Prediction method of axial compression bearing capacity of concrete-filled stainless steel tube based on extreme gradient algorithm

Abstract

The present invention discloses a method for predicting the axial compressive bearing capacity of stainless steel tube concrete based on an extreme gradient algorithm, comprising the following steps: based on the mechanical mechanism analysis of stainless steel tube concrete compression components, using the test and numerical simulation results to create data related to the compressive bearing capacity of circular and square section components, using ten-fold cross validation to divide the data set into a training set and a test set; using the training set, respectively establish extreme gradient boosting regression models for the compressive bearing capacity of circular and square sections, and use the data set to predict and iterate the model until the network meets the convergence condition. The test set is input into the regression model of the compressive bearing capacity to obtain the predicted value of the compressive bearing capacity; finally, the prediction accuracy of the regression model is analyzed using an error evaluation index, and by comparing with the actual bearing capacity test value and the existing relevant specification calculation value, it is shown that this method can calculate the bearing capacity of stainless steel tube concrete faster and more accurately, and the prediction result is closer to the actual situation.

Description

Technical Field

[0001] The invention relates to the technical field of stainless steel compression detection, in particular to a method for predicting the axial compression bearing capacity of stainless steel tube concrete based on an extreme gradient algorithm. Background Art

[0002] Stainless steel has excellent corrosion resistance, durability, fire resistance and fatigue resistance due to the addition of no less than 10.5% chromium. Stainless steel tube concrete is formed by pouring concrete into stainless steel pipes, which can combine the superior mechanical properties of steel tube concrete and the superior durability of stainless steel. Therefore, it has broad application prospects in marine and offshore infrastructure. Stainless steel tube concrete has been used as the main pressure-bearing structure in many large-scale projects. However, there are currently no design regulations and design clauses for stainless steel tube concrete, and the prediction method for its compressive bearing capacity is not clear and unified. This is mainly because stainless steel materials have significant nonlinear characteristics and their plastic properties far exceed those of ordinary carbon steel. Therefore, the constraint effect of stainless steel pipes on the core concrete is complex and significantly different from ordinary steel tube concrete structures. At the same time, there are many factors that affect the bearing capacity of stainless steel tube concrete structures, and the correlation is relatively complex. Therefore, providing a scientific and reasonable prediction method for the compressive bearing capacity of stainless steel tube concrete structures is very critical for its safe application in actual engineering. Summary of the invention

[0003] In view of this, the purpose of the present invention is to provide a method for predicting the axial compressive bearing capacity of stainless steel tube concrete based on an extreme gradient algorithm. This method can accurately, stably and efficiently predict the ultimate bearing capacity of stainless steel tube concrete structures under compressive loads, and provide engineers with technologies and methods that can scientifically and quickly predict the performance of stainless steel tube concrete structures.

[0004] To achieve the above object, the present invention adopts the following technical solution: a method for predicting the axial compressive bearing capacity of stainless steel tube concrete based on an extreme gradient algorithm, comprising the following steps:

[0005] Step 1: Obtain the existing test data of the compressive bearing capacity of stainless steel tube concrete, and according to the bearing capacity calculation formula of carbon steel tube concrete, combine the mechanical mechanism analysis to condense all the parameters related to the ultimate compressive bearing capacity of stainless steel tube concrete components, and obtain the specific expression of the characteristic influencing parameters;

[0006] Step 2: Establish a finite element model of the stainless steel tube concrete compression member, and conduct a systematic parameter analysis to obtain numerical simulation data under different parameters;

[0007] Step 3: Fully integrate the finite element data obtained in step 2 and the test data set in step 1 to form a database with comprehensive coverage, reasonable parameter distribution and good generalization ability; take the characteristic influencing parameters as input parameters and the compressive ultimate bearing capacity as output parameters to establish the compressive bearing capacity data set of circular cross-section stainless steel tube concrete and the compressive bearing capacity data set of square cross-section stainless steel tube concrete respectively;

[0008] Step 4: The Pearson correlation analysis is performed on the compressive bearing capacity data set of stainless steel tube concrete to determine the correlation coefficient between the characteristic influencing parameters and the compressive bearing capacity; the data set is randomly divided into a training set and a prediction set in a ratio of 8:2 by using a ten-fold cross validation method, and the model is trained with the divided training set to establish an extreme gradient boosting model for predicting the compressive bearing capacity of stainless steel tube concrete;

[0009] Step 5: Use the prediction set to input the extreme gradient boosting model in step 4 for prediction iteration until the network meets the convergence conditions and the prediction accuracy of the compressive bearing capacity meets the requirements; finally, output the ultimate compressive bearing capacity of the stainless steel tube concrete to complete the prediction.

