An Artificial Intelligence-Based Vibration Frequency Optimization Method for Bistable Composite Material Structures

Abstract

This invention belongs to the field of composite material structure and artificial intelligence interaction, specifically an AI-based method for optimizing the vibration frequency of bistable composite material structures. The method includes the following steps: classifying randomly generated fiber layup angle data for a cylindrical shell structure; training the data to obtain a classifier that satisfies the constraints; establishing a finite element model and inputting the data satisfying the constraints into the finite element model to obtain the vibration frequencies corresponding to different fiber layup angles, arc lengths, and wrap angles, and inputting this data into a backpropagation (BP) neural network to obtain a calculation model for the vibration frequency; using the classifier and calculation model as the objective function of an improved particle swarm optimization algorithm, and the fiber layup angle, initial radius, and wrap angle as design variables, to obtain the parameter values ​​corresponding to the maximum vibration frequency satisfying the constraints. This invention is applicable to the optimization of vibration frequencies in bistable composite materials and also to the optimization of the performance of other composite material structures.

Description

Technical Field

[0001] This invention belongs to the field of composite material structure and artificial intelligence interaction technology, specifically, it is an artificial intelligence-based method for optimizing the vibration frequency of bistable composite material structures. Background Technology

[0002] Bistable composite cylindrical shell structures are space-deployable structures similar to a measuring tape. These structures can naturally maintain a rolled-up stable state without additional devices, and can autonomously deploy under certain conditions. Therefore, bistable composite cylindrical shell structures not only improve the storage efficiency of deployment systems but also reduce their mass. They are widely used in thin-film spacecraft such as solar sails and rollable flexible solar arrays. Flexible solar arrays require sufficiently high stiffness in the bistable cylindrical shell structure to avoid coupling between the spacecraft control system and the solar array structure modes. Lower modal orders result in greater effective modal mass; therefore, without sacrificing other performance aspects, the first-order natural frequency should be increased as much as possible to improve structural stability.

[0003] Currently, for optimizing the aforementioned vibration frequency, existing technologies combine support vector machines and particle swarm optimization to find the maximum natural frequency corresponding to a fixed fiber angle. However, this method has limitations, only considering the relationship between fiber angle and vibration frequency, and the particle swarm optimization algorithm is prone to getting trapped in local optima in optimization problems.

[0004] This method introduces the initial radius and wrap angle into variables, avoiding the limitation of only considering the relationship between fiber angle and vibration frequency. After expanding the dataset corresponding to different models, the BP neural network algorithm is used for prediction. By introducing an improved particle swarm algorithm that calculates inertia weight through beta distribution to predict vibration frequency, the effect of suppressing the particle swarm algorithm from getting trapped in local optima is achieved. Summary of the Invention

[0005] The purpose of this invention is to provide an optimization method for the vibration frequency of a bistable cantilever composite cylindrical shell structure based on an improved particle swarm optimization algorithm. Specifically, the method utilizes a bagged decision tree algorithm to classify composite cylindrical shell structures with different ply angles and select ply angles that meet the constraints. Then, using finite element simulation software, a vibration frequency analysis model of the bistable composite cylindrical shell structure is established to obtain numerical simulation calculations of the initial radius, wrap angle, and fiber ply angle of the composite material with respect to the vibration frequency. Subsequently, a BP neural network is used to establish a model to optimize and solve the problem. Finally, by improving the particle swarm optimization algorithm, the vibration frequency of the bistable composite cylindrical shell structure is increased by 15.5%.

[0006] The technical solution adopted by this invention to achieve the above objectives is: an optimization method for the vibration frequency of a bistable cantilever composite material column shell structure based on artificial intelligence, used to optimize the structural parameters of composite materials, including the following steps:

[0007] S1: By analyzing the cantilever bistable composite column shell structure, the curling diameter and flattening arc length are set as constant values, fiber layup angle data are randomly generated, and it is determined whether the fiber layup angle data corresponding to different initial radii and wrap angles meet the constraint conditions. The fiber layup angle data are then classified.

[0008] S2: Input the fiber laying angle data obtained in step S1 into MATLAB and model and train it using the bagging tree method to obtain a classifier that judges whether the fiber laying angle data meets the conditions.

[0009] S3: Establish a finite element model and input the fiber laying angle data that meets the constraint conditions in step S1 into the finite element model with the corresponding initial radius and wrap angle to obtain the vibration frequency corresponding to different models and different fiber laying angles.

