Composite material step-by-step electromagnetic forming method and system
Through finite element simulation and deep neural network model, combined with movable platform and screw lift, the problems of low computational efficiency and difficult parameter optimization in electromagnetic forming technology are solved, and efficient forming and precise control of composite materials are achieved, and forming performance is improved. .
Patent Information
- Application Number
- CN202411043987.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-31
- Publication Date
- 2024-12-03
- Estimated Expiration
- 2044-07-31
AI Technical Summary
The existing electromagnetic forming technology has low calculation efficiency, difficult parameter optimization and difficult process control during the forming process of metal composite materials, resulting in poor forming performance.
The composite material step-by-step electromagnetic forming method is adopted, and the relationship between multiple parameters is established using finite element simulation and deep neural network model. Intelligent regulation is achieved through movable platform and screw lift, and electromagnetic parameters and workpiece parameters are predicted and adjusted to optimize the forming process.
The calculation efficiency and parameter optimization capabilities of electromagnetic forming are improved, efficient forming of composite materials is achieved, and forming performance and processing control accuracy are enhanced.
Smart Images

Figure CN119056943A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of electromagnetic forming and intelligent manufacturing, and in particular to a step-by-step electromagnetic forming method and system for composite materials. Background Art
[0002] Electromagnetic forming (EMF), a machining technology that utilizes pulsed electromagnetic forces, has garnered widespread attention from both industry and academia due to its advantages, including extremely fast workpiece forming speeds, high surface quality, and significant improvements in the forming limit of materials. EMF is currently widely used in the machining of lightweight sheet materials. Studies have demonstrated that EMF can improve the forming limit of AA1060-H24 sheet by as much as 68.3%. Some researchers have also investigated the EMF of metal-polymer composites, demonstrating similarly impressive improvements in the forming limit. However, research on EMF of metal composites remains limited. In the next generation of aerospace technology, metal composites are gradually replacing monometallic materials, and demand for metal composites is also growing in other fields. For example, multilayer Al-Ti sheets, characterized by their high ductility and high specific strength, are widely used in the automotive and aerospace industries. Al / Mg composites effectively combine the advantages of Mg's high specific strength and light weight with Al's ductility, resulting in performance far superior to that of monometallic materials. Therefore, research on EMF of composite materials, making EMF more suitable for composite machining and improving their forming performance, has become a crucial engineering challenge.
[0003] There are three major problems in the electromagnetic forming of metal composites: (1) The computational efficiency of electromagnetic forming simulation is low, which involves structural and electromagnetic multi-physics fields. Improving the computational efficiency of electromagnetic forming has become a key issue in the industry; (2) The ductility of different parts of the composite material is different, and process control is required during forming to reduce problems such as material separation and deformation during the forming process; (3) Due to the coupling of multiple physical fields, there are many parameters, such as electromagnetic parameters such as current, workpiece parameters such as width and thickness, and process parameters such as friction and pressure. Controlling and optimizing multiple parameters has become a major difficulty in electromagnetic forming. Summary of the Invention
[0004] In order to solve the technical problems of low computational efficiency, difficulty in parameter optimization and difficulty in process control in existing electromagnetic forming solutions, an embodiment of the present invention provides a step-by-step electromagnetic forming method and system for composite materials.
