A method and system for predicting buckling characteristics of thin-walled reinforced structures

Abstract

This invention discloses a method and system for predicting the buckling characteristics of thin-walled stiffened structures, relating to the field of deep learning technology. The method includes: acquiring stiffening parameters of the target thin-walled stiffened structure; the stiffening parameters include the number of stiffeners, stiffener spacing, stiffener web height, and stiffener flange width; inputting the stiffening parameters into a trained buckling load prediction model to obtain the buckling load of the target thin-walled stiffened structure; inputting the stiffening parameters into a trained buckling modal prediction model to obtain the buckling modal image of the target thin-walled stiffened structure; the buckling load and the buckling modal image constitute the buckling characteristics of the target thin-walled stiffened structure. This invention uses a neural network to predict the buckling characteristics of the target thin-walled stiffened structure, improving the accuracy and efficiency of buckling characteristic prediction for thin-walled stiffened structures.

Description

Technical Field

[0001] This invention relates to the field of deep learning technology, and in particular to a method and system for predicting the buckling characteristics of thin-walled reinforced structures. Background Technology

[0002] Thin-walled stiffened structures are widely used in the aerospace field due to their advantages such as light weight, high strength, and high design flexibility. Under the action of compression, shear, and compressive-shear loads, the common failure mode is buckling instability. Therefore, when designing thin-walled stiffened structures, it is necessary to consider their buckling characteristics, which include buckling load and buckling modes.

[0003] Currently, research methods for the buckling characteristics of thin-walled stiffened structures mainly include experimental methods, finite element method (FEM), analytical methods, and engineering calculation methods. Experimental methods are time-consuming and labor-intensive, and inefficient. The FEM simulates experiments, greatly reducing time and labor costs, but still requires significant time for calculation, especially for nonlinear problems, and exhibits poor convergence. Analytical methods can quickly and accurately calculate buckling loads, but their applicability is limited, only applicable to a single type of thin-walled stiffened structure. Engineering calculation methods, based on extensive engineering experience and the introduction of empirical coefficients, can approximate buckling loads, but cannot meet the needs of research requiring high precision. Therefore, a high-precision and efficient method for predicting the buckling characteristics of thin-walled stiffened structures is urgently needed. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for predicting the buckling characteristics of thin-walled reinforced structures, which can improve the accuracy and efficiency of predicting the buckling characteristics of thin-walled reinforced structures.

[0005] To achieve the above objectives, the present invention provides the following solution:

[0006] A method for predicting the buckling characteristics of thin-walled reinforced structures, the method comprising:

[0007] Obtain the reinforcement parameters of the target thin-walled reinforced structure; the reinforcement parameters include the number of reinforcements, the reinforcement spacing, the web height of the reinforcement strip, and the flange width of the reinforcement strip.

[0008] The stiffening parameters are input into the trained buckling load prediction model to obtain the buckling load of the target thin-walled stiffened structure; the trained buckling load prediction model is a model trained by using sample stiffening parameters as input and sample buckling loads corresponding to the sample stiffening parameters as labels.

[0009] The stiffening parameters are input into the trained buckling mode prediction model to obtain the buckling mode image of the target thin-walled stiffened structure; the trained buckling mode prediction model is a model trained by taking the sample stiffening parameters as input and the sample buckling mode image corresponding to the sample stiffening parameters as labels; the buckling load and the buckling mode image constitute the buckling characteristics of the target thin-walled stiffened structure.

[0010] Optionally, before inputting the stiffening parameters into the trained buckling load prediction model to obtain the buckling load of the target thin-walled stiffened structure, the method further includes: training the buckling load prediction model, the training process of which is as follows:

[0011] Obtain a first sample set; the first sample set includes several sample stiffening parameters and the sample buckling load corresponding to each sample stiffening parameter;

[0012] The buckling load prediction model is trained using the first sample set to obtain a trained buckling load prediction model.

[0013] Optionally, obtaining the first sample set specifically includes:

[0014] Set several sample reinforcement parameters;

[0015] For each of the sample stiffening parameters, finite element simulation is performed using the sample stiffening parameters as input to obtain the sample buckling load corresponding to the sample stiffening parameters.

[0016] Optionally, the trained buckling load prediction model is a BP neural network model.