[0010] In a preferred embodiment, the important characteristic influencing parameters affecting the compressive bearing capacity of stainless steel pipe concrete in step 1 are the circular section diameter D or the square section width B, the wall thickness t, the component length l, the stainless steel elastic modulus E, the section gyration radius i, the standard value of the concrete axial compressive strength f ck , nominal yield strength σ 0.2 , constraint effect coefficient ξ, regularized slenderness ratio Slenderness ratio of the component λ, diameter-to-thickness ratio of a circular section D / t or width-to-thickness ratio of a square section B / t, ratio of compressive strength to yield strength f ck / σ 0.2 and the dimensionless yield strength σ 0.2 / E;

[0011] The constraint effect coefficient ξ is expressed as follows according to the Fujian Province Technical Code for Concrete-Filled Steel Tube Structures:

[0012]

[0013] The regularized slenderness ratio According to the technical specifications of China's steel tube concrete structure, the expression is:

[0014]

[0015] In a preferred embodiment: the data set constructed in step 3 is θ=[θ1,θ2,…,θ N ]=[(X1,Y1,Z1),(X2,Y2,Z2),…,(X N,Y N ,Z N )], where (X, Y, Z…) refers to the input parameter set of each group of components, and the training set and prediction set are randomly generated.

[0016] In a preferred embodiment: in step 4, the extreme gradient boosting model is trained using a randomly generated training set to obtain the desired prediction model.

[0017] In a preferred embodiment: the relevant parameter information is input into the extreme gradient boosting regression model, the output information of the decision tree in the extreme gradient regression model is accumulated, and the bearing capacity of the stainless steel tube concrete axial compression member corresponding to the prediction set is determined;

[0018] Wherein, the preset accumulation formula is as follows:

[0019]

[0020] Among them, k represents the number of decision trees in the extreme gradient boosting regression model. represents the set of all decision trees CART, f k (x i ) represents the output information of the decision tree;

[0021] Among them, the extreme gradient boosting model uses several decision trees as learning units, fits the next decision tree according to the residual between the output result of the previous decision tree and the actual value, and obtains the predicted value of the compressive bearing capacity of stainless steel tube concrete by summing the output results of several decision trees.

[0022] In a preferred embodiment: the method for predicting the compressive bearing capacity of stainless steel tube concrete based on the extreme gradient boosting algorithm also includes the following steps: judging whether the accuracy of the prediction result of the extreme gradient boosting model reaches the set accuracy, and if so, outputting the predicted value of the compressive bearing capacity of the stainless steel tube concrete; otherwise, modifying the number of decision trees of the extreme gradient boosting model and re-obtaining the predicted value of the compressive bearing capacity of the stainless steel tube concrete.

[0023] The present invention also provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the method for predicting the axial compressive bearing capacity of stainless steel tube concrete based on the extreme gradient algorithm as described above is implemented.

[0024] The present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable by the processor. When the processor executes the computer program, it implements the method for predicting the axial compressive bearing capacity of stainless steel tube concrete based on the extreme gradient algorithm as described above.

[0025] Compared with the prior art, the present invention has the following beneficial effects: effectively improving the prediction accuracy, and making the prediction parameter range fully cover the extreme situations in actual engineering, which is of great significance for accurately predicting the performance of stainless steel tube concrete engineering, ensuring its structural safety and avoiding major engineering accidents. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 A flowchart of a preferred embodiment of the present invention;

[0027] Figure 2 Comparison of stress-strain curves of stainless steel and ordinary steel in finite element modeling of a preferred embodiment of the present invention;

[0028] Figure 3 This is a schematic diagram of the initial defect distribution of a stainless steel tube concrete component in a preferred embodiment of the present invention;

[0029] Figure 4 The stress-strain relationship curve of the core concrete in the finite element model in the preferred embodiment of the present invention;

[0030] Figure 5 It is a schematic diagram of a finite element model of a stainless steel tube concrete compression member in a preferred embodiment of the present invention;

[0031] Figure 6 The typical finite element parameter analysis results of the stainless steel tube concrete compressive member in the preferred embodiment of the present invention, wherein (1a) is the influence of different ξ of circular stainless steel tube concrete on the compressive bearing capacity-axial strain relationship curve of stainless steel tube concrete, wherein (1b) is the influence of different ξ of square stainless steel tube concrete (D / B=1) on the compressive bearing capacity-axial strain relationship curve of stainless steel tube concrete, wherein (1c) is the influence of different ξ of rectangular stainless steel tube concrete (D / B=1.5) on the compressive bearing capacity-axial strain relationship curve of stainless steel tube concrete, wherein (1d) is the influence of different ξ of rectangular stainless steel tube concrete (D / B=2) on the compressive bearing capacity-axial strain relationship curve of stainless steel tube concrete, wherein (2a) is the influence of σ0.2 of the circular member on the compressive stability coefficient φ-slenderness ratio λ relationship curve of stainless steel tube concrete, wherein (2b) is the influence of σ0.2 of the square member on the compressive stability coefficient φ-slenderness ratio λ relationship curve of stainless steel tube concrete, wherein (3a) is the f of the circular member. cu The influence of the compressive stability coefficient φ of stainless steel tube concrete and the slenderness ratio λ, where (3b) is the f of the square member cu The influence of the compressive stability coefficient φ of stainless steel tube concrete and the slenderness ratio λ, where (4a) is the α of the circular component sThe influence of the compressive stability coefficient φ of stainless steel tube concrete and the slenderness ratio λ, where (4b) is the α of the square member s The influence on the relationship curve between the compressive stability coefficient φ and the slenderness ratio λ of stainless steel tube concrete;

