0029] In the model construction aspect, by extracting spatial feature parameters and scientifically combining to form multi-dimensional structure feature parameters, a dynamic threshold screening and cross-condition correlation analysis method is used to construct a structure optimization benchmark model. The model can comprehensively consider the influence of various factors on the structure performance, and can more accurately reflect the mechanical properties of the curtain wall keel structure in actual use than the traditional model, providing a reliable basis for subsequent design verification and optimization.

[0030] In the design verification process, the multi-level data correlation technology is used to verify the keel structure data in real time, and the abnormal matching of node displacement and stress distribution is quantitatively evaluated. Through real-time monitoring and accurate calculation, potential problems of the structure under different working conditions can be found in time, and compared with the traditional static and local analysis method, design defects can be found earlier, which can save time for ensuring the safety of the curtain wall structure.

[0031] In the aspect of design defect judgment and optimization, the structural abnormality index is generated to judge the design defect, and the optimization design scheme is automatically generated when the defect exists. The scheme optimizes multiple targets of minimizing the structure weight, maximizing the stiffness and minimizing the cost, adopts the genetic algorithm to obtain the Pareto optimal solution set, and sorts according to the structural abnormality index. This way changes the limitation of traditional single target optimization, can find a balance between multiple important targets, ensures the safety and performance of the structure, realizes effective control of the cost, improves the material utilization rate and reduces resource waste. At the same time, the mechanism of automatically generating and sorting the scheme greatly improves the efficiency of design optimization, reduces the time and energy cost of manual design, provides an efficient, scientific and intelligent solution for the design of the building curtain wall keel structure, and promotes the development of the building curtain wall design technology. BRIEF DESCRIPTION OF DRAWINGS

[0032] Fig. 1 The working principle diagram of the intelligent optimization design method of the curtain wall keel structure is described.

[0033] Fig. 2 The generation diagram of data verification and feature set is described.

[0034] Fig. 3 The flowchart of the construction of the structure optimization benchmark model is described. DETAILED DESCRIPTION

[0035] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0036] Please refer to Figs. 1-3 The intelligent optimization design method of the curtain wall keel structure is described.

[0037] Step S1: Obtain the geometric parameters, load distribution data and material performance indicators of the building curtain wall, and perform data verification on the geometric parameters, load distribution data and material performance indicators. The geometric parameters include panel size deviation and connector positioning error; the load distribution data covers wind pressure gradient change and temperature stress distribution; the material performance indicators involve elastic modulus discrete value and yield strength fluctuation range. The data verification process includes data normalization, outlier correction, dimension unification and integrity inspection. After the geometric parameter verification, a geometric feature sequence is generated, and after the load distribution data verification, a load feature sequence is generated; the geometric feature sequence and the load feature sequence of different dimensions are spatially associated to form a multi-source structure feature set.

[0038] Step S2: analyzing the checked data to generate multi-dimensional structure characteristic parameters, selecting features according to the multi-dimensional structure characteristic parameters to form a feature vector set, and constructing a structure optimization benchmark model according to the feature vector set. In specific implementation, first, spatial characteristic parameters including stiffness variation coefficient, connection point offset and stress concentration degree are extracted from the checked geometric feature sequence, load feature sequence and material performance index, and each spatial characteristic parameter is combined according to a preset weight to generate multi-dimensional structure characteristic parameters; then, dynamic threshold values are set for the multi-dimensional structure characteristic parameters in the multi-source structure characteristic set, and feature parameters within the threshold range are selected to form an initial feature vector set, and feature parameters exceeding the threshold range are excluded; then, cross-condition correlation analysis is performed on each feature parameter in the initial feature vector set, the deviation degree of the feature parameter in different conditions under the same load condition is extracted, the variation coefficient of the deviation degree is calculated and marked as a condition difference parameter; finally, the condition difference parameter is compared with a preset tolerance threshold, and feature parameters exceeding the tolerance threshold are added to the structure optimization benchmark model, and yield strength mutation features are supplemented to the structure optimization benchmark model according to the material performance index.

[0039] Step S3: real-time verification of the keel structure data according to the structure optimization benchmark model, identification of the topological correlation of node displacement and stress distribution by using multi-level data correlation technology, and quantitative evaluation of abnormal matching of node displacement and stress distribution. The specific operation is to monitor the node displacement change amount and the stress change amount at the corresponding position in real time, calculate the difference value of the change rates of the two, and if the difference value of the change rates exceeds the preset reasonable interval, it is determined as an abnormal mechanical event; the number of abnormal mechanical events in a preset analysis period is counted and recorded as M, and the elastic modulus fluctuation amplitude in the material performance index is obtained synchronously and recorded as E; according to the correlation between the number of abnormal mechanical events M and the elastic modulus fluctuation amplitude E, a nonlinear function mapping is used to generate a mechanical abnormality index, wherein the product component and the ratio component of M and E are generated by superposition operation to generate a comprehensive evaluation value; if the mechanical abnormality index exceeds the preset safety threshold, a design defect flag is triggered.