[0010] S4: Input the vibration frequency obtained in step S3 into the BP neural network, fit the data, and obtain a fast calculation model of the vibration frequency corresponding to the initial radius, wrap angle and fiber laying angle.

[0011] S5: Using the classifier in step S2 and the fast calculation model in S4 as the objective function of the improved particle swarm optimization algorithm, and the initial radius, wrap angle and fiber laying angle data as design variables, obtain the model parameter values ​​corresponding to the maximum vibration frequency that meets the constraints.

[0012] 2. The method for optimizing the vibration frequency of a bistable composite material structure based on artificial intelligence according to claim 1, wherein the angle range in the randomly generated fiber laying angle data is 0° to 90°.

[0013] 3. The method for optimizing the vibration frequency of a bistable composite material structure based on artificial intelligence according to claim 1, characterized in that, in step S1, determining whether the initial radius, wrap angle, and fiber layup angle data satisfy the constraint conditions specifically involves:

[0014] Using randomly generated fiber layup angles as initial data, the stiffness matrices of multiple composite column shell structures are obtained according to the classical laminate theory of composite materials. It is then determined whether the multiple composite column shell structures satisfy the bistable structure, and the fiber layup angle data that satisfy / do not satisfy the bistable structure are then determined.

[0015] The formula for satisfying a bistable structure is:

[0016]

[0017] Among them, D 11 D 12 and D 22 D is the bending stiffness coefficient of the column shell structure. 66 The torsional stiffness coefficient represents the column shell structure.

[0018] Considering the storage space constraints of the unfolding system, a specific curling diameter is added as a constraint condition. That is, during the optimization process, the curling diameter D of the bistable composite material structure is set as a constraint. C Set it to a fixed value.

[0019] The formula for calculating the curl diameter is:

[0020]

[0021] Where R is the initial radius of the cylindrical shell structure.

[0022] The specific arc length of the bistable cylindrical shell is introduced as another constraint for vibration optimization, enabling the structure to better fit the deployment mechanism. The formula for calculating the arc length is as follows:

[0023] L C =Rγ

[0024] Where γ is the wrap angle of the cylindrical shell structure, and R is the initial radius of the cylindrical shell structure.

[0025] The classification of fiber laying data specifically involves dividing the data into two equal parts, with half satisfying the constraints and half not satisfying them.

[0026] In step S2, the classifier for determining the constraints is the piecewise function:

[0027]

[0028] Step S3 specifically includes:

[0029] 3-1) Establishing the finite element model as described above:

[0030] A finite element model was established, and the sample parameter combination of the cylindrical shell structure was set as follows: the length of the cylindrical shell is L, the initial radius of the cylindrical shell is R, the wrap angle of the cylindrical shell is γ, the thickness of the cylindrical shell is t, and the number of ply and the plying method of the cylindrical shell structure are set as [α,β,0,-β,-α]. The parameter attributes with the number of ply and the plying method of [45,-45,0,45,-45] are used as the benchmark. The mesh element is set as shell element S4R, and the boundary condition is set as one end fixed and the other end free. A linear perturbation analysis step is established, and the vibration frequency of the cylindrical shell structure is solved by the Lanczos algorithm to complete the establishment of the finite element model.

[0031] 3-2) Input the fiber laying angle data that meet the constraints determined in step S1 into the finite element model with the corresponding initial radius and wrap angle to obtain the vibration frequency corresponding to different parameters.

[0032] The vibration frequencies corresponding to different parameter values ​​are input into a BP neural network, and the data is fitted to obtain a fast calculation model for the vibration frequencies corresponding to the initial radius, wrap angle, and fiber laying angle. For the expression of the three-layer BP neural network, the matrix form is as follows:

[0033] f(x) = W (o) tansig(W (h) x+b (h) )+b (o)

[0034] Where W is the weight matrix, x and b are vectors, b is the threshold, and the superscripts o and h represent the output layer and hidden layer, respectively.

[0035] The classifier in step S2 and the fast computation model in step S4 are used as the objective functions of the improved particle swarm optimization algorithm, and the initial radius, wrap angle, and fiber layup angle data are used as design variables, i.e.:

[0036]

[0037] Where f(x) represents the fast computation model trained by the BP neural network, μ(x) is the piecewise function corresponding to the classifier in step S2, S is the bistable criterion, α and β are the fiber angles, [α,β,0,-β,-α] is the laying method of the bistable cantilever composite column shell structure, and D C It is the curl diameter of the cylindrical shell structure, L C It is the flattened arc length of the cylindrical shell structure.

