Method and system for predicting axial compression bearing capacity of concrete filled steel tube short column with hollow interlayer

Abstract

The invention provides a method and system for predicting the axial compression bearing capacity of a hollow sandwich concrete-filled steel tube short column, and the method comprises the steps: collecting the working parameter data of the hollow sandwich concrete-filled steel tube short column, and carrying out the normalization processing of the data; establishing an LM algorithm theoretical prediction model based on the original data, dividing the normalized data into a training set and a test set, and training the prediction model based on the training set to obtain an optimal hyper-parameter; and after the optimal parameter is determined, evaluating the model performance by using the test set. According to the technical scheme, through a data driving method and in combination with a BP neural network algorithm, accurate prediction of the axial compression bearing capacity of the concrete filled steel tube short column with the hollow interlayer is achieved. The scheme has the specific steps of parameter collection, preprocessing, model establishment and training, performance evaluation and technical effect analysis, and provides beneficial reference for solving similar problems.

Description

Technical Field

[0001] The invention belongs to the field of steel tube concrete, and relates to a method and a system for predicting the axial compressive bearing capacity of a hollow sandwich steel tube concrete short column. Background Art

[0002] The hollow sandwich steel tube concrete short column is an innovative composite structure. The essence of its construction lies in the ingenious combination of the inner and outer layers of steel tubes and the concrete poured between the two layers of steel tubes. This design not only makes full use of the tensile properties of steel and the compressive properties of concrete, but also significantly improves the strength and toughness of concrete through the tight constraints of the steel tube on the inner filling concrete. At the same time, the inner filling concrete acts as a solid support, effectively delaying the tendency of the steel tube to buckle outward when subjected to external forces, thereby ensuring the stability and durability of the entire structure. Therefore, this type of composite column can show excellent strength characteristics, good ductility performance and outstanding energy absorption capacity when bearing loads, becoming one of the most popular structural forms in modern construction projects.

[0003] In the field of structural engineering, bearing capacity is a core indicator for measuring structural performance and safety, and is directly related to the design and evaluation of structures. Traditional structural bearing capacity modeling methods often follow the principle of "mechanism-driven", that is, in-depth analysis of the force mechanism of the structure or component, and on this basis, the calculation formula for bearing capacity is derived. This method focuses on the rigor of theoretical analysis and can more accurately reflect the mechanical behavior of the structure. Another commonly used modeling method is the semi-theoretical and semi-empirical method, which is based on mechanism analysis, combined with a large amount of experimental data and engineering practice, and summarizes empirical formulas that are easy to apply through parameter fitting. Although these methods can achieve good results in many cases, when faced with structural problems with complex force mechanisms and many influencing factors, their numerical implementation process often becomes quite complicated, requiring not only a large number of iterative calculations or solving implicit equations, but also low calculation efficiency and large error accumulation.

[0004] Therefore, exploring more efficient and accurate bearing capacity modeling methods for such complex problems has become an important research direction in the current field of structural engineering. Summary of the invention

[0005] The purpose of the present invention is to provide a method and system for predicting the axial compressive bearing capacity of a hollow sandwich steel tube concrete short column to solve the above-mentioned problem.

[0006] To achieve the above object, the present invention adopts the following technical solutions: In a first aspect, the present invention provides a method for predicting the axial compressive bearing capacity of a hollow sandwich steel tube concrete short column, comprising: Collect the working parameter data of hollow sandwich steel tube concrete short columns and normalize the data; The LM algorithm theoretical prediction model is established based on the original data, the normalized data is divided into a training set and a test set, and the prediction model is trained based on the training set to obtain the optimal hyperparameters; After determining the optimal parameters, the model performance is evaluated using the test set.

[0007] Optionally, collect working parameter data of hollow sandwich steel tube concrete short columns, including: Axial compression bearing capacity N u , outer steel pipe diameter D o and thickness t o , Inner steel pipe diameter D i and thickness t i , concrete compressive strength f c , yield strength of outer steel pipe f yo , Inner steel pipe yield strength f yi and specimen height L, the data set is normalized to the interval [-1,1].