[0040] Step S4: generating a structure abnormality index according to the verification result of the structure optimization benchmark model, and determining whether there is a design defect according to the structure abnormality index. Specifically, the verification results of the structure optimization benchmark model in the same building curtain wall historical design data are extracted, the deviation values of each benchmark model are weighted and accumulated to generate a structure abnormality index, and if the structure abnormality index continuously exceeds the preset trigger number, it is determined that there is a design defect.

[0041] The application will be further described below in combination with Examples 1 to 5: Example

[0042] In the data checking stage, for the panel size deviation in geometric parameters, due to the production process, measurement error and other factors in the actual production and installation process, the panel size may deviate, and these deviation data may exist in different measurement units or forms. In the normalization process, first of all, the original unit and representation of the data should be clear, for example, part of the panel size deviation data may be in millimeters, and another part may be in inches, or some data is in the form of percentage deviation relative to the standard size. For different situations, corresponding conversion rules are adopted to convert all panel size deviation data into a unified unit and standard form, so that these data have comparability and consistency in subsequent processing.

[0043] The connection positioning error also has similar problems, and the sources and representations of the error data are diverse. In the normalization process, according to the design position and actual measurement position of the connection in the building curtain wall structure, the positioning error data is unified to the same measurement system through accurate calculation and conversion. This can eliminate the interference caused by the difference in data representation and lay a foundation for subsequent accurate analysis and processing of geometric parameters.

[0044] In the aspect of abnormal value correction, for panel size deviation and connection positioning error data, statistical methods are used for in-depth analysis. First, calculate the mean, median, standard deviation and other statistical quantities of the data to describe the overall distribution characteristics of the data. By setting a reasonable threshold range, which can usually be determined based on the mean and standard deviation, such as the mean plus or minus several times the standard deviation as the threshold range, to identify data points that deviate significantly from the normal range. When abnormal data points are found, instead of simply deleting these data, reasonable correction is made according to the overall distribution characteristics of the data. For example, if the abnormal data points are caused by measurement error, and the surrounding data shows a certain distribution rule, the values and distribution trend of the adjacent data points can be referred to adjust the abnormal data to conform to the overall data distribution, so as to ensure the reliability and accuracy of the data.

[0045] In the dimension unification process, for geometric parameters, the dimensions of each parameter are converted according to physical principles and engineering specifications. For example, for length-related parameters such as panel size deviation and connection positioning error, they are converted to meters in the International System of Units. When performing dimension conversion, ensure the accuracy of the conversion process to avoid affecting subsequent calculations and analyses due to conversion errors. At the same time, the correlation between different parameters should be considered to ensure that the data after dimension unification can correctly reflect the geometric characteristics of the building curtain wall structure.

[0046] Integrity check is an important part of data verification. For geometric parameters, carefully check if the data fields are complete and ensure that no key information is missing. For example, when recording the panel size deviation, not only the length and width direction deviations, but also the thickness direction deviation should be recorded; for connector positioning error, the positioning deviation in each direction and the related attribute information of the connector should be recorded. Through comprehensive inspection, missing data can be found and supplemented in time, ensuring the integrity of the data and providing a complete data basis for subsequent analysis and modeling.

[0047] For load distribution data, wind pressure gradient change data may be affected by various factors such as weather conditions and measurement equipment during collection, resulting in differences and irregularities in the data. During normalization processing, according to the collection principle of wind pressure data and actual application requirements, wind pressure gradient change data from different sources and under different measurement conditions are converted into a unified standard format and unit. For example, wind pressure data is uniformly converted to Pascal (Pa) as the unit, and is arranged according to the specified time interval and spatial position.

[0048] Temperature stress distribution data also needs to be normalized. Since temperature changes will cause thermal expansion and contraction of materials, resulting in stress, and there may be various factors leading to inconsistency in the data during temperature measurement and stress calculation. Through reasonable conversion and processing, temperature stress distribution data is converted into a unified representation form to facilitate subsequent analysis and comparison.