[0005] The technical solution of the embodiment of the present invention is achieved as follows:
[0006] An embodiment of the present invention provides a step-by-step electromagnetic forming method for composite materials, which is applied to a step-by-step electromagnetic forming system for composite materials, wherein the step-by-step electromagnetic forming system for composite materials includes a movable platform and a composite network model; the movable platform includes a base mold, a discharge coil and a screw elevator; the base mold is used to provide a forming shape for a workpiece; the discharge coil is used to generate an electromagnetic force so that the workpiece is formed under the action of the electromagnetic force; the screw elevator is used to adjust the distance from the discharge coil to the workpiece; the composite network model includes a first network model, a second network model and a third network model; the method includes: using the first network model to perform finite element simulation on the electromagnetic forming process of the workpiece to obtain multiple sets of simulation data; each set of simulation data includes electromagnetic parameters, workpiece parameters, process parameters, adjustment parameters, and the corresponding maximum deformation and maximum stress of the workpiece; the multiple sets of simulation data are input into the second network model to obtain the second network model generated similar simulation data; splicing the simulation data and the similar simulation data and performing data format processing to obtain training samples; using the training samples to train the third network model, and using the trained third network model to predict the adjustment parameters and maximum stress of the workpiece under the conditions of electromagnetic parameters, workpiece parameters, process parameters and maximum deformation; the trained third network model includes a mapping relationship between electromagnetic parameters, workpiece parameters, process parameters, maximum deformation to adjustment parameters and a mapping relationship between electromagnetic parameters, workpiece parameters, process parameters, maximum deformation to maximum stress; judging whether it meets the processing permission range of the workpiece according to the maximum stress; if it meets the processing permission range of the workpiece, controlling the screw lift to adjust the distance from the discharge coil to the workpiece so that the distance from the discharge coil to the workpiece is the adjustment parameter; repeating the above operation multiple times until the workpiece reaches the bottom of the base mold, completing the electromagnetic forming of the workpiece.
[0007] In one embodiment, the first network model includes a first script and a first solver; the first network model is used to perform finite element simulation on the electromagnetic forming process of the workpiece to obtain multiple sets of simulation data, including: establishing an initial simulation process through a first software to obtain a first script; running the first script in the first solver, setting parameters and loop conditions, inputting multiple sets of electromagnetic parameters, workpiece parameters, process parameters and adjustment parameters, and obtaining the corresponding output maximum stress and maximum deformation; based on the multiple sets of electromagnetic parameters, workpiece parameters, process parameters, adjustment parameters, and the corresponding maximum stress and maximum deformation, multiple sets of simulation data are obtained.
[0008] In one embodiment, the second network model includes an encoder, a generator and a discriminator, and the multiple groups of simulation data are input into the second network model to obtain similar simulation data generated by the second network model, including: inputting the multiple groups of simulation data into the encoder to obtain intermediate variables output by the encoder; inputting the intermediate variables into the generator to obtain similar simulation data output by the generator; inputting the simulation data and the similar simulation data into the discriminator to obtain the identification probability output by the discriminator; the identification probability is used to identify whether the similar simulation data meets the preset requirements, and outputting the similar simulation data that meets the preset requirements.
[0009] In one embodiment, the encoder, the generator and the discriminator are all deep neural networks, the input layer of the encoder includes 8 neurons or 6 neurons, and the output layer of the encoder includes 2 neurons; the input layer of the generator includes 2 neurons, and the output layer of the generator includes 8 neurons or 6 neurons; the input layer of the discriminator includes 16 neurons or 12 neurons, and the output layer of the discriminator includes one neuron; the number of hidden layers and hidden layer neurons of the encoder, the generator and the discriminator is positively correlated with the data volume of the simulation data.
[0010] In one embodiment, the loss function during training of the second network model includes a first reconstruction loss part and a second loss part:
[0011] The first reconstruction loss part includes:
[0012] Loss reconstruction →||Output Generator -G(E(Input Encoder ))||
[0013] Among them, G() is the generator, E() is the encoder, Output Generator is the output of the generator, Input Encoder is the input of the encoder;
[0014] The second part of the loss includes:
[0015] Loss GAN →log(D(Input Encoder ))+log(1-D(G(E(Output Encoder ))))+log(1-D(G(Z p1 ,Z p2 )))
[0017] Among them, D() is the discriminator, Z p1 ,Zp2 is an intermediate variable, which is obtained by resampling in the normal distribution space. Input Encoder is the input of the encoder, Output Encoder is the output of the encoder, G() is the generator, and E() is the encoder.