[0017] Optionally, the step of inputting the stiffening parameters into the trained buckling mode prediction model to obtain the buckling mode image of the target thin-walled stiffened structure specifically includes:

[0018] The trained buckling modal prediction model includes multiple trained buckling modal prediction sub-models; each trained buckling modal prediction sub-model corresponds to a pixel in the buckling modal image; the stiffening parameters are input into all trained buckling modal prediction sub-models to obtain the grayscale value of each pixel in the buckling modal image.

[0019] Optionally, before inputting the stiffening parameters into the trained buckling modal prediction model, the method further includes: training the buckling modal prediction model, which includes multiple buckling modal prediction sub-models, as follows:

[0020] Obtain a second sample set; the second sample set includes several sample reinforcement parameters and sample buckling mode images corresponding to each sample reinforcement parameter;

[0021] For each pixel, the buckling modality prediction sub-model is trained using the sample reinforcement parameters as input and the grayscale value of the pixel as the label, to obtain the trained buckling modality prediction sub-model corresponding to the pixel; all the trained buckling modality prediction sub-models corresponding to the pixels constitute the trained buckling modality prediction model.

[0022] Optionally, obtaining the second sample set specifically includes:

[0023] Set several sample reinforcement parameters;

[0024] For each of the sample reinforcement parameters, finite element simulation is performed using the sample reinforcement parameters as input to obtain the sample buckling mode image corresponding to the sample reinforcement parameters.

[0025] Optionally, before training the buckling mode prediction sub-model, the following steps are also included:

[0026] The buckling mode image of the sample is blurred.

[0027] Optionally, before inputting the stiffening parameters into the trained buckling load prediction model, the method further includes:

[0028] The reinforcement parameters are normalized.

[0029] The present invention also provides a buckling characteristic prediction system for thin-walled reinforced structures, the system comprising:

[0030] The reinforcement parameter acquisition module is used to acquire the reinforcement parameters of the target thin-walled reinforced structure; the reinforcement parameters include the number of reinforcements, the reinforcement spacing, the web height of the reinforcement strip, and the flange width of the reinforcement strip.

[0031] The buckling load prediction model module is used to input the stiffening parameters into the trained buckling load prediction model to obtain the buckling load of the target thin-walled stiffened structure; the trained buckling load prediction model is a model trained by using sample stiffening parameters as input and sample buckling loads corresponding to the sample stiffening parameters as labels.

[0032] The buckling mode prediction module is used to input the stiffening parameters into the trained buckling mode prediction model to obtain the buckling mode image of the target thin-walled stiffened structure; the trained buckling mode prediction model is a model trained by using the sample stiffening parameters as input and the sample buckling mode image corresponding to the sample stiffening parameters as labels; the buckling load and the buckling mode image constitute the buckling characteristics of the target thin-walled stiffened structure.

[0033] According to specific embodiments provided by the present invention, the following technical effects are disclosed: The present invention provides a method and system for predicting the buckling characteristics of a thin-walled stiffened structure. The method includes: obtaining stiffening parameters of the target thin-walled stiffened structure; the stiffening parameters include the number of stiffeners, the spacing between stiffeners, the web height of the stiffeners, and the flange width of the stiffeners; inputting the stiffening parameters into a trained buckling load prediction model to obtain the buckling load of the target thin-walled stiffened structure; the trained buckling load prediction model is a model trained using sample stiffening parameters as input and sample buckling loads corresponding to the sample stiffening parameters as labels; inputting the stiffening parameters into a trained buckling modal prediction model to obtain a buckling modal image of the target thin-walled stiffened structure; the trained buckling modal prediction model is a model trained using sample stiffening parameters as input and sample buckling modal images corresponding to the sample stiffening parameters as labels; the buckling load and the buckling modal image constitute the buckling characteristics of the target thin-walled stiffened structure. This invention uses neural networks to predict the buckling characteristics (buckling load and buckling mode) of target thin-walled stiffened structures, thereby improving the accuracy and efficiency of buckling characteristic prediction for thin-walled stiffened structures. Attached Figure Description

[0034] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0035] Figure 1 This is a schematic diagram of the buckling characteristic prediction method for thin-walled reinforced structures provided in an embodiment of the present invention;

[0036] Figure 2 This is a conceptual diagram for predicting the buckling characteristics of a thin-walled reinforced structure provided in an embodiment of the present invention;

[0037] Figure 3 A schematic diagram of the BP neural network structure provided in an embodiment of the present invention;

[0038] Figure 4 The buckling mode prediction model provided in the embodiments of the present invention. Detailed Implementation

[0039] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0040] Artificial neural networks have become a research hotspot in the field of artificial intelligence since the 1980s. Artificial neural network methods have been extensively studied in areas such as the static properties, fatigue phenomena, dynamic properties, creep phenomena, delamination phenomena, impact, crack / damage detection, friction characteristics, and vibration control of thin-walled structures.