[0038] The present invention has the following beneficial effects and advantages:

[0039] 1. This invention has high computational efficiency and accuracy, and can maximize the vibration frequency under the condition of satisfying constraints.

[0040] 2. This invention is not only applicable to the optimization of vibration frequency of bistable composite materials, but also to the optimization of the performance of other composite material structures. On the other hand, under the condition of satisfying the constraints, the method for fast calculation of vibration frequency proposed by finite element simulation and BP neural network is accurate and efficient. At the same time, the optimization algorithm for the parameters of bistable cantilever composite column shell structure based on particle swarm optimization algorithm has fast operation speed and good convergence.

[0041] 3. This invention introduces an improved particle swarm optimization algorithm that calculates inertia weights using beta distribution to predict vibration frequencies, thereby preventing the particle swarm optimization algorithm from getting trapped in local optima.

[0042] 4. By introducing the initial radius and wrap angle as variables, this invention avoids the limitations of only considering the relationship between fiber angle and vibration frequency. Attached Figure Description

[0043] Figure 1 This is a flowchart of the present invention;

[0044] Figure 2 This is a schematic diagram of the cylindrical shell structure in this invention;

[0045] Figure 3 This is a schematic diagram illustrating the batch processing of a BP neural network model dataset using Python according to the present invention;

[0046] Figure 4 This is a diagram of the BP neural network structure of the present invention;

[0047] Figure 5 This is an iterative diagram comparing the improved particle swarm optimization algorithm used in this invention with the basic particle swarm optimization algorithm. Detailed Implementation

[0048] The specific implementation measures of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. The following embodiments are used to illustrate the present invention, but do not limit the scope of the present invention. The accompanying drawings are only schematic diagrams related to specific embodiments and do not represent the entirety of the present invention.

[0049] The above description is merely an overview of the technical solution of the invention. In order to better understand the technical means of the present invention and to implement it in accordance with the description, the following describes in detail, with reference to the accompanying drawings and embodiments, a method for optimizing the vibration frequency of a cantilever bistable composite material structure based on artificial intelligence provided by the present invention.

[0050] In this invention, terms such as bistable composite material, bagged decision tree classification method, BP neural network, and particle swarm optimization algorithm are well-known to those skilled in the art and need not be limited in any way.

[0051] like Figure 1 The diagram shown is a flowchart of the method of the present invention. The present invention provides an artificial intelligence-based method for optimizing the vibration frequency of bistable composite material structures, which is described in detail below. This method is used to optimize the structural parameters of single-layer plate composite materials, specifically optimizing the structural parameters of cantilever bistable composite column-shell structures that meet the constraint conditions. The method includes the following steps:

[0052] S1: Use randomly generated fiber laying angle, initial radius and wrap angle as initial data. The range of randomly generated fiber laying angle is 0° to 90°.

[0053] Using randomly generated fiber layup angles, initial radii, and wrap angles as initial data, the stiffness matrices (ABD matrices) of multiple composite column shell structures are obtained according to the classical laminate theory of composite materials. The existence of a second stable state in bistable composite materials indicates that the strain energy expression of the structure has two extreme values. The derived formula is used to determine whether multiple composite column shell structures satisfy the bistable structure, and then the fiber layup angle data that satisfy / do not satisfy the bistable structure are determined.

[0054] The formula for satisfying a bistable structure is:

[0055]

[0056] Among them, D 11 D 12 and D 22 D is the bending stiffness coefficient of the column shell structure. 66 The torsional stiffness coefficient represents the column shell structure.

[0057] Considering the storage space constraints of the unfolding system, a specific curling diameter is added as a constraint condition. That is, during the optimization process, the curling diameter D of the bistable composite material structure is set as a constraint. C Set it to a fixed value.

[0058] The formula for calculating the curl diameter is:

[0059]

[0060] Where R is the initial radius of the cylindrical shell structure.

[0061] The specific arc length of the bistable cylindrical shell is introduced as another constraint for vibration optimization, enabling the structure to better fit the deployment mechanism. The formula for calculating the arc length is as follows:

[0062] LC =Rγ

[0063] Where γ is the wrap angle of the cylindrical shell structure.

[0064] In this embodiment, the calculated fiber laying angles are classified, and the 4000 data points are divided into two equal parts, with half of the data meeting the constraints and half not meeting the constraints, in order to prepare for the next step of training the classifier model.