[0008] Optionally, the dataset is randomly divided into a training set and a test set, with the ratio generally being 70% and 30%; the training set is used to train the prediction model, and the test set is used to evaluate the performance of the generated model.

[0009] Optionally, the BP neural network algorithm is used to train the model based on the training set; the ten-fold cross-validation method is used for parameter optimization, and the empirical formula is combined to preliminarily determine the relevant range of the model parameters, and the exhaustive method is used to run the used parameters once to obtain the optimal effect value.

[0010] Optionally, the range of the number of hidden layer nodes refers to the following empirical formula:

[0011] In the formula, m is the number of input layer nodes; n is the number of output layer nodes; l is the number of hidden layer nodes; a is a constant between 0 and 10. The number of hidden layer nodes l is calculated to be between 3 and 13, and an adjustment trial is performed.

[0012] Optionally, after determining the optimal parameters, use the test set to evaluate the actual performance of the model. For regression problems, the performance rating indicator of the model is the determination coefficient R 2 , mean square error MSE, root mean square error RMSE; the smaller the RMSE and MSE values, the higher the R 2 The larger the value, the higher the prediction accuracy of the model, which determines the topological structure of the BP neural network model.

[0013] Optionally, an explicit formula is obtained from the neural network model by using its activation function and weights, thresholds and normalization factors.

[0014] In a second aspect, the present invention provides a system for predicting the axial compressive bearing capacity of hollow sandwich steel tube concrete short columns, comprising: The data acquisition module is used to collect the working parameter data of the hollow sandwich steel tube concrete short column and normalize the data; The hyperparameter acquisition module is used to establish the LM algorithm theoretical prediction model based on the original data, divide the normalized data into a training set and a test set, train the prediction model based on the training set, and obtain the optimal hyperparameters; The evaluation module is used to evaluate the model performance using the test set after determining the optimal parameters.

[0015] In a third aspect, the present invention provides a computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the method for predicting the axial compressive bearing capacity of hollow sandwich steel tube concrete short columns when executing the computer program.

[0016] In a fourth aspect, the present invention provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the method for predicting the axial compressive bearing capacity of hollow sandwich steel tube concrete short columns.

[0017] Compared with the prior art, the present invention has the following technical effects: This technical solution aims at predicting the axial compressive bearing capacity of hollow sandwich steel tube concrete short columns. It proposes a data-driven method, combined with the BP neural network algorithm, to achieve a direct mapping relationship prediction from the basic design parameters of the structure to the target performance indicators (axial compressive bearing capacity).

[0018] Firstly, the working parameter data that need to be collected are clarified, including axial compressive bearing capacity, diameter and thickness of inner and outer steel pipes, compressive strength of concrete, yield strength of inner and outer steel pipes and specimen height, which are the key factors affecting axial compressive bearing capacity.

[0019] The data set is normalized to the interval [-1, 1] to eliminate the differences in dimensions and magnitudes between different parameters and improve the efficiency and accuracy of model training.

[0020] Compared with the traditional mechanism-driven method, this scheme is data-centric, fully utilizing the advantages of the big data era and avoiding complicated mechanical derivation and numerical implementation.

[0021] The relationship between various parameters is comprehensively considered through the neural network, which reduces the prediction error and improves the accuracy and stability of the prediction.

[0022] An accurate and stable prediction model was established, which provided strong technical support for the prediction of the axial compressive bearing capacity of hollow sandwich concrete-filled steel tube short columns.

[0023] In summary, this technical solution achieves accurate prediction of the axial compressive bearing capacity of hollow sandwich steel tube concrete short columns through a data-driven approach combined with a BP neural network algorithm. This solution has clear steps of parameter collection, preprocessing, model building and training, performance evaluation, and technical effect analysis, providing a useful reference for solving similar problems. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 It is a flow chart of the present invention.

[0025] Figure 2 It is the topology diagram of BP neural network.

[0026] Figure 3 Implementing a flow chart for Matlab.

[0027] Figure 4 Schematic diagram of the ten-fold cross validation method.