[0041] The purpose of this invention is to provide a method and system for predicting the buckling characteristics of thin-walled stiffened structures, employing an artificial neural network approach to predict buckling characteristics (including buckling load and buckling modes). First, parametric modeling is implemented in ABAQUS finite element software using Python scripts. This involves combining different stiffening parameters (stiffening parameters include the number of stiffeners N, stiffening spacing D, stiffener web height H, and stiffener flange width W), and batch modeling and simulation are performed using scripts to obtain a large amount of data (including training and validation sets) required for the neural network. Second, a buckling load prediction model is established to predict buckling loads. The network is trained using the training set, and the validation set is used to test the network's prediction accuracy. High-precision prediction can be achieved by adjusting the neural network type, the number of input layer nodes, the number of hidden layers, and the number of hidden layer nodes. Finally, the buckling mode data is digitized by reducing the pixel count of the buckling mode image and acquiring modal information, thus establishing a buckling mode prediction model to predict buckling modes. This invention uses neural networks to predict the buckling characteristics (buckling load and buckling mode) of target thin-walled stiffened structures, thereby improving the accuracy and efficiency of buckling characteristic prediction for thin-walled stiffened structures.

[0042] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0043] like Figure 1 and Figure 2 As shown, the present invention provides a method for predicting the buckling characteristics of thin-walled reinforced structures, the method comprising:

[0044] S1: Obtain the reinforcement parameters of the target thin-walled reinforced structure; the reinforcement parameters include the number of reinforcements, the reinforcement spacing, the web height of the reinforcement strip, and the flange width of the reinforcement strip.

[0045] S2: Input the stiffening parameters into the trained buckling load prediction model to obtain the buckling load of the target thin-walled stiffened structure; the trained buckling load prediction model is a model trained by taking the sample stiffening parameters as input and the sample buckling load corresponding to the sample stiffening parameters as labels.

[0046] S3: Input the stiffening parameters into the trained buckling mode prediction model to obtain the buckling mode image of the target thin-walled stiffened structure; the trained buckling mode prediction model is a model trained by using the sample stiffening parameters as input and the sample buckling mode image corresponding to the sample stiffening parameters as labels; the buckling load and the buckling mode image constitute the buckling characteristics of the target thin-walled stiffened structure.

[0047] Before step S2, which inputs the stiffening parameters into the trained buckling load prediction model to obtain the buckling load of the target thin-walled stiffened structure, the method further includes: training the buckling load prediction model, the training process of which is as follows:

[0048] Obtain a first sample set; the first sample set includes several sample stiffening parameters and the sample buckling load corresponding to each sample stiffening parameter.

[0049] The buckling load prediction model is trained using the first sample set to obtain a trained buckling load prediction model.

[0050] The acquisition of the first sample set specifically includes:

[0051] Set reinforcement parameters for several samples.

[0052] For each of the sample stiffening parameters, finite element simulation is performed using the sample stiffening parameters as input to obtain the sample buckling load corresponding to the sample stiffening parameters.

[0053] The step of inputting the stiffening parameters into the trained buckling mode prediction model to obtain the buckling mode image of the target thin-walled stiffened structure specifically includes:

[0054] The trained buckling modal prediction model includes multiple trained buckling modal prediction sub-models; each trained buckling modal prediction sub-model corresponds to a pixel in the buckling modal image; the stiffening parameters are input into all trained buckling modal prediction sub-models to obtain the grayscale value of each pixel in the buckling modal image.

[0055] Before inputting the stiffening parameters into the trained buckling modal prediction model, the method further includes: training the buckling modal prediction model, which includes multiple buckling modal prediction sub-models, as follows:

[0056] Obtain a second sample set; the second sample set includes several sample reinforcement parameters and sample buckling mode images corresponding to each sample reinforcement parameter.