[0065] S2: Input all the data obtained in step S1 into the Classification Learner toolbox in MATLAB, and model and train it using the bagged decision tree classification method. Train the classifier in the toolbox to determine whether the structure meets the constraints. Before performing finite element modeling, exclude data that do not meet the constraints in advance to save time in subsequent finite element analysis calculations.

[0066] The classifier that determines the constraints, i.e., the piecewise function:

[0067]

[0068] S3: Establish a finite element model and input the data that meets the constraints in step S1 into the finite element model to obtain the vibration frequencies corresponding to different initial radii, wrap angles and fiber laying angles;

[0069] 3-1) Establishing the finite element model as described above:

[0070] like Figure 2 The diagram shown is a schematic of the cylindrical shell structure in this invention. A finite element model is established, and the sample parameter combination of the cylindrical shell structure is set as follows: the length of the cylindrical shell is L, the initial radius of the cylindrical shell is R, the wrap angle of the cylindrical shell is γ, and the thickness of the cylindrical shell is t. The number of layers and the laying method of the cylindrical shell structure are set as [±α, ​​±β, 0, ±β, ±α]. Parameter attributes with [±45, ±45, 0, ±45, ±45] layers and laying method are used as the benchmark. The mesh element is set as shell element S4R, and the boundary conditions are set as one end fixed and the other end free. A linear perturbation analysis step is established, and the vibration frequency of the cylindrical shell structure is solved using the Lanczos algorithm, thus completing the establishment of the finite element model.

[0071] 3-2) Input the fiber laying angle data that meet the constraints determined in step S1 into the finite element model with the corresponding initial radius and wrap angle to obtain the vibration frequency corresponding to different parameters.

[0072] In this embodiment, a finite element model is established with sample parameters (L = 1.5m), an initial radius ranging from 16 to 50 mm, and a wrap angle γ ranging from 120° to 358°. It is assumed that the laminate has 5 layers and a laying pattern of [45° / -45° / 0° / 45° / -45°]. A columnar shell structure with the above parameter properties is used as the reference. The boundary conditions are set to one end fixed and the other end free. A linear perturbation analysis step is established, and the Lanczos algorithm is selected to solve for the vibration frequency of the cantilever structure. The mesh element is set to shell element S4R, and the mesh number is set to 24000. The job is submitted. Using the created finite element model, batch processing is performed to obtain the vibration frequencies corresponding to different fiber laying angles.

[0073] The finite element model in S3 adopts an antisymmetric unidirectional strip laying method, taking into account the characteristics of bistable composite materials.

[0074] As a further optimization of the present invention, the parameterized model of the bistable composite material cylindrical shell structure established by the finite element simulation software in S3 includes the cylindrical shell length L, the cylindrical shell wrap angle γ, the initial radius R of the cylindrical shell, the thickness t, and the fiber laying sequence [±α, ​​±β, 0, ±β, ±α].

[0075] S4: Input the vibration frequencies corresponding to different parameter values ​​into the BP neural network, such as... Figure 3 As shown, the BP neural network contains one hidden layer and 11 hidden nodes. Using this structure to fit the data, a fast calculation model for the vibration frequency corresponding to the initial radius, wrap angle, and fiber layup angle is obtained. The expression for the three-layer BP neural network is in matrix form:

[0076] f(x) = W (o) tansig(W (h) x+b (h) )+b (o)

[0077] Where W is the weight matrix, x and b are vectors, b is the threshold, and the superscripts o and h represent the output layer and hidden layer, respectively.

[0078] like Figure 4 As shown, this invention utilizes Python to batch process a BP neural network dataset. Based on the data obtained in S3, the data is input into the BP neural network to fit the data and obtain a fast calculation model of the vibration frequency corresponding to the initial radius, wrap angle, and fiber laying angle. In this step, considering that the finite element calculation time is too long, the BP neural network fast calculation model is used instead of the finite element model.

[0079] As a further optimization of the present invention, the fast calculation model described in S4 divides the dataset using the hold-out method, using 90% of the data as the training set and the remaining 10% as the validation set. Since the reserved validation set does not participate in the model training, its prediction effect is more convincing.