[0032] Figure 7 It is a specific procedure of a cross-validation method for predicting the extreme gradient improvement of the compressive bearing capacity of stainless steel tube concrete in a preferred embodiment of the present invention;

[0033] Figure 8 RMSE learning curve of the extreme gradient boosting model in each period on the training and test data sets in the preferred embodiment of the present invention;

[0034] Fig. 9 It is a line chart of predicted values ​​of partial sample data of bearing capacity of stainless steel tube concrete based on extreme gradient boosting algorithm in a preferred embodiment of the present invention;

[0035] Fig.10 This is a discrete distribution diagram of the bearing capacity prediction results and standard calculated values ​​of stainless steel tube concrete based on the extreme gradient boosting algorithm in a preferred embodiment of the present invention, where (a) is a square screenshot and (b) is a circular screenshot. DETAILED DESCRIPTION

[0036] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0037] It should be noted that the following detailed descriptions are illustrative and are intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present application belongs.

[0038] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application; as used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, they indicate the presence of features, steps, operations, devices, components and / or their combinations.

[0039] The prediction method of axial compressive bearing capacity of stainless steel tube concrete based on extreme gradient algorithm is as follows: Figure 1-10 , including the following steps:

[0040] Step 1: Obtain existing test data on the compressive bearing capacity of stainless steel tube concrete from previous literature. Refer to the bearing capacity calculation formula of ordinary carbon steel tube concrete provided in the existing relevant design specifications, combine with mechanical mechanism analysis, condense all parameters related to the ultimate compressive bearing capacity of stainless steel tube concrete components, and obtain specific expressions of characteristic influencing parameters.

[0041] Step 2: Establish a finite element model of the stainless steel tube concrete compression member and conduct a systematic parameter analysis to obtain numerical simulation data under different parameters. On the one hand, the finite element simulation data can verify whether the parameters determined by the test data are comprehensive and accurate. On the other hand, it can also effectively expand the parameter range of the database and make the parameter distribution more uniform and cover a more comprehensive range, thereby improving the generalization ability of the database;

[0042] Step 3: Fully integrate the finite element data obtained in step 2 and the test data set in step 1 to form a database with comprehensive coverage, reasonable parameter distribution and good generalization ability. Take the characteristic influencing parameters as input parameters and the compressive ultimate bearing capacity as output parameters to establish the compressive bearing capacity data set of circular cross-section stainless steel tube concrete and the compressive bearing capacity data set of square cross-section stainless steel tube concrete respectively;

[0043] Step 4: The Pearson correlation analysis is performed on the compressive bearing capacity data set of stainless steel tube concrete to determine the correlation coefficient between the characteristic influencing parameters and the compressive bearing capacity; the data set is randomly divided into a training set and a prediction set in a ratio of 8:2 by using a ten-fold cross validation method, and the model is trained with the divided training set to establish an extreme gradient boosting model for predicting the compressive bearing capacity of stainless steel tube concrete;

[0044] Step 5: Use the prediction set to input the extreme gradient boosting model in step 4 for prediction iteration until the network meets the convergence conditions and the prediction accuracy of the compressive bearing capacity meets the requirements. Finally, the ultimate compressive bearing capacity of stainless steel tube concrete is output to complete the prediction.

[0045] The compressive bearing capacity test data of stainless steel tube concrete collected in the step 1 is derived from previously publicly reported test literature at home and abroad; the referenced existing relevant design specifications are specifically derived from the four commonly used ordinary carbon steel tube concrete specifications: American Standard (ANSI / AISC 360-16), European Standard EC4 (2004) Specification, China Technical Specification for Steel Tube Concrete Structure (GB 50936-2014) and Fujian Province Technical Code for Steel Tube Concrete Structure (DBJ / T13-51-2020).

[0046] Furthermore, the characteristic parameters affecting the compressive bearing capacity of the stainless steel pipe concrete in step 1 are: circular section diameter D, square section width B, stainless steel pipe wall thickness t, component length l, stainless steel elastic modulus E, section gyration radius i, concrete axial compressive strength standard value f ck , Nominal yield strength of stainless steel σ 0.2 , constraint effect coefficient ξ, regularized slenderness ratio Slenderness ratio of the member λ, diameter-to-thickness ratio of the circular section D / t (width-to-thickness ratio of the square section B / t), ratio of the concrete compressive strength to the nominal yield strength of the stainless steel f ck / σ 0.2 , dimensionless nominal yield strength σ 0.2 / E;

[0047] The constraint effect coefficient ξ mentioned therein is modified to the expression of ordinary carbon steel tube concrete in the Technical Code for Steel Tube Concrete Structures of Fujian Province (DBJ / T13-51-2020) as follows:

[0048]

[0049] The regularized slenderness ratio The expression is:

[0050]

[0051] In step 2, since the previous test work on stainless steel tube concrete is still insufficient, the test data on the compressive bearing capacity of stainless steel tube concrete that can be collected are relatively small, and the parameter distribution is uneven, and it is difficult to cover the parameter range required for actual engineering. If only these data are used to train the extreme gradient boosting model, the training results will be inaccurate and the generalization performance will be poor due to the small number of database samples, uneven distribution, and comprehensive parameters. Therefore, the present invention uses a finite element model that has been verified by experiments to perform extended parameter analysis on the database samples to obtain a more complete and applicable data set, and make the data distribution more reasonable, so as to provide a guarantee for the prediction accuracy of the extreme gradient boosting model, and is conducive to training a prediction model that better meets the needs of actual engineering.

[0052] 1) Establishment of a stainless steel tube concrete compression database. The basic parameters of the stainless steel tube concrete compression bearing capacity test data obtained in previous literature are collected and sorted out to construct a stainless steel tube concrete compression test data set. The embodiment of the present invention collects a total of 189 stainless steel tube concrete compression bearing capacity test data, and the basic parameter range of the test data is shown in Table 1.

[0053] Table 1 Basic parameter range of test data

[0054]

[0055] Due to the high cost of physical experiments on stainless steel tube concrete, there are currently few experimental data that have been completed. If only these data are used to train the extreme gradient lifting model, the training results will be inaccurate due to insufficient database samples, and the model generalization performance will be poor. Therefore, the present invention uses a finite element model that has been verified by experiments to perform extended parameter analysis on database samples to obtain a more complete and applicable data set, and make the data distribution more reasonable, which provides a guarantee for the prediction accuracy of the extreme gradient lifting model and is conducive to training a prediction model that better meets the needs of actual engineering.

[0056] The specific steps to establish the finite element model of stainless steel tube concrete under compression are:

[0057] (1) The stainless steel pipe adopts a four-node fully integrated shell element (S4). To meet certain calculation accuracy, Simpson integration with 9 integration points is used in the thickness direction of the shell element. The cover plate and core concrete both adopt a three-dimensional solid element (C3D8R) with an eight-node reduced integration format. The finite element model is divided into units using the structured meshing technology provided by the finite element software.

[0058] (2) The material constitutive model of concrete takes into account the restraint effect of the stainless steel pipe, especially the strong restraint effect that the superior plastic properties of stainless steel can still maintain on the core concrete in the later stage of loading. The finite element model of the stainless steel component and the stainless steel pipe concrete component involved in the present invention does not consider the influence of residual stress.

[0059] Due to the addition of alloy elements, the stress-strain relationship of stainless steel is different from that of ordinary carbon steel. The stress-strain relationship curve of ordinary carbon steel has an obvious yield point and a yield platform; while the stress-strain relationship curve of stainless steel shows significant nonlinear characteristics at the beginning of loading, and the proportional limit is low. The stress-strain curves of stainless steel and ordinary steel are compared as follows: Figure 2 The stress-strain relationship model expression of the stainless steel material in the present invention is as follows:

[0060] When σ≤σ 0.2 hour:

[0061]

[0062] When σ>σ 0.2 hour:

[0063]

[0064] Where:

[0065]

[0066] where σ 0.01For the residual strain of 0.01% stress, the unified parameter values ​​of the commonly used stainless steel types in building structures are adopted, that is, for austenitic stainless steel, n = 5.4 when under tension and n = 4.3 when under compression.

[0067] The stainless steel pipe in the finite element model adopts an isotropic elastic-plastic model to meet the VonMises yield criterion. Since the model requires the input of the true stress-true plastic strain relationship of the material, it is necessary to convert the σ-ε relationship calculated by equations (3) and (4) into the true stress-true plastic strain relationship, that is, σ true =σ(1+ε),ε p1 true =ln(1+ε)-σ true / E s , where σ true and ε p1 true are true stress and true plastic strain, respectively.

[0068] The stainless steel pipe of the present invention takes initial defects into consideration, that is, elastic buckling mode analysis is first performed on the model before nonlinear analysis, and then the first-order buckling mode is input as the initial defect of the component during nonlinear analysis. Figure 3 , which can be expressed by the following expression:

[0069]

[0070] Where a is the plate width; L is the plate length; m and n are the plate buckling half-wave numbers in the x and y directions, respectively. For example, when the length-to-width ratio of the square stainless steel tube concrete wall is 1:3, m and n can be taken as 1 and 3, respectively.

[0071] For stainless steel pipe

[0072]

[0073] in, is the elastic critical buckling stress.