[0049] In the aspect of abnormal value correction, for load distribution data, statistical methods are also used to identify abnormal data points. By analyzing the distribution of data, a suitable threshold range is set to find data that deviates from the normal range. For these abnormal data, combined with actual engineering conditions and related theories, the causes are judged. If it is caused by measurement error or abnormal weather conditions, it is corrected according to reasonable methods to ensure that the load distribution data accurately reflects the actual load conditions.

[0050] Dimensional uniformity is also crucial for load distribution data. According to the principles of mechanics and engineering specifications, the dimensions of wind pressure gradient change and temperature stress distribution data are unified to meet the requirements of structural analysis. For example, the dimensions of wind pressure data are coordinated with the dimensions of stress data so that the data can be accurately calculated and compared when performing structural stress analysis.

[0051] During integrity check, all aspects of load distribution data are comprehensively checked, including wind pressure gradient change data at different positions and times, and temperature stress distribution data at different parts. It is ensured that no key load information is missed, and the data can completely describe the stress conditions of the building curtain wall under different load conditions.

[0052] For material performance indicators, the elastic modulus discrete value and yield strength fluctuation range data may have inaccuracies and inconsistencies during the acquisition process due to factors such as the material's own characteristics, testing methods, etc. During normalization processing, according to the standards and specifications of material performance testing, the elastic modulus discrete value and yield strength fluctuation range data obtained under different testing conditions are uniformly processed. For example, the elastic modulus data is uniformly converted to the international unit of Pascal (Pa), and is arranged according to the specified accuracy requirements.

[0053] In the aspect of abnormal value correction, statistical methods are used to analyze the material performance indicator data and identify abnormal data points. For these abnormal data, combined with factors such as the production process of the material and the testing process, the reasonableness is judged. If it is an abnormal data caused by testing error, it is corrected according to reasonable methods to ensure that the material performance indicator data can accurately reflect the actual performance of the material.

[0054] In the dimensional unification process, according to the principles of materials mechanics and engineering specifications, the dimensions of the elastic modulus discrete value and yield strength fluctuation range data are unified to the form required by the structure design. Ensure that the material performance indicator data can accurately participate in the calculation and evaluation when performing structure design and analysis.

[0055] During the integrity test, each aspect of the material performance indicator data is carefully checked, including the performance data of different batches of materials and the data obtained by different testing methods. Ensure that there is no missing key material performance information, and provide accurate and complete material performance data for subsequent structure design and optimization.

[0056] After completing the verification of geometric parameters, load distribution data and material performance indicators, the geometric feature sequence and load feature sequence of different dimensions are spatially correlated. First, a three-dimensional coordinate system is established, taking the actual spatial position of the building curtain wall as the reference to determine the origin and coordinate axis direction of the coordinate system. Then, according to the actual position of each parameter in the curtain wall structure, such as panel size deviation and connector positioning error, the parameters in the geometric feature sequence are mapped to the corresponding positions in the spatial coordinate system. For the load feature sequence, the wind pressure gradient change and temperature stress distribution data are also mapped to the spatial coordinate system according to the position and direction of the load action. In this way, the geometric features and load features are corresponded and integrated in space, so that each spatial position corresponds to the corresponding geometric feature and load feature information, forming a multi-source structure feature set. This multi-source structure feature set comprehensively describes the geometric and load characteristics of the building curtain wall structure, providing a rich and accurate data basis for subsequent construction of structure optimization benchmark model, so that the constructed model can more truly reflect the actual situation of the building curtain wall, and provide strong support for subsequent structure optimization design. EMBODIMENT

[0057] In the process of building the structure optimization benchmark model, firstly, the space characteristic parameters of the verified geometric feature sequence, load feature sequence and material performance index are extracted. For the geometric feature sequence, due to the differences in shape, size and connection mode of different parts of the structure, the stiffness varies. When calculating the stiffness variation coefficient, the division of each part of the structure needs to be determined. According to the engineering practice, the curtain wall keel structure is reasonably divided according to the plate, connection area, etc. Then, for each division, according to the panel size deviation, connector positioning error and other data in the geometric feature sequence, combined with the elastic modulus dispersion value and other information in the material performance index, the stiffness of each division is analyzed through mechanical principle. By comparing the stiffness values of different divisions, the variation degree of the stiffness in the whole structure is calculated, and the stiffness variation coefficient is obtained. The determination of the connection point offset depends on the accurate measurement and comparison of the actual position and the design position of the connector. In the construction process of building curtain wall, through high-precision measuring equipment such as total station and laser range finder, the actual coordinates of the connector in three-dimensional space are obtained, and the coordinates in the design drawing are compared one by one to obtain the offset value of each connector in each direction, so as to determine the connection point offset. The analysis of stress concentration degree is mainly aimed at the structure nodes. Since the node is the key part of force transmission, the stress is complex. Through finite element analysis and other methods, the stress distribution state of the node under different load conditions is analyzed, the degree of stress concentration is determined, and the stress concentration degree parameter is obtained.