[0080] S5: The classifier from step S2 and the fast computation model from S4 are used as the objective function of the improved particle swarm optimization algorithm, and the initial radius, wrap angle, and fiber layup angle data are used as design variables, i.e.:

[0081]

[0082] Where f(x) represents the fast computation model trained by the BP neural network, μ(x) is the piecewise function corresponding to the classifier in step S2, S is the bistable criterion, α and β are the fiber angles, [α,β,0,-β,-α] is the laying method of the cantilever bistable composite column shell structure, and D C It is the curl diameter of the cylindrical shell structure, L C It is the flattened arc length of the cylindrical shell structure.

[0083] The initial radius, wrap angle, and fiber layup angle corresponding to the maximum vibration frequency satisfying the constraints are determined. The particle swarm optimization process is as follows: Figure 5 As shown, the modified particle swarm optimization algorithm constructs an inertia weight based on an exponential function and introduces a differential evolution algorithm to update the particle positions. This algorithm has higher search accuracy and faster convergence speed. After 50 iterations, the particles are concentrated at the optimal solution position, obtaining the maximum value of the vibration frequency. The improved particle swarm optimization algorithm effectively shortens the solution time for the vibration frequency and solves the problems of premature convergence and getting trapped in local optima in the particle swarm optimization algorithm.

[0084] This embodiment presents an artificial intelligence-based method for optimizing the vibration frequency of cantilever bistable composite material structures. The method uses particle swarm optimization and an improved particle swarm optimization algorithm with dynamically adjusted inertia weights (IDWPSO) to compare the optimization of the vibration frequency of the same material.

[0085] Table 1 shows the optimization results of this embodiment using particle swarm optimization, considering the actual production and manufacturing of bistable composite material column shell structures, and the fact that there are 5 possible radii and wrap angles that meet the constraints.

[0086] Table 1

[0087]

[0088]

[0089] Table 2 shows the optimization results of the improved particle swarm optimization method with dynamically adjusted inertia weights, compared with the particle swarm optimization algorithm in this embodiment.

[0090] Table 2

[0091] R Corner α β Vibration frequency 19mm 301.53° 28.40 1.52 12.41 22.5mm 254.52° 49.7 -71.47 9.48 28.5mm 201.21° -57.95 -63.25 7.41 38mm 150.19° 59.3 89.41 5.7017 44.5mm 128.98° 60 34 4.987

[0092] As shown in Table 1, the artificial intelligence-based vibration frequency optimization method for bistable composite material structures determined in this invention improves the vibration frequency of the benchmark cylindrical shell structure and effectively reduces the solution time compared with the traditional finite element model. Through comparison of optimization results and model evaluation analysis, the accuracy of the method combining finite element analysis and artificial intelligence for optimizing the vibration frequency of bistable cantilever structures is demonstrated. Analysis of the data in Tables 1 and 2 shows that when using the particle swarm optimization algorithm, the fiber layup angles are close to the critical values, potentially leading to local optima. The improved particle swarm optimization algorithm avoids this problem. Furthermore, a smaller initial radius and a larger wrap angle make it easier to obtain a larger vibration frequency. Therefore, the maximum vibration frequency is obtained when the radius is 19 mm and the wrap angle is 301.53°.

[0093] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the invention. Those skilled in the art can make various changes and modifications without departing from the spirit and scope of the invention. Therefore, all technical solutions obtained through equivalent substitution or transformation fall within the protection scope of the present invention.

Claims

1. A method for optimizing the vibration frequency of bistable composite material structures based on artificial intelligence, characterized in that, Includes the following steps: S1: By analyzing the bistable cantilever composite column shell structure, the curling diameter and flattening arc length are set as constant values, fiber layup angle data are randomly generated, and it is determined whether the fiber layup angle data corresponding to different initial radii and wrap angles meet the constraint conditions. The fiber layup angle data are then classified. S2: Input the fiber laying angle data obtained in step S1 into MATLAB and model and train it using the bagging decision tree method to obtain a classifier that judges whether the fiber laying angle data meets the conditions. S3: Establish a finite element model and input the fiber laying angle data that meets the constraint conditions in step S1 into the finite element model with the corresponding initial radius and wrap angle to obtain the vibration frequency corresponding to different models and different fiber laying angles. S4: Input the vibration frequency obtained in step S3 into the BP neural network, fit the data, and obtain a fast calculation model of the vibration frequency corresponding to the initial radius, wrap angle and fiber laying angle. S5: Using the classifier in step S2 and the fast calculation model in S4 as the objective function of the improved particle swarm optimization algorithm, and the initial radius, wrap angle and fiber laying angle data as design variables, obtain the model parameter values ​​corresponding to the maximum vibration frequency that meets the constraints.