[0074] The core concrete of the present invention adopts a plastic damage model, and its stress-strain relationship model under compression adopts a core concrete model of steel pipe constraint effect, and its stress-strain relationship curve is as follows: Figure 4 , which is expressed as follows:

[0075]

[0076] Where: σ o =f c ';ε o =ε c +800×ξ 0.2 ×10 -6 ; εc =(1300+12.5f c ')×10 -6 ;

[0077]

[0078] In the present invention, the parameters of the concrete elastic stage in the plastic damage model are determined according to ACI318-05 (2005), and the elastic modulus of the core concrete is taken as E c =4700(f c ′) 1 / 2 , where f c ' is the compressive strength of the concrete cylinder, in N / mm 2 The unit is , and the Poisson's ratio of concrete in the elastic stage is 0.2.

[0079] For the constitutive relation of tensile concrete, the tensile softening property of concrete, i.e. stress-fracture energy relation, is defined by energy failure criterion. The tensile property of concrete is defined by fracture energy (G f ) description, for concrete grade C20, G f Take 40N / m, for concrete grade C40, G f Take 120N / m, and other levels in between are calculated using linear interpolation. The concrete tensile calculation formula used to calculate the concrete cracking stress is:

[0080] σ to =0.26×(1.25f c ') 2 / 3 (6)

[0081] (3) The interface model between the stainless steel tube and the core concrete consists of bond slip in the tangential direction and contact in the normal direction of the interface. The normal contact model between the stainless steel tube and the core concrete is simulated using "hard contact", that is, the pressure perpendicular to the cross section can be completely transmitted between the interfaces, allowing the stainless steel tube and concrete to separate during the loading process. The expression is as follows:

[0082]

[0083] Among them, F N is the unit contact force; σ is the unit contact compressive stress; A is the unit area normal to the unit compressive stress.

[0084] The tangential contact model adopts the Coulomb friction model. The interface can transmit shear stress until the shear stress reaches a critical value and relative sliding occurs between the interfaces. The interface friction coefficient μ is 0.25.

[0085] (4) The component adopts the boundary condition of one end fixed and the other end free. When modeling, the upper and lower covers are coupled with the reference points (RP1 and RP2) respectively, and then the boundary and loading conditions are set for the reference points, that is, RP1 constrains the displacement in the X, Y and Z axis directions, and RP2 constrains the displacement in the X and Y axis directions, while both are allowed to rotate. Displacement loading is used in the Z axis direction of RP2 during calculation.

[0086] (5) Before meshing, the mesh density that meets the accuracy must be determined first, and then the finite element model is divided into units using the structured meshing technology provided by the finite element software. The stiffness matrix of each unit is assembled into a total stiffness matrix, and then the displacement matrix and load matrix are combined. The Newton-Raphson method is used to solve the nonlinear equations to obtain the mechanical response of the stainless steel tube concrete component under compression.

[0087] (6) Using the finite element model, a systematic parametric analysis of the mechanical behavior of the component under varying parameters is conducted. The range of parameter analysis is: nominal yield strength of stainless steel (180-900 MPa), concrete cube compressive strength (30-150 MPa), cross-sectional steel content (0.02-2), slenderness ratio (10-200), etc.

[0088] The schematic diagram of the finite element axial compression model of stainless steel tube concrete is as follows: Figure 5 shown.

[0089] The test data are used to verify the reliability and rationality of the finite element model. The test verification results of the finite element simulation of the compressive bearing capacity of stainless steel tube concrete are shown in Table 2 and Table 3.

[0090] Table 2 Circular section finite element simulation verification results

[0091]

[0092]

[0093]

[0094] Table 3. Square section finite element simulation verification results

[0095]

[0096]

[0097] The finite element method established above has been verified to simulate the actual working conditions well. A total of 1200 sets of simulation data were calculated, and the parameter range is: f cu =30~150MPa,σ 0.2 =180~900MPa、λ=10~200、α s=0.02~2.0. Typical finite element parameter analysis results are as follows Figure 6 The finite element model was used to expand the parameters of the collected test results, and 1200 sets of finite element simulation results and 189 sets of collected test results were randomly combined to establish the compressive bearing capacity database of stainless steel tube concrete.

[0098] 2) Refer to the definitions and sources of parameters in the compressive bearing capacity calculation formula provided in the existing ordinary steel tube concrete design specifications, and sort out all parameters related to the compressive bearing capacity. Determine the characteristic parameters that affect the bearing capacity of stainless steel tube concrete compressive members: circular section diameter D (square section width B), wall thickness t, member length l, stainless steel elastic modulus E, section gyration radius i, standard value of concrete axial compressive strength f ck , nominal yield strength σ 0.2 In order to achieve more accurate and stable prediction accuracy, seven dimensionless factors are added on the basis of the above influencing factors: constraint effect coefficient ξ, regularized slenderness ratio Slenderness ratio of the component λ, diameter-to-thickness ratio of the circular section D / t (width-to-thickness ratio of the square section B / t), ratio of compressive strength to yield strength f ck / σ 0.2 , dimensionless yield strength σ 0.2 / E;

[0099] The characteristic parameters that affect the bearing capacity of stainless steel tube concrete compression members are used as input parameters, and the ultimate bearing capacity of the members is used as output parameters to construct a data set. The constructed data set is θ=[θ1,θ2,…,θ N ]=[(X1,Y1,Z1),(X2,Y2,Z2),…,(X N ,Y N ,Z N )], where (X, Y, Z…) refers to the set of input parameters for each group of components.