[0058] For the load feature sequence, the related space characteristic parameters are extracted based on the mechanical principle and engineering actual situation. When considering the wind pressure gradient change and temperature stress distribution, the influence of load on each part of the structure at different positions and times is analyzed to determine the characteristic parameters related to the structure performance. For the material performance index, combined with the mechanical properties of the material, the parameters related to the spatial performance of the structure are extracted, such as the influence parameter of the change of the elastic modulus of the material on the stiffness of the structure, etc.

[0059] The extracted spatial feature parameters are combined according to preset weights to generate multi-dimensional structure feature parameters. The determination of the preset weights is a rigorous process that needs to comprehensively consider the importance of each parameter in the structure performance. Experts in the fields of organizational structure design, material science, engineering mechanics, etc. evaluate each spatial feature parameter according to previous engineering experience, theoretical analysis and in-depth research on the structure performance of building curtain walls. For example, in some curtain wall structures with high stiffness requirements, the weight of the stiffness variation coefficient may be relatively high; and in structures where the connecting nodes are prone to problems, the weights of the connecting point offset and stress concentration may be correspondingly increased. Through scientific methods such as expert scoring and analytic hierarchy process, the specific weight values of each spatial feature parameter are determined, and the parameters are combined according to the weights to form multi-dimensional structure feature parameters, which comprehensively reflect the performance characteristics of the structure in different aspects.

[0060] Dynamic thresholds are set for the multi-dimensional structure feature parameters in the multi-source structure feature set. The setting of the dynamic thresholds is based on a historical engineering database that collects a large amount of engineering data of building curtain walls under different types and different environmental conditions. Through statistical analysis of the historical data, the confidence interval of each feature parameter is calculated using the probability density function. The probability density function can describe the probability distribution of the feature parameters in the data set, and the confidence interval obtained by calculation reflects the value range of the feature parameters under normal conditions. The upper and lower limits of the confidence interval are used as the initial values of the dynamic thresholds. In actual application, the initial threshold range is dynamically adjusted according to the specific feature parameter distribution of the current project. For example, if the material performance of the current project is significantly different from that of the historical projects, or the environmental conditions of the building are special, causing changes in the distribution of some feature parameters, the threshold range is adjusted accordingly to ensure that the threshold accurately reflects the actual situation of the current project. According to the adjusted threshold range, the feature parameters that meet the requirements are selected to form an initial feature vector set, and the feature parameters that exceed the threshold range are temporarily excluded, so that the initial feature vector set contains the parameters that best represent the structure features of the current project.

[0061] Each feature parameter in the initial feature vector set is subjected to cross-condition correlation analysis. In the actual use of building curtain walls, various different conditions may be encountered, such as temperature changes in different seasons and wind pressure actions of different intensities. Under the same load condition, the deviation of the feature parameters in different conditions is analyzed. For example, under the same wind pressure load condition, the changes of the stiffness variation coefficient, connecting point offset and other feature parameters of the structure under high-temperature summer conditions and low-temperature winter conditions are analyzed. By calculating the variation coefficient of the deviation degree of the feature parameters under different conditions, the difference parameters between conditions are obtained. The variation coefficient can reflect the dispersion degree of the data, and by calculating the variation coefficient of the deviation degree, the variation degree of the feature parameters under different conditions can be quantified, so as to evaluate the stability and reliability of the feature parameters under different conditions.

[0062] Finally, the working condition difference parameters are compared with the preset tolerance threshold. The preset tolerance threshold is determined according to the engineering design standard and the actual use requirement, and is used to measure the limit value of whether the characteristic parameters change reasonably under different working conditions. The characteristic parameters exceeding the tolerance threshold are screened out. The changes of these characteristic parameters under different working conditions are more significant, and the influence on the structural performance is larger. They are added to the structural optimization benchmark model. At the same time, according to the related data in the material performance index, the yield strength mutation feature is supplemented to the structural optimization benchmark model. The yield strength mutation is an important change feature of the material performance under some special conditions, which has an important influence on the safety and stability of the structure. By comprehensively considering these factors, the construction of the structural optimization benchmark model is completed, which can accurately reflect the performance characteristics of the building curtain wall keel structure under different working conditions and conditions, and provide a reliable basic model for the subsequent structural optimization design. Embodiment

[0063] When quantitatively evaluating the abnormal matching of node displacement and stress distribution, real-time monitoring is first performed. High-precision displacement sensors and stress sensors are installed at key node positions of the building curtain wall keel structure. These sensors have the characteristics of high precision, high reliability and long-time stable work, and can real-time and accurately collect the displacement change amount of the keel node and the stress change amount data of the corresponding position. During the installation process, the installation position of the sensor needs to be accurately determined to ensure that it can truly reflect the mechanical state of the node. For the displacement sensor, it should be installed at a position that can accurately measure the displacement change of the node in all directions; for the stress sensor, it should be installed at the key position that can effectively measure the stress change of the node. After installation, the sensors are calibrated and debugged to ensure the accuracy and reliability of the measurement data.