2. The method for optimizing the vibration frequency of a bistable composite material structure based on artificial intelligence according to claim 1, characterized in that, The angle range in the randomly generated fiber laying angle data is 0° to 90°.

3. The method for optimizing the vibration frequency of a bistable composite material structure based on artificial intelligence according to claim 1, characterized in that, In step S1, determining whether the initial radius, wrap angle, and fiber layup angle data meet the constraints specifically involves: Using randomly generated fiber layup angles as initial data, the stiffness matrices of multiple composite column shell structures are obtained according to the classical laminate theory of composite materials. It is then determined whether the multiple composite column shell structures satisfy the bistable structure, and the fiber layup angle data that satisfy / do not satisfy the bistable structure are then determined. The formula for satisfying a bistable structure is: Among them, D 11 D 12 and D 22 D is the bending stiffness coefficient of the column shell structure. 66 The torsional stiffness coefficient representing the cylindrical shell structure; Considering the storage space constraints of the unfolding system, a specific curling diameter is added as a constraint condition. That is, during the optimization process, the curling diameter D of the bistable composite material structure is set as a constraint. C Set to a fixed value; The formula for calculating the curl diameter is: Where R is the initial radius of the cylindrical shell structure; The specific arc length of the bistable cylindrical shell is introduced as another constraint for vibration optimization, enabling the structure to conform to the deployment mechanism. The arc length L C The formula for calculation is as follows: L C =R γ Where γ is the wrap angle of the cylindrical shell structure, and R is the initial radius of the cylindrical shell structure.

4. The method for optimizing the vibration frequency of a bistable composite material structure based on artificial intelligence according to claim 1, characterized in that, The data classification process specifically involves dividing the data into two equal parts, with half satisfying the constraints and half not satisfying them.

5. The method for optimizing the vibration frequency of a bistable composite material structure based on artificial intelligence according to claim 1, characterized in that, In step S2, the classifier used to determine whether the constraints are met is the piecewise function:

6. The method for optimizing the vibration frequency of a bistable composite material structure based on artificial intelligence according to claim 1, characterized in that, Step S3 specifically includes: 3-1) Establishing the finite element model as described above: A finite element model was established, and the sample parameter combination of the cylindrical shell structure was set as follows: the length of the cylindrical shell is L, the initial radius of the cylindrical shell is R, the wrap angle of the cylindrical shell is γ, the thickness of the cylindrical shell is t, and the number of ply and the plying method of the cylindrical shell structure are set as [α,β,0,-β,-α]. The parameter attributes with the number of ply and the plying method of [45,-45,0,45,-45] are used as the benchmark. The mesh element is set as shell element S4R, and the boundary conditions are set as one end fixed and the other end free. A linear perturbation analysis step is established, and the vibration frequency of the cylindrical shell structure is solved by the Lanczos algorithm to complete the establishment of the finite element model. 3-2) Input the fiber laying angle data that meet the constraints determined in step S1 into the finite element model with the corresponding initial radius and wrap angle to obtain the vibration frequency corresponding to different parameters.

7. The method for optimizing the vibration frequency of a bistable composite material structure based on artificial intelligence according to claim 1, characterized in that, The vibration frequencies corresponding to different parameter values ​​are input into a BP neural network, and the data is fitted to obtain a fast calculation model for the vibration frequencies corresponding to the initial radius, wrap angle, and fiber laying angle. For the expression of the three-layer BP neural network, the matrix form is as follows: f(x)=W (o) tansig(W (h) x+b (h) )+b (o) Where W is the weight matrix, x and b are vectors, b is the threshold, and the superscripts o and h represent the output layer and hidden layer, respectively.

8. The method for optimizing the vibration frequency of a bistable composite material structure based on artificial intelligence according to claim 1, characterized in that, The classifier in step S2 and the fast computation model in step S4 are used as the objective functions of the improved particle swarm optimization algorithm, and the initial radius, wrap angle, and fiber layup angle data are used as design variables, i.e.: Where f(x) represents the fast computation model trained by the BP neural network, μ(x) is the piecewise function corresponding to the classifier in step S2, S is the bistable criterion, α and β are the fiber angles, [±α, ​​±β, 0, ±β, ±α] is the laying method of the bistable composite column shell structure, and D C It is the curl diameter of the cylindrical shell structure, L C It is the flattened arc length of the cylindrical shell structure.

Images