[0100] 3) According to the Pearson correlation coefficient formula, the correlation coefficient between the characteristic parameters affecting the bearing capacity of stainless steel tube concrete compressive components and the compressive bearing capacity is calculated. The Pearson correlation coefficient expression is:

[0101]

[0102] Among them, P ij is the correlation coefficient between the i-th characteristic parameter and the j-th characteristic parameter; X i and X j are the vectors of the i-th feature parameter and the j-th feature parameter respectively; the Pearson correlation coefficient varies in the range of -1 to +1, 0 indicates no correlation, and -1 / +1 indicates an exact negative / positive linear relationship.

[0103] The correlation coefficient values ​​of the characteristic parameters affecting the bearing capacity of stainless steel tube concrete compressive members calculated using the Pearson correlation coefficient formula are listed in Table 4. Since the focus here is on the correlation between the characteristic parameters of the bearing capacity of stainless steel tube concrete compressive members and the compressive bearing capacity, only the absolute values ​​of the Pearson correlation coefficients between each parameter and the bearing capacity are listed.

[0104] Table 4 Pearson correlation coefficient between ultimate bearing capacity and characteristic parameters of stainless steel tube concrete compression members

[0105]

[0106] 4) The data set is divided into a training set and a prediction set at a set ratio of 8:2. At the same time, in order to prevent overfitting, the data set is resampled using a ten-fold cross-validation validation method, and an extreme gradient boosting model is trained using the training set. The specific process of the ten-fold cross-validation method is as follows: Figure 7 As shown: Divide the sample into 10 subsets of equal size, use one subsample as the training set and the remaining subsamples as the test set, and repeat this process several times until all samples have been tested once;

[0107] 5) The extreme gradient boosting model uses the decision tree as the basic learning unit, and uses the ten-fold cross-validation method to iteratively establish the decision tree to form an extreme gradient boosting model. Among them, the stainless steel tube concrete compressive bearing capacity prediction device establishes the first decision tree based on the parameter information and the ultimate bearing capacity information, and calculates the residual between the predicted value and the true value, and establishes the next decision tree based on the residual output by the previous decision tree; within the preset number of iterations, the number of iterations corresponding to the minimum residual is selected as the target number of iterations, and the number of decision trees is determined based on the target number of iterations, thereby forming an extreme gradient boosting regression model. The specific steps are as follows:

[0108] Setting up the dataset The loss function of the learning unit of the extreme gradient boosting model is Let the number of iterations t and the number of decision trees k be traversed in the set data set, verify the accuracy and select the number of iterations t and the number of decision trees k with the highest accuracy as the extreme gradient boosting model parameters; where x i is the feature vector, n is the number of samples in the data set, y i For the actual output result, Outputting results for the extreme gradient boosting model;

[0109] Get the t-th simulated prediction value of the extreme gradient boosting model as follows

[0110] in is the output result of the extreme gradient boosting model t-1 times, f t (x i ) is the output result of the tth iteration of the decision tree;

[0111] Several subsets K are randomly extracted from the training data set with replacement, and a decision tree is generated for each subset. Each decision tree is trained to obtain the residual

[0112] The training loss function is minimized as follows:

[0113]

[0114] Where γ is the regularization term coefficient;

[0115] The prediction results of the several decision trees are summed to obtain the predicted value of the compressive bearing capacity:

[0116]

[0117] Among them, f k is the prediction result of a single decision tree, It is the set of all decision trees CART. By using decision trees to predict the bearing capacity of a single subset, a bearing capacity prediction value with higher accuracy is obtained by averaging the prediction results of the decision trees.

[0118] 6) Use the divided prediction set to input the trained extreme gradient boosting model for prediction. When the prediction accuracy of the compressive bearing capacity meets the requirements, stop and output the bearing capacity of the stainless steel tube concrete compressive member to complete the prediction. If the prediction result does not meet the accuracy requirements, modify the number of decision trees of the extreme gradient boosting model and re-obtain the prediction value of the compressive bearing capacity of the stainless steel tube concrete. Through the iterative optimization process of feedback, improvement of model parameters, retraining, and output of results, the prediction value is made more accurate and stable until the accuracy requirements are met.

[0119] When the extreme gradient boosting model is trained on the training data set and the test data set, it is necessary to evaluate the performance of the model for each iteration of the extreme gradient boosting model and plot the results as a learning curve. The learning curve can be used to evaluate the prediction effect of the extreme gradient boosting model, and then the model can be optimized to improve the prediction performance. The root mean square error RMSE learning curve of the extreme gradient boosting model in each period on the training and test data sets in the embodiment of the present invention is as follows: Figure 8 ; In this example, to prevent the model from overfitting, the early_stopping_rounds=10 parameter is used for the extreme gradient boosting model in the fit phase.