[0018] In one embodiment, the simulation data and the similar simulation data are spliced and then data formatted to obtain training samples, including: splicing the simulation data and the similar simulation data row by row to obtain spliced data; adjusting the input-output relationship of the spliced data to obtain training samples; the training samples include a first training sample and a second training sample; the input items of the first training sample are electromagnetic parameters, workpiece parameters, process parameters and maximum deformation, and the output item of the first training sample is an adjustment parameter; the input items of the second training sample are electromagnetic parameters, workpiece parameters, process parameters and maximum deformation, and the output item of the second training sample is maximum stress.
[0019] In one embodiment, the third network model includes a first sub-model and a second sub-model; training the third network model using the training samples includes: training the first sub-model using the first training samples; training the second sub-model using the second training samples; the trained first sub-model includes a mapping relationship between electromagnetic parameters, workpiece parameters, process parameters, and maximum deformation to adjustment parameters; the trained second sub-model includes a mapping relationship between electromagnetic parameters, workpiece parameters, process parameters, and maximum deformation to maximum stress.
[0020] In one embodiment, the loss function during training of the first sub-model is:
[0021]
[0022] Among them, H i To adjust the parameters, that is, the distance from the coil to the workpiece (represented by the coil to the lowest point of the mold), H' i Generate parameters for the corresponding generator, H″ i is the LSTM1 prediction parameter, and m is the amount of LSTM data;
[0023] The loss function during training of the second sub-model is:
[0024]
[0025] Among them, SMAX i is the maximum stress of the workpiece, SMAX′ i Generate stress for generator, SMAX″ i The stress prediction of LSTM2 is m, and the amount of data of LSTM is m.
[0026] In one embodiment, controlling the screw lifter to adjust the distance from the discharge coil to the workpiece includes:
[0027] The linear quadratic regulator is used to control the screw lift to adjust the distance from the discharge coil to the workpiece.
[0028] An embodiment of the present invention also provides a step-by-step electromagnetic forming system for composite materials, which includes a movable platform, a composite network model and a processor; the movable platform includes a base mold, a discharge coil and a screw lift; the base mold is used to place the workpiece for processing; the discharge coil is used to generate electromagnetic force so that the workpiece is formed under the action of the electromagnetic force; the screw lift is used to adjust the distance from the discharge coil to the workpiece for processing; the composite network model includes a first network model, a second network model and a third network model; the processor is used to execute the steps of any of the above methods.
[0029] The embodiment of the present invention provides a method and system for step-by-step electromagnetic forming of composite materials, which uses the first network model to perform finite element simulation on the electromagnetic forming process of the workpiece to obtain multiple sets of simulation data; each set of simulation data includes electromagnetic parameters, workpiece parameters, process parameters, adjustment parameters, and the maximum deformation and maximum stress of the corresponding workpiece; the multiple sets of simulation data are input into the second network model to obtain similar simulation data generated by the second network model; the simulation data and the similar simulation data are spliced and then processed in data format to obtain training samples; the training samples are used to train the third network model, and the trained third network model is used to predict the electromagnetic parameters, workpiece parameters, process parameters, adjustment parameters, and the maximum deformation and maximum stress of the workpiece. The training process comprises the following steps: determining the relationship between electromagnetic parameters, workpiece parameters, process parameters, and the maximum deformation under the conditions of adjustment parameters and maximum stress; determining the relationship between electromagnetic parameters, workpiece parameters, process parameters, and the maximum deformation and the maximum stress; judging whether the processing permission range of the workpiece is met based on the maximum stress; controlling the screw elevator to adjust the distance between the discharge coil and the workpiece when the processing permission range is met, so that the distance between the discharge coil and the workpiece is the adjustment parameter; repeating the above operation multiple times until the workpiece reaches the bottom of the base mold, thereby completing the electromagnetic forming of the workpiece. The solution provided by the present invention establishes the relationship between multiple parameters through finite element simulation and neural network architecture, realizes intelligent control of the distance between the coil and the workpiece and stress prediction, and realizes multiple step-by-step forming through multiple adjustments. BRIEF DESCRIPTION OF THE DRAWINGS