[0028] Figure 5 It is a graphical user interface based on Matlab.

[0029] Figure 6 Schematic diagram of BP neural network hyperparameter optimization.

[0030] Figure 7 Comparison chart of actual value and predicted value. DETAILED DESCRIPTION

[0031] The present invention is further described below in conjunction with the accompanying drawings: Example 1, please refer to Figure 1 , the prediction method of axial compressive bearing capacity of hollow sandwich concrete-filled steel tube short columns includes: Collect the working parameter data of hollow sandwich steel tube concrete short columns and normalize the data; The LM algorithm theoretical prediction model is established based on the original data, the normalized data is divided into a training set and a test set, and the prediction model is trained based on the training set to obtain the optimal hyperparameters; After determining the optimal parameters, the model performance is evaluated using the test set.

[0032] Embodiment 2, a method for predicting the axial compressive bearing capacity of a hollow sandwich steel tube concrete short column, specifically comprises: 1) Collect appropriate data sets that can cover the main influencing parameters of the axial compression performance of hollow sandwich steel tube concrete short columns and their use range in normal design. The main influencing parameters include axial compression bearing capacity Nu, outer steel tube diameter D o and thickness t o , Inner steel pipe diameter D i and thickness t i, concrete compressive strength f c , yield strength of outer steel pipe f yo , Inner steel pipe yield strength f yi And specimen height L, etc.

[0033] The size of a general data set depends on the number of features, preferably 10 to 20 times the number of features. At the same time, in order to avoid the scale effect caused by the numerical value, the data set is normalized to the interval [-1, 1].

[0034] 2) Randomly divide the data set into a training set and a test set, with the ratio generally being 70% and 30%; the training set is used to train the prediction model, while the test set is used to evaluate the performance of the generated model. It should be noted that the distribution of the original data set and the subset data must be the same to ensure that the method can learn a larger range and the test set can cover a wider range of data.

[0035] 3) The BP neural network algorithm is used to train the model based on the training set. In order to find the optimal hyperparameters, the ten-fold cross-validation method is used for parameter optimization, that is, the empirical formula is combined to preliminarily determine the relevant range of the model parameters, and the exhaustive method is used to run the used parameters once to find the optimal effect value.

[0036] At the same time, in order to avoid the deviation caused by random division of the training set, the training set is divided into 10 folds, 9 folds are used for training and 1 fold for validation each time, and the average value is taken to represent the generalization performance of each group of parameters after running 10 times in sequence.

[0037] 4) After determining the optimal parameters, use the test set to evaluate the actual performance of the model. For regression problems, the performance rating indicator of the model is generally the coefficient of determination (R 2 ), mean square error (MSE), root mean square error (RMSE), etc. The smaller the RMSE and MSE values ​​are, the higher the R 2 The larger the value, the higher the prediction accuracy of the model. Determine the topological structure of the BP neural network model.

[0038] 5) The preparation of the graphical user interface, and finally the prediction results of the axial compressive bearing capacity of hollow sandwich steel tube concrete short columns are obtained. It can provide a reference for the design of hollow sandwich steel tube concrete short columns. Researchers and designers can use this interface to estimate the strength of hollow sandwich steel tube concrete columns without conducting experiments or using complex empirical equations, helping users to verify the prediction results with each other to ensure the reliability of the prediction.

[0039] The detailed Matlab steps in steps 1) to 3) are as follows: Step 1: Select the experimental data as the data set and normalize the data set to the interval [-1, 1].

[0040] x(i)=[Do , t o , D i , t i , f c , f yo , f yi , L…]' (1) Where D o , t o is the outer steel pipe diameter and thickness, D i , t i is the diameter and thickness of the inner steel pipe, f c is the compressive strength of sandwich concrete, f yo is the yield strength of the outer steel pipe, f yi is the yield strength of the inner steel pipe and L is the specimen height, x(i) is a set of column vectors including 8 eigenvalues.

[0041] X=[x(1),x(2),…,x(n)](2) Y=[Nu(1), Nu(2),…,Nu(n)](3) Where X is the input vector, X is the target vector and n represents the total number of trials.