[0064] After collecting the displacement change and stress change data of the keel node in real time, the difference between the two change rates is calculated. The calculation of the difference between the change rates is an important basis for evaluating whether the mechanical state of the node is abnormal. By analyzing the displacement change and stress change over a period of time, the change rates of each are calculated. The displacement change rate reflects the speed of the node position change over time, and the stress change rate reflects the speed of the node stress change over time. Subtracting the two change rates gives the change rate difference. If the change rate difference exceeds the pre-set reasonable interval, it is determined to be an abnormal mechanical event. The determination of the pre-set reasonable interval is a complex process that requires a comprehensive consideration of structural mechanics theory and engineering experience. Structural mechanics theory provides a theoretical basis for determining the reasonable interval. Through mechanical analysis of the curtain wall keel structure, the reasonable relationship between the displacement change rate and the stress change rate under normal working conditions is determined; engineering experience is based on the actual operation of similar projects in the past to correct and improve the theoretical calculation results, thereby determining a pre-set reasonable interval that meets the actual engineering requirements.

[0065] Next, the number of abnormal mechanical events in the pre-set analysis period is counted, denoted as M. The selection of the pre-set analysis period is determined according to the characteristics of the curtain wall keel structure, the use environment, and the actual engineering requirements. In this period, all cases determined to be abnormal mechanical events are counted to obtain the number of abnormal mechanical events M. At the same time, the elastic modulus fluctuation amplitude in the material performance index is obtained, denoted as E. Elastic modulus is an important mechanical property of materials, and its fluctuation amplitude reflects the stability of material performance. Through real-time monitoring and analysis of material performance, the elastic modulus fluctuation amplitude E in the pre-set analysis period is obtained.

[0066] According to the correlation between the number of abnormal mechanical events M and the elastic modulus fluctuation amplitude E, a nonlinear function mapping is used to generate a mechanical abnormality index. Specifically, a coupling function is established with the number of abnormal mechanical events M as the independent variable and the elastic modulus fluctuation amplitude E as the dependent variable. This coupling function describes the internal relationship between M and E. Through analysis and mining of a large amount of historical data, the specific form of the function is determined. In order to obtain an accurate mechanical abnormality index prediction model, machine learning algorithms are used to train the coupling function. Machine learning algorithms have powerful data analysis and pattern recognition capabilities, and can learn the complex relationship between M and E from a large amount of data.

[0067] During the training process, a prediction model is constructed using a neural network structure. The input layer of this neural network includes the normalized values of M and E. Normalization is performed to eliminate the influence of the dimensions and orders of magnitude between different variables, so that the model can treat each input variable more fairly. By mapping the values of M and E to a specific interval, their normalized values are obtained. The output layer of the neural network is the mechanical anomaly index. During the training process of the neural network, the weights and biases of the network are continuously adjusted so that the output of the model can accurately reflect the relationship between M and E as much as possible.

[0068] During the generation of the mechanical anomaly index, the product component and the ratio component of M and E are generated by superposition operation to generate a comprehensive evaluation value. The product component reflects the combined effect of M and E, and the ratio component reflects the relative relationship between them. By superimposing these two components, the final mechanical anomaly index is obtained. This superposition operation method can comprehensively and comprehensively consider the influence of M and E on the mechanical properties of the structure.