[0057] For each pixel, the buckling modality prediction sub-model is trained using the sample reinforcement parameters as input and the grayscale value of the pixel as the label, to obtain the trained buckling modality prediction sub-model corresponding to the pixel; all the trained buckling modality prediction sub-models corresponding to the pixels constitute the trained buckling modality prediction model.

[0058] The acquisition of the second sample set specifically includes:

[0059] Set reinforcement parameters for several samples.

[0060] For each of the sample reinforcement parameters, finite element simulation is performed using the sample reinforcement parameters as input to obtain the sample buckling mode image corresponding to the sample reinforcement parameters.

[0061] Specifically, the first and second sample sets are obtained first. Secondary development in ABAQUS can be divided into two main categories: user subroutine development based on the Fortran language and script development based on the Python language. The second category, script development based on the Python language, can be further subdivided into three main categories: implementing ABAQUS parametric modeling through Python scripts, implementing user-customized post-processing through Python scripts, and using Python to write a visual user interface via the FoxGUI Toolkit. This invention uses Python to implement ABAQUS parametric modeling. Taking a T-shaped unidirectional stiffened straight plate as an example, the stiffening parameters are taken as the number of stiffeners N, the stiffening spacing D, the web height H of the stiffener, and the flange width W of the stiffener.

[0062] Batch finite element simulations were performed on samples with different stiffening parameters. For example, the stiffening parameters of the samples were: 3 stiffeners, 20mm web height, 40mm flange width, and stiffener spacing ranging from 40mm to 150mm. Finite element simulations were performed every 10mm.

[0063] The number of stiffeners, the web height of the stiffeners, and the flange width of the stiffeners are then modified and combined, and a new batch of simulations is continuously performed. Each simulation is post-processed to view and record the sample buckling load and sample buckling mode (sample buckling mode image). This yields the first and second sample sets. 90% of the data from the parametric modeling simulation is selected as the training set for the buckling load prediction model and the buckling mode prediction model, and 10% of the data is selected as the validation set for the buckling load prediction model.

[0064] The buckling load prediction model trained in this invention employs a BP neural network model. A BP neural network is a multi-layer feedforward network trained using the backpropagation algorithm, and is one of the most widely used neural network models currently available. The structure of the BP neural network model is as follows: Figure 3As shown, it includes an input layer, hidden layers, and an output layer, where n is the sum of the number of input layers, hidden layers, and output layers. X1, X2, X3, etc., are multiple input variables of the neural network, and Y1, Y2, etc., are multiple output variables of the neural network.

[0065] Each layer of a neural network is connected by weights, a threshold, and a transfer function, as shown in the following equation:

[0066] x i =f i (w i x i-1 +b i Let i = 2, 3, ..., n. (1)

[0067] In the formula: x i This is a column vector consisting of the data from the i-th layer of the neural network, such as the input vector x1 = (X1, X2, X3, ...). T Output vector x n =(Y1,Y2,…) T ;w i b i These are the weight and threshold matrices between the (i-1)th and ith layers of the BP neural network model, respectively. i The vectorized representation of the transfer function of the BP neural network model is specifically expressed as follows:

[0068] f i ((x1,x2,…,x s ) T )=(f i (x1),f i (x2),…,f i (x s )) T (2)

[0069] In the formula: f i Let be the transfer function between the (i-1)th layer and the ith layer of the neural network. The transfer function is generally:

[0070]

[0071] In the learning process of a BP neural network model, training data enters the input layer, is processed by the hidden layers, and outputs the result through the output layer. When the result is not within the expected error range, an error backpropagation process is initiated, distributing the error equally among the hidden layer units. The hidden layer nodes adjust their weights according to the error magnitude, and then the forward learning process begins again. The training of the BP neural network involves repeating these two processes—forward data learning propagation and error backpropagation—until the output result reaches the expected value, at which point training ends. The trained buckling load prediction model can be validated using a validation set to verify its true performance.

[0072] Before inputting the stiffening parameters into the trained buckling load prediction model, the method further includes: normalizing the stiffening parameters. Specifically:

[0073] To improve the computational efficiency of the neural network and ensure consistent data magnitude, the input and output data are normalized to be between -1 and 1. When normalizing a variable (the number of stiffeners, stiffener spacing, stiffener web height, and stiffener flange width), the maximum and minimum values ​​of the variable are first determined, i.e., the maximum value α. max The minimum value α of the variable min Normalization is performed according to the following formula:

[0074]

[0075] in, α represents the value of the normalized variable. i The values ​​of the variables before normalization.