[0030] Figure 1Schematic diagram of the process of the step-by-step electromagnetic forming method of a composite material according to an embodiment of the present invention;
[0031] Figure 2 A schematic diagram of the structure of a movable platform according to an embodiment of the present invention;
[0032] Figure 3 Schematic diagram of the technical implementation process of the composite material step-by-step electromagnetic forming technology based on a movable platform and MC-VAE-GAN-LSTM according to an embodiment of the present invention;
[0033] Figure 4 This is a partial mesh diagram of a COMSOL composite plate according to an embodiment of the present invention;
[0034] Figure 5 This is a schematic diagram of electromagnetic forming in steps according to an embodiment of the present invention;
[0035] Figure 6 This is a schematic diagram of the network architecture of the MC-VAE-GAN-LSTM model according to an embodiment of the present invention;
[0036] Figure 7 This is a diagram of the internal structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION
[0037] The present invention will be described in further detail below with reference to the accompanying drawings and embodiments.
[0038] An embodiment of the present invention provides a composite material step-by-step electromagnetic forming method, which is applied to a composite material step-by-step electromagnetic forming system, wherein the composite material step-by-step electromagnetic forming system includes a movable platform and a composite network model; the movable platform includes a base mold, a discharge coil, and a screw elevator; the base mold is used to provide a forming shape for a workpiece; the discharge coil is used to generate an electromagnetic force so that the workpiece is formed under the action of the electromagnetic force; the screw elevator is used to adjust the distance from the discharge coil to the workpiece; the composite network model includes a first network model, a second network model, and a third network model; Figure 1 As shown, the method includes:
[0039] Step 101: Using the first network model, perform finite element simulation on the electromagnetic forming process of the workpiece to obtain multiple sets of simulation data; each set of simulation data includes electromagnetic parameters, workpiece parameters, process parameters, adjustment parameters, and the corresponding maximum deformation and maximum stress of the workpiece;
[0040] Step 102: inputting the plurality of sets of simulation data into the second network model to obtain similar simulation data generated by the second network model;
[0041] Step 103: splicing the simulation data and the similar simulation data and performing data format processing to obtain training samples;
[0042] Step 104: Using the training samples to train the third network model, and using the trained third network model to predict the adjustment parameters and maximum stress of the workpiece under the conditions of electromagnetic parameters, workpiece parameters, process parameters, and maximum deformation; the trained third network model includes a mapping relationship between electromagnetic parameters, workpiece parameters, process parameters, and maximum deformation to adjustment parameters, and a mapping relationship between electromagnetic parameters, workpiece parameters, process parameters, and maximum deformation to maximum stress;
[0043] Step 105: determining whether the maximum stress meets the permitted processing range of the workpiece according to the maximum stress; if the maximum stress meets the permitted processing range of the workpiece, controlling the screw lifter to adjust the distance between the discharge coil and the workpiece so that the distance between the discharge coil and the workpiece is the adjustment parameter;
[0044] Step 106: Repeat the above operation multiple times until the workpiece reaches the bottom of the base mold, completing the electromagnetic forming of the workpiece.
[0045] This embodiment achieves multi-step forming using a mobile platform that adjusts the distance between the coil and the workpiece. Based on a composite network model (i.e., the MC-VAE-GAN-LSTM model), automated finite element simulation and a neural network architecture establish relationships between multiple parameters, enabling intelligent control of the distance between the coil and the workpiece and predicting stress in the workpiece.