[0042] [x, PS1]=mapminmax(X,-1,1)(4) [y, PS2]=mapminmax(Y,-1,1)(5) Step 2: Randomly divide the data set into training set and test set, with the division ratio generally being 70% and 30%; cv = cvpartition(y,'Holdout',0.3,'Stratify',false)(6) xtrain = x(training(cv), :) ; ytrain = y(training(cv), :) (7) xtest= x(testing(cv), :); ytest= y(testing(cv), :) Step 3: Use BP neural network algorithm to train the model based on the training set; net = newff(trainx, trainy, i, {'tansig', 'purelin'}, 'trainlm') (8) net.trainparam.epochs = 400(9) Where i is the number of hidden layer nodes.

[0043] The ten-fold cross-validation method is used for parameter optimization, that is, the relevant range of the model parameters is first determined, and the parameters used are run once using the exhaustive method to find the optimal effect value: The detailed steps of Matlab are as follows: The range of the number of hidden layer nodes refers to the following empirical formula:

[0044] In the formula, m is the number of input layer nodes; n is the number of output layer nodes; l is the number of hidden layer nodes; a is a constant between 0 and 10. In this example, the number of hidden layer nodes l is calculated to be between 3 and 13, and an adjustment trial calculation is performed.

[0045] c = cvpartition(numXtrain,'KFold',10); for i = 3:1:13 for ModelIndex = 1:c.NumTestSets TrainIndex = training(c, ModelIndex); TestIndex= test(c, ModelIndex); %Extract training and test data trainx = xtrain(TrainIndex,:)'; trainy = ytrain(TrainIndex,:)'; testx= xtrain(TestIndex,:)'; testy= ytrain(TestIndex,:)'; %Build a network and predict net = newff(trainx,trainy,i,{'tansig','purelin'},'trainlm'); trainpred = sim(net, trainx);% training set prediction testpred = sim(net,testx);% Test set prediction % Calculate the training set and test set errors e1=trainpred-trainy;RMSE1=sqrt(sum(e1.^2) / length(e1)); e2=testpred-testy;RMSE2= sqrt(sum(e2.^2) / length(e2)); Step 4: After determining the optimal parameters, use the test set to evaluate the actual performance of the model. For regression problems, the performance rating indicators of the model are generally the coefficient of determination (R2), mean square error (MSE), root mean square error (RMSE), etc. Determine the topological structure of the BP neural network model.

[0046] pred = sim(net,x)(11) Pred=mapminmax('reverse', pred, PS2) (12) r = corrcoef(Ytest, Predtest); R2 = r(1,2)^2(13) MSE = sum(e.^2) / length(e)(14) RMSE= sqrt(sum(e.^2) / length(e))(15) Step 5: Compilation of the graphical user interface. Finally, the prediction results of the axial compressive bearing capacity of hollow sandwich steel tube concrete short columns are obtained. The explicit formula is obtained directly from the developed neural network model by using its activation function and parameters (weights, thresholds and normalization factors). The weights and thresholds are extracted as follows: w12 = net.iw{1, 1}; w23 = net.lw{2, 1} (16) b2 = net.b{1}; b3 = net.b{2} (17) Using the formula

[0047] .

[0048] N u : Output layer output, i.e. axial pressure bearing capacity b2: The threshold between the input layer and the hidden layer b3: The threshold between the hidden layer and the output layer W i2 : The weight between the input layer and the hidden layer W j3 : The weight between the hidden layer and the output layer X i : Input variable of the i-th node in the input layer H j : Output of the jth node in the hidden layer Example 3, the present invention can predict and design the axial compressive bearing capacity of hollow sandwich steel tube concrete short columns. In order to illustrate the effect of the present invention, the experimental data in some previous papers are used as the original series for explanation. The following is a modeling prediction based on some historical data of the axial compressive bearing capacity of hollow sandwich steel tube concrete short columns. A total of 150 groups are sorted out, each group of values ​​has 8 characteristic values ​​(external steel tube diameter Do and thickness to, inner steel tube diameter Di and thickness ti, concrete compressive strength fc, outer steel tube yield strength fyo, inner steel tube yield strength fyi and specimen height L) and 1 target value (axial compressive bearing capacity Nu).