[0069] Finally, the generated mechanical anomaly index is compared with the preset safety threshold. The preset safety threshold is determined according to the engineering design standards and actual use requirements, and is used as the limit value to measure whether the mechanical properties of the structure are safe. If the mechanical anomaly index exceeds the preset safety threshold, the design defect flag is triggered. This indicates that in the current working state, the node displacement and stress distribution of the curtain wall keel structure are abnormally matched, and there may be design defects that need to be further analyzed and processed. Through this quantitative evaluation method, potential problems in the curtain wall keel structure can be found in a timely and accurate manner, providing strong support for the optimization design and safety evaluation of the structure. Embodiment

[0070] In the process of generating the structure anomaly index and judging the design defects, a functional and complete historical data storage database must first be established. Taking a building curtain wall project of a large commercial complex as an example, this project includes curtain wall designs in multiple different areas, and the curtain walls in each area differ in geometric parameters, load conditions, and material selection. During the project advancement process, after completing the design of the curtain wall keel structure in each area, a structure optimization benchmark model is constructed according to the method described in the present invention, and the model is comprehensively verified. At this time, various data generated during the verification process, such as node displacement monitoring data, stress distribution calculation results, model parameter setting conditions, etc., are stored in the historical data storage database in a complete and accurate manner according to a unified data format and standard. The storage architecture of the database adopts a hierarchical classification method, and is classified and stored according to information such as project name, design stage, and curtain wall area, facilitating subsequent data retrieval and calling.

[0071] When the new design of the curtain wall structure needs to be judged for design defects, the verification results of the structure optimization benchmark model in the historical design data of the same building curtain wall are extracted from the historical data storage database. Assuming that the current design is a newly added floor curtain wall of the commercial complex, the verification results of the structure optimization benchmark model of the curtain wall design of each floor and each region of the building in the past are retrieved from the database. These results cover the analysis and verification data of the curtain wall structure by different design teams at different times, providing rich reference for subsequent analysis.

[0072] Next, the deviation values of each benchmark model are weighted and accumulated to generate a structure abnormality index. The determination of the deviation value is based on the comparison and analysis of the newly designed structure optimization benchmark model and the historical benchmark model. Taking the connection node design of the curtain wall as an example, the stress distribution of a certain type of connection node in the new design is different from the stress distribution of the same type of node in the historical design under the same load condition. This difference is part of the deviation value. For different structural parameters, such as the cross-sectional size of the gusset and the elastic modulus value of the material, the deviation value is calculated according to the corresponding comparison method.

[0073] In the process of weighted accumulation, the weight of each deviation value is determined according to the importance of its influence on the structure performance. For example, the cross-sectional size of the gusset directly affects the load-carrying capacity and stiffness of the structure, so it has a greater impact on the structure performance. Therefore, when calculating the structure abnormality index, a higher weight is given to the deviation value of the gusset cross-sectional size. The size deviation of some secondary decorative components has a relatively small impact on the structure performance, so a lower weight is given. The determination of the weight requires a professional team composed of structural engineers, material specialists, etc. to determine each structural parameter after in-depth analysis and evaluation based on engineering experience, structural mechanics principles, and relevant design specifications.

[0074] Taking the curtain wall design of the commercial complex as an example, when analyzing the curtain wall design of a newly added floor, the deviation values of multiple structural parameters are calculated by comparing historical data, and are accumulated according to the weights. Assuming that the calculated structure abnormality index is a specific numerical value. At this time, the numerical value is compared with the preset trigger number. The preset trigger number is set in combination with the actual engineering situation and design standards. For this commercial complex, which has high requirements for building safety, the preset trigger number may be set to 3 times in a row. If the structure abnormality index exceeds the preset trigger number continuously, it is determined that the curtain wall structure design has defects.

[0075] In practical applications, there may be a situation where the structural abnormality index calculated for the first time exceeds the preset value, but the second calculation is within the normal range, and the third calculation again exceeds. In this case, since the preset trigger number is not continuously exceeded, it is temporarily determined that there is no design defect, and the design continues to be monitored and analyzed. Only when the structural abnormality index continuously reaches or exceeds the preset trigger number, it is determined that the current curtain wall keel structure design has defects, so that appropriate measures are taken to optimize and improve in time, to ensure the safety and reliability of the building curtain wall. Embodiments

[0076] When the system determines that the curtain wall keel structure design has defects according to the structural abnormality index, it enters the process of automatically generating an optimized design scheme. First, a multi-objective optimization function is constructed, which focuses on three core objectives of minimizing structural weight, maximizing stiffness, and minimizing cost. In practical operation, minimizing structural weight aims to reduce the amount of material used and reduce the overall load burden of the building. Taking a large commercial building curtain wall as an example, the keel structure is usually composed of a large amount of metal material. By reasonably optimizing the cross-sectional shape, size, and layout of the keel, the amount of material used is minimized under the premise of meeting the structural safety requirements, thereby achieving the reduction of structural weight.

[0077] Maximizing stiffness is the key to ensuring the stability of the curtain wall under various load actions. Under different environmental conditions, such as strong winds and temperature changes, the curtain wall needs to have sufficient stiffness to resist deformation. For example, in the design of a high-rise building curtain wall, the wind pressure load generated by strong winds will exert a large force on the curtain wall keel structure. At this time, by optimizing the connection method of the keel, increasing the support structure, etc., the overall stiffness of the structure can be effectively improved, ensuring that the curtain wall remains stable in extreme weather.