[0120] 7) A line graph of partial prediction sample data and test sample data of the bearing capacity of the stainless steel tube concrete compression member in the embodiment of the present invention is obtained as shown in FIG. Fig. 9 As shown, from Fig. 9 It can be seen that the predicted sample data is very consistent with the test sample data, indicating that the extreme gradient lifting model has a good prediction effect on the bearing capacity of stainless steel tube concrete compressive components; the prediction results of the compressive bearing capacity of stainless steel tube concrete based on the extreme gradient lifting model are shown in Table 5. From the error index of the prediction results in Table 5, it can be seen that the prediction accuracy of the compressive bearing capacity of stainless steel tube concrete based on the extreme gradient lifting model in the embodiment of the present invention is high, and the prediction performance of the extreme gradient lifting model is good.

[0121] Among them, the error indicators are the average value Ave. of the ratio of the predicted value to the test value, the standard deviation Std. of the ratio of the predicted value to the test value, and the root mean square error RMSE between the predicted value and the test value, and their formulas are:

[0122]

[0123] Where n is the total number of test data, χ pi and χ ti They are the predicted value and test value of the bearing capacity (unit: kN), respectively.

[0124] Table 5 Prediction results of compressive bearing capacity of stainless steel tube concrete based on extreme gradient lifting model

[0125]

[0126] 8) The embodiment of the present invention uses the formulas in the four commonly used ordinary carbon steel tube concrete specifications (American Standard (ANSI / AISC 360-16), European Standard EC4 (2004), China Steel Tube Concrete Structure Technical Specification (GB 50936-2014), Fujian Province Steel Tube Concrete Structure Technical Code (DBJ / T13-51-2020)) to obtain the specification formula calculation value corresponding to each sample, and performs error analysis on it and the predicted value of the corresponding sample. The error analysis results are shown in Table 6. The mean value, standard deviation, and variance respectively represent the mean value, standard deviation, and variance of the ratio of each sample prediction value to the corresponding specification calculation value; the coefficient of variation represents the ratio of the standard deviation to the mean value. From the coefficient of variation in Table 6, the prediction accuracy of the extreme gradient boosting model adopted in the embodiment of the present invention is significantly better than the specification design standard; from the mean value, the degree of agreement between the predicted value and the test value of the extreme gradient boosting model adopted in the embodiment of the present invention is higher than the degree of agreement between the calculated value and the test value of the specification formula. This indicates that the extreme gradient boosting model used in the embodiment of the present invention has good prediction performance and can accurately predict the compressive bearing capacity of stainless steel tube concrete components.

[0127] Table 6 Error analysis results of the extreme gradient boosting model prediction values ​​and the four standard formula calculation values

[0128]

[0129]

[0130] Fig.10 The discrete distribution diagrams between the predicted values ​​and the experimental values ​​of the extreme gradient boosting model and the discrete distribution diagrams between the calculated values ​​and the experimental values ​​of the existing specifications are shown respectively. Fig.10 It can be seen that the prediction accuracy and consistency of the extreme gradient lifting model are significantly better than the design specifications, indicating that the extreme gradient lifting model used in the embodiment of the present invention can more accurately predict the compressive bearing capacity of the stainless steel tube concrete component.

[0131] In summary, the present invention provides a method for predicting the extreme gradient improvement of the compressive bearing capacity of stainless steel tube concrete. The calculation results show that this method effectively improves the accuracy, versatility and reliability of the design work, and can calculate the bearing capacity of stainless steel tube concrete faster and more stably. With the increase of test data and the expansion of the test parameter range, a more complete and extensive extreme gradient improvement prediction system can be established to improve the prediction accuracy of the model so that it can be promoted and applied in actual engineering.

[0132] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any one of the above-mentioned methods for predicting the bearing capacity index of stainless steel tube concrete based on the extreme gradient boosting algorithm.

[0133] The present invention also provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable by the processor, wherein when the processor executes the computer program, the steps of the method for predicting the compressive bearing capacity of stainless steel tube concrete based on the extreme gradient boosting algorithm as described in any one of the above are implemented.

[0134] The present invention implements all or part of the processes in the above-mentioned embodiment method, which can be completed by instructing related hardware through a computer program, and the computer program can be stored in a computer-readable storage medium. When the computer program is executed by a processor, the steps of each of the above-mentioned method embodiments can be implemented. The present invention can take the form of a computer program product implemented on one or more storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing program code. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form, etc. Computer-readable storage media include permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. Information can be computer-readable instructions, data structures, modules of programs or other data.