[0046] Specifically, see Figure 2 , the movable platform in this embodiment includes the following components:
[0047] Processing part:
[0048] 1-Matrix mold: composite material forming mold, the matrix shape is the forming shape;
[0049] 2-Mold base: Place the base mold and connect it to the 7-screw lift;
[0050] 3-Clamp: Apply force F to clamp the two ends of the composite material;
[0051] 4-Discharge coil: generates electromagnetic force;
[0052] 5-Coil fixing bracket: fix 4-discharge coil;
[0053] 6- Composite sheet: A composite material sheet that needs to be processed and formed, consisting of different metal material layers 1 and metal material 2.
[0054] Adjustment part:
[0055] 7-Screw lift: realizes the adjustment of the distance H between the coil and the workpiece (the distance from the coil to the lowest point of the mold);
[0056] 8-Motor: realizes the rotation movement of 7-screw elevator;
[0057] 9- Spring damping support: supports 2- mold base and absorbs vibration when 7- screw lift is working.
[0058] 10-Lifter base: fixes 7-screw lift, 8-motor and 9-spring damping support.
[0059] Specifically, the composite network model in this embodiment, namely the MC-VAE-GAN-LSTM model, includes a first network model MC model, a second network model VAE-GAN model and a third network model LSTM model.
[0060] Further, see Figure 3 The MC-VAE-GAN-LSTM model in this embodiment includes the following steps:
[0061] S1: Build the MC model. The MC model consists of a MATLAB script and a COMSOL solver. MATLAB creates a COMSOL structural and electromagnetic multiphysics simulation loop script, enabling real-time communication with COMSOL. The maximum deformation and stress of the composite material sheet are obtained for multiple sets of different electromagnetic parameters, workpiece parameters, process parameters, and the adjustment parameter (H). The results are automatically written to a .txt text file.
[0062] S2: Build and train a VAE-GAN model, which is a deep neural network. The VAE-GAN model consists of a deep encoder, a deep generator, and a deep discriminator. The input to the VAE-GAN network is a set of different electromagnetic parameters, workpiece parameters, process parameters, and a tuning parameter (H), along with the corresponding maximum deformation and maximum stress of the composite sheet. The output is a set of electromagnetic parameters, workpiece parameters, process parameters, tuning parameters (H), and the maximum deformation and maximum stress of the composite sheet generated by the deep generator.
[0063] S3: Use Python Script to reshape the data obtained in S1 and S2. The data of S1 and S2 are concatenated row by row, and the electromagnetic parameters, workpiece parameters, process parameters, and maximum deformation of the composite plate are used as input columns, and the adjustment parameter (H) and maximum stress are used as output columns.
[0064] S4: Build and train LSTMs. LSTM1 establishes a mapping relationship between electromagnetic parameters, workpiece parameters, process parameters, and the maximum deformation of the composite material sheet to the adjustment parameter (H). LSTM2 establishes a mapping relationship between electromagnetic parameters, workpiece parameters, process parameters, and the maximum deformation of the composite material sheet to the maximum stress. Both LSTM1 and LSTM2 must pass through a fully connected layer to obtain their output.
[0065] The specific implementation steps of S1 are:
[0066] S11: First, the entire solution process is established using the first software, COMSOL, i.e., the initial simulation process is established, including structural dimensions, structural-electromagnetic physics, geometric (workpiece) parameter settings, solution domain and deformation domain selection, structural boundary conditions, material definition, electromagnetic parameter settings, meshing, and solver settings. This is then recorded and saved as a basic MATLAB script. The MATLAB input is:
[0067] Input MATLAB =[A i ,T i ,W i ,F i ,f i ,H i ](i=1,2,...,n)
[0068] Among them, A i is the current (electromagnetic parameter), T i is the thickness of the workpiece (workpiece parameter), W i is the width of the workpiece (workpiece parameter), F i is the clamping force of the fixture (process parameter), f i is the friction coefficient (process parameter), H i is the adjustment parameter, i.e., the distance from the coil to the workpiece (expressed from the coil to the lowest point of the mold), and n is the data volume. (To simplify the solution, the process parameters can be simplified to fixed supports in structural mechanics).