[0078] Cost minimization considers multiple aspects such as material cost, processing cost, and installation cost. In terms of material selection, in addition to considering the performance indicators of materials, the prices of materials provided by different suppliers are also compared, and materials with high cost performance are selected under the premise of ensuring quality. In terms of processing cost, by optimizing the processing technology, unnecessary processing procedures are reduced, and production efficiency is improved, thereby reducing processing cost. Installation cost is closely related to the design of the keel structure. Reasonably simplifying the installation process of the keel structure can effectively reduce the installation difficulty and time, thereby reducing the installation cost. The multi-objective optimization function comprehensively considers these three objectives and balances the relationship between each objective to achieve the optimal solution of the curtain wall keel structure in terms of performance and economy.

[0079] Genetic algorithm is used to solve multiple schemes iteratively. Genetic algorithm simulates the process of biological evolution. First, a set of initial curtain wall keel structure optimization design schemes are generated, which are like individuals in biological populations. Each scheme includes a series of design parameters such as keel section size, connection node form, material type, etc. Next, selection operation is performed. According to the performance of each scheme in the multi-objective optimization function, the scheme with higher fitness is selected, just like the biological individuals that are more suitable for the environment are more likely to survive and reproduce. The evaluation of fitness is mainly based on the comprehensive performance of each scheme in the three objectives of structural weight, stiffness and cost. The better the performance, the higher the probability of being selected.

[0080] The crossover operation simulates the reproduction process of organisms, and exchanges and combines part of the design parameters of the selected two schemes to generate new schemes. For example, one scheme is reasonable in keel section size design, and the other scheme has advantages in connection node form. Through crossover operation, the advantages of the two schemes can be combined to form a new scheme with different characteristics. Mutation operation is a random and small change to part of the design parameters in the scheme, which increases the diversity of the scheme and avoids the algorithm falling into local optimal solution, just like the gene mutation that occurs in the evolution process of organisms, which may produce new characteristics that are more suitable for the environment.

[0081] Through repeated selection, crossover and mutation operations, multiple iterations are performed, and each iteration generates a new generation of design schemes. As the iteration proceeds, the performance of the schemes in the multi-objective optimization function gradually improves, i.e. the structural weight continuously decreases, the stiffness continuously increases, and the cost continuously decreases. The final Pareto optimal solution set contains multiple schemes that balance different objectives. These schemes do not have absolute advantages and disadvantages. For example, a scheme may have the lightest structure weight, but the cost is relatively high; another scheme may have the largest stiffness, but the material usage is more.

[0082] After obtaining the Pareto optimal solution set, the optimization design schemes are prioritized according to the size of the structural abnormality index. The structural abnormality index reflects the severity of design defects. The larger the index, the more serious the impact of design defects on structural performance, and the corresponding optimization design scheme needs to be given priority for implementation. Taking a curtain wall project as an example, if a scheme proposes effective optimization measures for design defects that seriously affect the safety of the structure, and the corresponding structural abnormality index is large, then it will be in an early position in the priority ranking. In this way, scientific and reasonable scheme selection can be provided for the optimization of curtain wall keel structure design, ensuring that the scheme that best solves the design defect problem is implemented first in limited resources and time, improving the safety and economy of the curtain wall keel structure.

[0083] It is to be understood that the terminology used herein such as first and second, and the like, is only used to distinguish one entity or action from another entity or action, and does not necessarily require or imply any such actual relationship or order between such entities or actions. Moreover, the terms "comprises", "comprising", or any other variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can include other elements not expressly listed or inherent to such process, method, article, or apparatus.

[0084] While embodiments of the present application have been shown and described with reference to particular embodiments thereof, it will be understood by those skilled in the art that various changes in form and details can be made therein without departing from the spirit and scope of the application. The scope of the application is defined by the appended claims and their equivalents.