[0135] The above embodiments are only used to illustrate the technical solutions of the present invention, but not to limit them. Although the present invention has been described in detail with reference to the above embodiments, it should be understood by those skilled in the art that other embodiments can still be implemented using the technical solutions described in the above embodiments, or some of the technical features thereof can be replaced by equivalents, and these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting the axial compressive bearing capacity of stainless steel tube concrete based on extreme gradient algorithm, characterized in that: The steps include: Step 1: Obtain the existing test data of the compressive bearing capacity of stainless steel tube concrete, and according to the bearing capacity calculation formula of carbon steel tube concrete, combine the mechanical mechanism analysis to condense all the parameters related to the ultimate compressive bearing capacity of stainless steel tube concrete components, and obtain the specific expression of the characteristic influencing parameters; Step 2: Establish a finite element model of the stainless steel tube concrete compression member, and conduct a systematic parameter analysis to obtain numerical simulation data under different parameters; Step 3: Fully integrate the finite element data obtained in step 2 and the experimental data set in step 1 to form a database with comprehensive coverage, reasonable parameter distribution and good generalization ability; Taking the characteristic influencing parameters as input parameters and the ultimate compressive bearing capacity as output parameters, a data set of compressive bearing capacity of circular section stainless steel tube concrete and a data set of compressive bearing capacity of square section stainless steel tube concrete were established respectively. Step 4: The compressive bearing capacity data set of stainless steel tube concrete is analyzed by Pearson correlation to determine the correlation coefficient between the characteristic influencing parameters and the compressive bearing capacity; The data set was randomly divided into a training set and a prediction set in a ratio of 8:2 using a ten-fold cross validation method, and the model was trained using the divided training set to establish an extreme gradient boosting model for predicting the compressive bearing capacity of stainless steel tube concrete. Step 5: Use the prediction set to input the extreme gradient boosting model in step 4 to perform prediction iterations until the network meets the convergence conditions and the prediction accuracy of the compressive bearing capacity meets the requirements; finally, output the ultimate compressive bearing capacity of the stainless steel tube concrete to complete the prediction; The important characteristic parameters that affect the compressive bearing capacity of stainless steel pipe concrete in step 1 are the circular section diameter D or square section width B, wall thickness t, component length l, stainless steel elastic modulus E, section gyration radius i, and standard value of concrete axial compressive strength f. ck , nominal yield strength σ 0.2 , constraint effect coefficient ξ, regularized slenderness ratio Slenderness ratio of the component λ, diameter-to-thickness ratio of a circular section D / t or width-to-thickness ratio of a square section B / t, ratio of compressive strength to yield strength f ck / σ 0.2 and the dimensionless yield strength σ 0.2 / E; The constraint effect coefficient ξ is expressed as follows according to the Fujian Province Technical Code for Concrete-Filled Steel Tube Structures: The regularized slenderness ratio According to the technical specifications of China's steel tube concrete structure, the expression is:

2. According to the method for predicting the axial compressive bearing capacity of stainless steel tube concrete based on the extreme gradient algorithm in claim 1, it is characterized in that: The data set constructed in step 3 is θ=[θ1,θ2,…,θ N ]=[(X1,Y1,Z1),(X2,Y2,Z2),…,(X N ,Y N ,Z N )], where (X, Y, Z…) refers to the input parameter set of each group of components, and the training set and prediction set are randomly generated.

3. The method for predicting the axial compressive bearing capacity of stainless steel tube concrete based on extreme gradient algorithm according to claim 1 is characterized in that: In step 4, the extreme gradient boosting model is trained using a randomly generated training set to obtain the desired prediction model.

4. The method for predicting the axial compressive bearing capacity of stainless steel tube concrete based on the extreme gradient algorithm according to claim 1 is characterized in that: Input all parameters related to the ultimate compressive bearing capacity of the stainless steel tube concrete member into the extreme gradient boosting model, accumulate the output information of the decision tree in the extreme gradient boosting model, and determine the bearing capacity of the stainless steel tube concrete axial compression member corresponding to the prediction set; Among them, the preset accumulation formula is as follows: Among them, k represents the number of decision trees in the extreme gradient boosting model. represents the set of all decision trees CART, fk(xi) represents the output information of the decision tree; Among them, the extreme gradient boosting model uses several decision trees as learning units, fits the next decision tree according to the residual between the output result of the previous decision tree and the actual value, and obtains the predicted value of the compressive bearing capacity of stainless steel tube concrete by summing the output results of several decision trees.

5. The method for predicting the axial compressive bearing capacity of stainless steel tube concrete based on extreme gradient algorithm according to claim 4 is characterized in that: The method for predicting the compressive bearing capacity of stainless steel tube concrete based on the extreme gradient boosting algorithm also includes the following steps: judging whether the accuracy of the prediction result of the extreme gradient boosting model reaches the set accuracy, and if so, outputting the predicted value of the compressive bearing capacity of the stainless steel tube concrete; otherwise, modifying the number of decision trees of the extreme gradient boosting model and re-obtaining the predicted value of the compressive bearing capacity of the stainless steel tube concrete.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for predicting the axial compressive bearing capacity of stainless steel tube concrete based on the extreme gradient algorithm as described in any one of claims 1 to 5 is implemented.

7. A computer device, characterized in that: The method comprises a memory, a processor and a computer program stored in the memory and executable by the processor. When the processor executes the computer program, the method for predicting the axial compressive bearing capacity of stainless steel tube concrete based on the extreme gradient algorithm as described in any one of claims 1 to 5 is implemented.

Images