[0069] S12: Then, the basic MATLAB script is modified in MATLAB software, the electromagnetic parameters, workpiece parameters (geometric parameters), and process parameters (structural boundary conditions) are parameterized, and the For loop conditions are established. The maximum stress and maximum deformation are obtained through the mphmax function. The 'solid.mises' parameter in the mphmax function is defined to obtain the maximum stress of the workpiece. The 'solid.disp' parameter in the mphmax function is defined to obtain the maximum deformation of the workpiece. The results are saved line by line in a ".txt" file through the fprintf function. The MATLAB output is:
[0070] OutputMATLAB =[DMAX i ,SMAX i ](i=1,2,...,n)
[0071] Among them, DMAX i is the maximum deformation, SMAX i is the maximum stress.
[0072] S13: Finally, solve the problem. Run MATLAB software and call the COMSOL solver in real time to get the input [A i ,T i ,W i ,F i ,f i ,H i ]The corresponding output result [DMAX i ,SMAX i ].
[0073] Furthermore, the specific implementation steps of S2 are:
[0074] S21: Building VAE-GAN, see Figure 4 , VAE-GAN contains an encoder, a generator, and a discriminator, where the input of the encoder is:
[0075] Input Encoder =[A i ,T i ,W i ,F i ,f i ,H i ,DMAX i ,SMAX i ]
[0076] The output of the encoder is:
[0077] Output Encoder =[Z 1i ,Z 2i ]
[0078] where Z 1i ,Z 2i is an intermediate variable.
[0079] The input to the generator is:
[0080] Input Generator =[Z 1i ,Z 2i ]
[0081] The output of the generator is:
[0082] OutputGenerator =[A' i ,T' i ,W' i ,F' i ,f' i ,H' i ,DMAX' i ,SMAX' i ]
[0083] where A' i ,T' i ,W' i ,F' i ,f' i ,H' i ,DMAX' i ,SMAX' i Similar to A i ,T i ,W i ,F i ,f i ,H i ,DMAX i ,SMAX i A new set of parameters.
[0084] The input of the discriminator is divided into two parts, one from the input of the encoder and the other from the output of the generator:
[0085] Input Discriminator-Encoder =[A i ,T i ,W i ,F i ,f i ,H i ,DMAX i ,SMAX i ]
[0086] Input Discriminator-Generator =[A' i ,T' i ,W' i ,F' i ,f' i ,H' i ,DMAX' i ,SMAX' i ]
[0087] The output of the discriminator is the probability of identifying the encoder input data, which is between 0 and 1.
[0088] The encoder, generator, and discriminator are all deep neural networks. The input layer of the encoder contains 8 neurons (or 6 neurons when F i ,fi (When simplified to a fixed support, the output layer contains 2 neurons. The generator's input layer contains 2 neurons, and the output layer contains 8 neurons (or 6 neurons). The discriminator's input layer contains 16 neurons (or 12 neurons), and the output layer contains 1 neuron. The number of hidden layers and hidden layer neurons in the encoder, generator, and discriminator depends on the amount of data n.
[0089] S22: Train VAE-GAN. The VAE reconstruction loss part of the VAE-GAN loss function is:
[0090] Loss reconstruction →||Output Generator -G(E(Input Encoder ))||
[0091] Where G() is the generator, E() is the encoder, Output Generator is the output of the generator, Input Encoder is the input of the encoder.
[0092] In the second part, the GAN part loss of the VAE-GAN loss function is:
[0093]
[0094] Where D() is the discriminator, Z p1 ,Z p2 It is also an intermediate variable, obtained by resampling in the normal distribution space, Input Encoder is the input of the encoder, Output Encoder is the output of the encoder, G() is the generator, and E() is the encoder.