[0049] The data are shown in Table 1: Table 1 Dataset of hollow sandwich steel tube concrete short columns

[0050] Using the trial and error method, the mean square error MSE and correlation coefficient r of the LM algorithm regression under the number of hidden layer nodes are calculated. The range of the number of hidden layer nodes is selected based on the following formula:

[0051] In the formula, n is the number of input layer nodes; l is the number of hidden layer nodes; m is the number of output layer nodes; a is a constant between 0 and 10. In this example, the number of hidden layer nodes l is calculated to be between 3 and 13, and an adjustment trial calculation is performed.

[0052] The complete steps of the trial and error method MATLAB implementation process are as follows: clear;clc; data = readmatrix("Ndata.xlsx"); n = size(data,2); X = data(:,1:n-1); X = X'; Y = data(:,n); Y = Y'; net = newff(X, Y, 10, {'tansig', 'purelin'}, 'trainlm'); net.divideParam.trainRatio = 70 / 100; % training set net.divideParam.testRatio = 30 / 100; % test set net.trainParam.epochs=400; % Maximum number of runs [net, tr] = train(net, X, Y); % train function trains the network % Prediction results Pred = sim(net,X);% The sim function gets the predicted value of the network % Plot figure; t=1:length(Pred); plot(t, Y, 'r-*', t, Pred, 'b-o', 'LineWidth', 1) ;% Draw a graph to compare the original y and the y predicted by the network set(gca, 'FontSize', 16);;grid on;box on; xlabel("all samples"), ylabel("carrying capacity"), legend('true value', 'predicted value') ee = sqrt(mse(Pred-Y)); After trial calculation, the data seed is 62704 and the number of hidden layer neurons is 10. At this time, the model performance is good, and the mean square error and correlation coefficient are 119.57 and 0.9944 respectively, which are substituted into the BP neural network model for calculation.

[0053] In yet another embodiment of the present invention, a system for predicting the axial compressive bearing capacity of a short column of hollow sandwich steel tube concrete is provided, which can be used to implement the above-mentioned method for predicting the axial compressive bearing capacity of a short column of hollow sandwich steel tube concrete. Specifically, the system comprises: The data acquisition module is used to collect the working parameter data of the hollow sandwich steel tube concrete short column and normalize the data; The hyperparameter acquisition module is used to establish the LM algorithm theoretical prediction model based on the original data, divide the normalized data into a training set and a test set, train the prediction model based on the training set, and obtain the optimal hyperparameters; The evaluation module is used to evaluate the model performance using the test set after determining the optimal parameters.

[0054] The division of modules in the embodiments of the present invention is schematic and is only a logical function division. There may be other division methods in actual implementation. In addition, each functional module in each embodiment of the present invention may be integrated into one processor, or may exist physically separately, or two or more modules may be integrated into one module. The above-mentioned integrated modules may be implemented in the form of hardware or in the form of software functional modules.

[0055] In another embodiment of the present invention, a computer device is provided, the computer device including a processor and a memory, the memory is used to store a computer program, the computer program includes program instructions, and the processor is used to execute the program instructions stored in the computer storage medium. The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc., which are the computing core and control core of the terminal, which are suitable for implementing one or more instructions, and are specifically suitable for loading and executing one or more instructions in the computer storage medium to implement the corresponding method flow or corresponding function; the processor described in the embodiment of the present invention can be used for the operation of the axial compression bearing capacity prediction method of hollow sandwich steel tube concrete short column.