[0076] This yields the normalized reinforcement parameters, which include the normalized number of reinforcements. Normalized reinforcement spacing Normalized rib web height Normalized stiffener flange width By inputting the normalized stiffening parameters into the trained buckling load prediction model, the normalized buckling load of the target thin-walled stiffened structure can be obtained. That is, the input vector Output vector

[0077] Before training the buckling mode prediction sub-model, the following steps are also included:

[0078] The buckling mode image of the sample is blurred.

[0079] The structure of the buckling modal prediction model is as follows: Figure 4 As shown, in order to reduce the information implied by the buckling modes while retaining key information, the buckling mode image is first appropriately reduced in pixel count. For example, the buckling mode image of a stiffened thin-walled reinforced structure with stiffening parameters N=3, D=100mm, H=20mm, and W=40mm is 320×320 pixels, which is reduced to Num×Num ( Figure 4 (The pixel size is 32×32). In this embodiment, the buckling mode prediction model can employ neural networks such as BP neural networks and RBF neural networks.

[0080] In this embodiment, the buckling mode grayscale image of reduced pixels is represented by grayscale values ​​for each pixel, and transformed into a grayscale value matrix. Figure 4(Only a portion is shown in the image). After normalizing each matrix element, it is used as the output of the buckling mode prediction sub-model. Specifically, the grayscale value of each pixel is normalized to be between -1 and 1. A trained buckling mode prediction sub-model corresponding to each pixel is obtained using the second sample set, resulting in Num×Num trained buckling mode prediction sub-models. These Num×Num buckling mode prediction sub-models constitute the trained buckling mode prediction model.

[0081] The normalized reinforcement parameters (including the normalized number of reinforcements) Normalized reinforcement spacing Normalized rib web height Normalized stiffener flange width As input to the trained buckling mode prediction model, the buckling mode image of the target thin-walled stiffened structure can be obtained.

[0082] The present invention also provides a buckling characteristic prediction system for thin-walled reinforced structures, the system comprising:

[0083] The reinforcement parameter acquisition module is used to acquire the reinforcement parameters of the target thin-walled reinforced structure; the reinforcement parameters include the number of reinforcements, the reinforcement spacing, the web height of the reinforcement strip, and the flange width of the reinforcement strip.

[0084] The buckling load prediction model module is used to input the stiffening parameters into the trained buckling load prediction model to obtain the buckling load of the target thin-walled stiffened structure; the trained buckling load prediction model is a model trained by using sample stiffening parameters as input and sample buckling loads corresponding to the sample stiffening parameters as labels.

[0085] The buckling mode prediction module is used to input the stiffening parameters into the trained buckling mode prediction model to obtain the buckling mode image of the target thin-walled stiffened structure; the trained buckling mode prediction model is a model trained by using the sample stiffening parameters as input and the sample buckling mode image corresponding to the sample stiffening parameters as labels; the buckling load and the buckling mode image constitute the buckling characteristics of the target thin-walled stiffened structure.

[0086] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the descriptions are relatively simple; relevant parts can be referred to the method section.

[0087] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for predicting the buckling characteristics of a thin-walled reinforced structure, characterized in that, The method includes: Obtain the reinforcement parameters of the target thin-walled reinforced structure; the reinforcement parameters include the number of reinforcements, the reinforcement spacing, the web height of the reinforcement strip, and the flange width of the reinforcement strip. The stiffening parameters are input into the trained buckling load prediction model to obtain the buckling load of the target thin-walled stiffened structure; the trained buckling load prediction model is a model trained by using sample stiffening parameters as input and sample buckling loads corresponding to the sample stiffening parameters as labels. The stiffening parameters are input into a trained buckling modal prediction model to obtain a buckling modal image of the target thin-walled stiffened structure. Specifically, the trained buckling modal prediction model includes multiple trained buckling modal prediction sub-models; each trained buckling modal prediction sub-model corresponds to a pixel in the buckling modal image; the stiffening parameters are input into all trained buckling modal prediction sub-models to obtain the grayscale value of each pixel in the buckling modal image; the trained buckling modal prediction model is a model trained using the sample stiffening parameters as input and the sample buckling modal image corresponding to the sample stiffening parameters as a label; the buckling load and the buckling modal image constitute the buckling characteristics of the target thin-walled stiffened structure. Before inputting the stiffening parameters into the trained buckling modal prediction model, the method further includes: training the buckling modal prediction model, which includes multiple buckling modal prediction sub-models, as follows: Obtain a second sample set; the second sample set includes several sample reinforcement parameters and sample buckling mode images corresponding to each sample reinforcement parameter; For each pixel, the buckling modality prediction sub-model is trained using the sample reinforcement parameters as input and the grayscale value of the pixel as the label, to obtain the trained buckling modality prediction sub-model corresponding to the pixel; all the trained buckling modality prediction sub-models corresponding to the pixels constitute the trained buckling modality prediction model.