[0095] S23: Finally, the trained VAE-GAN obtains the output [A' i ,T' i ,W' i ,F' i ,f' i ,H' i ,DMAX' i ,SMAX' i ].
[0096] Furthermore, the specific implementation steps of S3 are:
[0097] S31: The input obtained by S13 [A i ,T i ,W i ,F i ,f i ,H i] and output [DMAX i ,SMAX i ] data and the output data [A' i ,T' i ,W' i ,F' i ,f' i ,H' i ,DMAX' i ,SMAX' i ] to splice and get:
[0098] [[A i ,T i ,W i ,F i ,f i ,H i ,DMAX i ,SMAX i ],...,[A' i ,T' i ,W' i ,F' i ,f' i ,H' i ,DMAX' i ,SMAX' i ]]
[0099] S32: Adjust the input-output relationship and divide the data into two groups, LSTM1 and LSTM2, namely the first training sample and the second training sample, which are used for subsequent training of LSTM1 and LSTM2 networks respectively:
[0100] LSTM1:[[A i ,T i ,W i ,F i ,f i ,DMAX i ,H i ],...,[A' i ,T' i ,W' i ,F' i ,f' i ,DMAX' i ,H' i ]]
[0101] LSTM2:[[A i ,T i ,W i ,F i ,f i ,DMAX i ,SMAX i],...,[A' i ,T' i ,W' i ,F' i ,f' i ,DMAX' i ,SMAX' i ]]
[0102] The last column is used as the output data of LSTM, and the remaining columns are input data.
[0103] Furthermore, the specific implementation steps of S4 are:
[0104] S41: Construct LSTM1 and LSTM2 (i.e., the first and second sub-models). Both networks complete the output after passing through a fully connected network. LSTM1 implements the prediction based on electromagnetic parameters A, workpiece parameters T, W, process parameters F, f, and the maximum deformation DMAX of the composite material to the adjustment parameter H. The input data of LSTM1 is:
[0105] [[A i ,T i ,W i ,F i ,f i ,DMAX i ],...,[A' i ,T' i ,W' i ,F' i ,f' i ,DMAX' i ]]
[0106] The output data is:
[0107] [[H i ],...,[H' i ]]
[0108] LSTM2 implements the prediction based on electromagnetic parameters A, workpiece parameters T, W, process parameters F, f, and the maximum deformation DMAX to maximum stress SMAX of the composite material. The input data of LSTM2 is:
[0109] [[A i ,T i ,W i ,F i ,f i ,DMAX i ],...,[A' i ,T' i ,W' i ,F' i ,f' i ,DMAX'i ]]
[0110] The output data is:
[0111] [[SMAX i ],...,[SMAX' i ]]
[0112] S42: Train LSTM1 and LSTM2. The loss functions of LSTM1 and LSTM2 are both MSE loss:
[0113]
[0114] Among them, H i To adjust the parameters, that is, the distance from the coil to the workpiece (represented by the coil to the lowest point of the mold), H' i Generate parameters for the corresponding generator, H″ i is the LSTM1 prediction parameter, and m is the amount of LSTM data.
[0115]
[0116] Among them, SMAX i is the maximum stress of the workpiece, SMAX′ i Generate stress for generator, SMAX″ i The stress prediction of LSTM2 is m, and the amount of data of LSTM is m.
[0117] Based on the above steps of the MC-VAE-GAN-LSTM model, see 5 and Figure 6 , then the step-by-step electromagnetic forming process of composite materials includes the following:
[0118] S1: Input electromagnetic parameters A, workpiece parameters T (thickness), W (width), process parameters F (clamp clamping force), f (friction), and the current actual measured maximum deformation of the workpiece DMAX into the MC-VAE-GAN-LSTM model to obtain the predicted H (height adjustment parameter, the distance from the coil to the processed workpiece, represented by the distance from the coil to the lowest point of the mold) and the maximum stress SMAX of the workpiece.