[0056] In another embodiment of the present invention, the present invention also provides a storage medium, specifically a computer-readable storage medium (Memory), which is a memory device in a computer device for storing programs and data. It is understandable that the computer-readable storage medium here can include both built-in storage media in the computer device and, of course, extended storage media supported by the computer device. The computer-readable storage medium provides a storage space, which stores the operating system of the terminal. In addition, one or more instructions suitable for being loaded and executed by the processor are also stored in the storage space, and these instructions can be one or more computer programs (including program codes). It should be noted that the computer-readable storage medium here can be a high-speed RAM memory or a non-volatile memory, such as at least one disk memory. The processor can load and execute one or more instructions stored in the computer-readable storage medium to implement the corresponding steps of the method for predicting the axial compressive bearing capacity of hollow sandwich steel tube concrete short columns in the above embodiment.

[0057] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0058] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0059] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0060] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0061] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for predicting the axial compressive bearing capacity of hollow sandwich steel tube concrete short columns, characterized in that: include: Collect the working parameter data of hollow sandwich steel tube concrete short columns and normalize the data; The LM algorithm theoretical prediction model is established based on the original data, the normalized data is divided into a training set and a test set, and the prediction model is trained based on the training set to obtain the optimal hyperparameters; After determining the optimal parameters, the model performance is evaluated using the test set.

2. The method for predicting the axial compressive bearing capacity of hollow sandwich steel tube concrete short columns according to claim 1 is characterized in that: Collect working parameter data of hollow sandwich steel tube concrete short columns, including: Axial compression bearing capacity N u , outer steel pipe diameter D o and thickness t o , Inner steel pipe diameter D i and thickness t i , concrete compressive strength f c , yield strength of outer steel pipe f yo , Inner steel pipe yield strength f yi and specimen height L, the data set is normalized to the interval [-1,1].

3. The method for predicting the axial compressive bearing capacity of hollow sandwich steel tube concrete short columns according to claim 1 is characterized in that: The data set is randomly divided into a training set and a test set, with the division ratio generally being 70% and 30%; the training set is used to train the prediction model, and the test set is used to evaluate the performance of the generated model.

4. The method for predicting the axial compressive bearing capacity of hollow sandwich steel tube concrete short columns according to claim 3 is characterized in that: The BP neural network algorithm is used to train the model based on the training set. The ten-fold cross-validation method is used for parameter optimization. Combined with the empirical formula, the relevant range of the model parameters is first determined. The exhaustive method is used to run the used parameters once to obtain the optimal effect value.

5. The method for predicting the axial compressive bearing capacity of hollow sandwich steel tube concrete short columns according to claim 3, characterized in that: The range of hidden layer node numbers refers to the following empirical formula: In the formula, m is the number of input layer nodes; n is the number of output layer nodes; l is the number of hidden layer nodes; a is a constant between 0 and 10. The number of hidden layer nodes l is calculated to be between 3 and 13, and an adjustment trial is performed.

6. The method for predicting the axial compressive bearing capacity of hollow sandwich steel tube concrete short columns according to claim 3 is characterized in that: After determining the optimal parameters, the actual performance of the model is evaluated using the test set. For regression problems, the performance rating indicator of the model is the determination coefficient R 2 , mean square error MSE, root mean square error RMSE; the smaller the RMSE and MSE values, the higher the R 2 The larger the value, the higher the prediction accuracy of the model, which determines the topological structure of the BP neural network model.

7. The method for predicting the axial compressive bearing capacity of hollow sandwich steel tube concrete short columns according to claim 6 is characterized in that: Obtain explicit formula from neural network model by using its activation function and weights, thresholds and normalization factors.

8. Axial compression bearing capacity prediction system for hollow sandwich steel tube concrete short columns, characterized by: include: The data acquisition module is used to collect the working parameter data of the hollow sandwich steel tube concrete short column and normalize the data; The hyperparameter acquisition module is used to establish the LM algorithm theoretical prediction model based on the original data, divide the normalized data into a training set and a test set, train the prediction model based on the training set, and obtain the optimal hyperparameters; The evaluation module is used to evaluate the model performance using the test set after determining the optimal parameters.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method for predicting the axial compressive bearing capacity of a hollow sandwich steel tube concrete short column as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method for predicting the axial compressive bearing capacity of a hollow sandwich steel tube concrete short column as claimed in any one of claims 1 to 7 are implemented.

Images