Claims

1. A method for intelligent optimization design of a curtain wall keel structure, characterized in that, The method comprises the following steps: S1: Obtain the geometric parameters, load distribution data and material performance indicators of the building curtain wall, and perform data verification on the geometric parameters, load distribution data and material performance indicators; S2: Analyze the verified data to generate multi-dimensional structure characteristic parameters, select features according to the multi-dimensional structure characteristic parameters to form a feature vector set, and construct a structure optimization benchmark model according to the feature vector set; S3: Real-time verify the keel structure data according to the structure optimization benchmark model, identify the topological correlation of node displacement and stress distribution by using multi-level data correlation technology, and quantitatively evaluate the abnormal matching of node displacement and stress distribution; S4: Generate a structure abnormality index according to the verification result of the structure optimization benchmark model, and determine whether there is a design defect according to the structure abnormality index; In S3, the quantitative evaluation of the abnormal matching of node displacement and stress distribution comprises the following steps: S301: Real-time monitor the node displacement change and the stress change at the corresponding position, calculate the difference between the two change rates, and if the difference exceeds the preset reasonable interval, determine that it is an abnormal mechanical event; S302: Count the number of abnormal mechanical events in a preset analysis period, denoted as M, and simultaneously obtain the elastic modulus fluctuation amplitude in the material performance indicators, denoted as E; S303: According to the correlation between the number of abnormal mechanical events M and the elastic modulus fluctuation amplitude E, a nonlinear function is used to map to generate a mechanical abnormality index, wherein the product component and the ratio component of M and E are generated by superposition operation to generate a comprehensive evaluation value; If the mechanical abnormality index exceeds the preset safety threshold, a design defect flag is triggered; In S4, the verification results of the structure optimization benchmark model in the historical design data of the same building curtain wall are extracted, the deviation values of each benchmark model are weighted and accumulated to generate a structure abnormality index, and if the structure abnormality index continuously exceeds the preset trigger number, it is determined that there is a design defect.

2. The method of claim 1, wherein: The geometric parameters include panel size deviation and connector positioning error; the load distribution data include wind pressure gradient change and temperature stress distribution; the material performance indicators include elastic modulus discrete value and yield strength fluctuation range; the data verification on the geometric parameters, load distribution data and material performance indicators comprises data normalization, abnormal value correction, dimension unification and integrity test; the geometric parameter verification generates a geometric feature sequence, and the load distribution data verification generates a load feature sequence; the geometric feature sequence and the load feature sequence of different dimensions are spatially associated to form a multi-source structure feature set.

3. The method of claim 2, wherein: In S2, the following steps are included: S201: Extract spatial feature parameters from the verified geometric feature sequence, load feature sequence and material performance indicators, respectively, the spatial feature parameters include stiffness variation coefficient, connection point offset and stress concentration degree, and combine each spatial feature parameter according to the preset weight to generate multi-dimensional structure characteristic parameters; S202: Set a dynamic threshold for the multi-dimensional structure characteristic parameters in the multi-source structure feature set, select the feature parameters that meet the threshold range to form an initial feature vector set, and exclude the feature parameters that exceed the threshold range; S203: cross-correlation analysis is performed on each feature parameter in the initial feature vector set, the deviation degree of the feature parameters in different working conditions under the same load condition is extracted, the variation coefficient of the deviation degree is calculated and marked as the inter-working condition difference parameter; S204: comparing the inter-working condition difference parameter with the preset tolerance threshold, screening the feature parameters exceeding the tolerance threshold to join the structure optimization benchmark model, and supplementing the yield strength mutation characteristics to the structure optimization benchmark model according to the material performance index.

4. The method of claim 3, wherein: In S202, the setting method of the dynamic threshold includes: based on the statistical distribution law of each feature parameter in the historical engineering database, the confidence interval of the feature parameter is calculated by using the probability density function, the upper limit and the lower limit of the confidence interval are taken as the initial value of the dynamic threshold, and the threshold range is dynamically adjusted according to the feature parameter distribution of the current project.

5. The method of claim 1, wherein: In S303, the specific implementation of the nonlinear function mapping is: a coupling function is established with the number of abnormal mechanical events M as the independent variable and the elastic modulus fluctuation amplitude E as the dependent variable, a prediction model of the mechanical abnormality index is obtained by machine learning algorithm training, the prediction model adopts a neural network structure, the input layer includes the normalized values of M and E, and the output layer is the mechanical abnormality index.

6. The method of claim 1, wherein: It also includes S5: when it is determined that there is a design defect, an optimized design scheme is automatically generated, the optimized design scheme includes the adjustment suggestion of the keel section size, the reinforcement measures of the connecting node and the material replacement scheme, and the priority of the optimized design scheme is sorted according to the size of the structure abnormality index.

7. The method of claim 6, wherein: In S5, the generation method of the optimized design scheme includes: constructing a multi-objective optimization function, taking the minimization of the structure weight, the maximization of the stiffness and the minimization of the cost as the optimization objectives, using genetic algorithm for multi-scheme iterative solution, and outputting the Pareto optimal solution set as the candidate optimized design scheme.

Images