2. The method for predicting buckling characteristics of thin-walled reinforced structures according to claim 1, characterized in that, Before inputting the stiffening parameters into the trained buckling load prediction model to obtain the buckling load of the target thin-walled stiffened structure, the method further includes: training the buckling load prediction model, the training process of which is as follows: Obtain a first sample set; the first sample set includes several sample stiffening parameters and the sample buckling load corresponding to each sample stiffening parameter; The buckling load prediction model is trained using the first sample set to obtain a trained buckling load prediction model.

3. The method for predicting buckling characteristics of thin-walled reinforced structures according to claim 2, characterized in that, The acquisition of the first sample set specifically includes: Set several sample reinforcement parameters; For each of the sample stiffening parameters, finite element simulation is performed using the sample stiffening parameters as input to obtain the sample buckling load corresponding to the sample stiffening parameters.

4. The method for predicting buckling characteristics of thin-walled reinforced structures according to claim 1, characterized in that, The trained buckling load prediction model is a BP neural network model.

5. The method for predicting buckling characteristics of thin-walled reinforced structures according to claim 1, characterized in that, The acquisition of the second sample set specifically includes: Set several sample reinforcement parameters; For each of the sample reinforcement parameters, finite element simulation is performed using the sample reinforcement parameters as input to obtain the sample buckling mode image corresponding to the sample reinforcement parameters.

6. The method for predicting buckling characteristics of thin-walled reinforced structures according to claim 1, characterized in that, Before training the buckling mode prediction sub-model, the following steps are also included: The buckling mode image of the sample is blurred.

7. The method for predicting buckling characteristics of thin-walled reinforced structures according to claim 1, characterized in that, Before inputting the stiffening parameters into the trained buckling load prediction model, the method further includes: The reinforcement parameters are normalized.

8. A buckling characteristic prediction system for thin-walled reinforced structures, characterized in that, The system includes: The reinforcement parameter acquisition module is used to acquire the reinforcement parameters of the target thin-walled reinforced structure; the reinforcement parameters include the number of reinforcements, the reinforcement spacing, the web height of the reinforcement strip, and the flange width of the reinforcement strip. The buckling load prediction model module is used to input the stiffening parameters into the trained buckling load prediction model to obtain the buckling load of the target thin-walled stiffened structure; the trained buckling load prediction model is a model trained by using sample stiffening parameters as input and sample buckling loads corresponding to the sample stiffening parameters as labels. A buckling modal prediction module is used to input the stiffening parameters into a trained buckling modal prediction model to obtain a buckling modal image of the target thin-walled stiffened structure. Specifically, the trained buckling modal prediction model includes multiple trained buckling modal prediction sub-models; each trained buckling modal prediction sub-model corresponds to a pixel in the buckling modal image; the stiffening parameters are input into all trained buckling modal prediction sub-models to obtain the grayscale value of each pixel in the buckling modal image; the trained buckling modal prediction model is a model trained using the sample stiffening parameters as input and the sample buckling modal image corresponding to the sample stiffening parameters as a label; the buckling load and the buckling modal image constitute the buckling characteristics of the target thin-walled stiffened structure. Before inputting the stiffening parameters into the trained buckling modal prediction model, the method further includes: training the buckling modal prediction model, which includes multiple buckling modal prediction sub-models, as follows: Obtain a second sample set; the second sample set includes several sample reinforcement parameters and sample buckling mode images corresponding to each sample reinforcement parameter; For each pixel, the buckling modality prediction sub-model is trained using the sample reinforcement parameters as input and the grayscale value of the pixel as the label, to obtain the trained buckling modality prediction sub-model corresponding to the pixel; all the trained buckling modality prediction sub-models corresponding to the pixels constitute the trained buckling modality prediction model.

Images