[0119] S2: Evaluate whether the workpiece's maximum stress, SMAX, is within the permitted range for composite material processing. If so, the obtained H is input into the motor controller, which controls the screw lift to move the height of H, achieving one-step forming.
[0120] S3: The process of S1-S2 is repeated to realize the next forming process until the composite sheet is formed (reaches the bottom of the mold). The data is stored and interacted with in real time through the database.
[0121] Furthermore, the specific implementation steps of S2 are:
[0122] S21: Control is performed using a linear quadratic regulator (LQR). Define the system's state equation, which includes a dynamic model for the effects of gravity and damping on the screw elevator. MC-VAE-GAN-LSTM predicts the current height value: H", and defines the target height value:
[0123] H target =H”
[0124] The dynamic behavior of the screw lift is described by establishing the following state space model:
[0125]
[0126] Where H(t) is the height of the screw lift at time t, Represents the rate of change of height, u(t) is the control input, A is the state matrix, and B is the input matrix.
[0127] Considering the effects of gravity and damping, the system is considered to be an approximate second-order system:
[0128]
[0129] Where m is the mass of the platform, δ is the damping coefficient, and k is the spring stiffness.
[0130] Then, convert the second-order differential equation into state-space form:
[0131]
[0132] Define the state vector:
[0133]
[0134] S22: Design the LQR controller. The control objective of the system is to design a linear feedback controller u(t) = -Kx(t) with a gain matrix K so that the state x(t) of the system reaches the target state x Target , or stabilize at the target state x Target Nearby. x Target for:
[0135]
[0136] The cost function of LQR control is:
[0137] ∫0 ∞ (x T Qx+u T Ru)dt
[0138] Where Q and R are diagonal matrices.
[0139] S23: Solve the gain matrix K. First solve the ARE equation:
[0140] A T P+PA-PBR -1 B T P+Q=0
[0141] The auxiliary matrix P is obtained by the ARE equation, and K is obtained by solving the specified A, B, Q, and R:
[0142] K=R -1 B T P
[0143] S24: The height is controlled to reach the ideal value H obtained by the MC-VAE-GAN-LSTM neural network model. The relationship between the magnitude of the electromagnetic force and H is generally:
[0144]
[0145] Therefore, when the distance H between the composite material workpiece and the discharge coil changes, the electromagnetic force on the workpiece changes, so that the electromagnetic force can be controlled in steps.
[0146] The magnitude of the electromagnetic force is obtained from the following relationship:
[0147] F=J×B
[0148] Where J is the induced eddy current and B is the magnetic flux density.
[0149] The embodiment of the present invention provides a step-by-step electromagnetic forming method for composite materials, which uses the first network model to perform finite element simulation on the electromagnetic forming process of the workpiece to obtain multiple sets of simulation data; each set of simulation data includes electromagnetic parameters, workpiece parameters, process parameters, adjustment parameters, and the corresponding maximum deformation and maximum stress of the workpiece; the multiple sets of simulation data are input into the second network model to obtain similar simulation data generated by the second network model; the simulation data and the similar simulation data are spliced and then processed in data format to obtain training samples; the training samples are used to train the third network model, and the trained third network model is used to predict the electromagnetic parameters, workpiece parameters, process parameters, and maximum deformation and maximum stress of the workpiece. The training process comprises the following steps: determining the electromagnetic parameters, workpiece parameters, process parameters, and the mapping relationship between the maximum deformation and the adjustment parameters, and determining the mapping relationship between the electromagnetic parameters, workpiece parameters, process parameters, and the maximum deformation and the maximum stress; judging whether the processing permission range of the workpiece is met based on the maximum stress; controlling the screw elevator to adjust the distance between the discharge coil and the workpiece so that the distance between the discharge coil and the workpiece is the adjustment parameter; repeating the above operation multiple times until the workpiece reaches the bottom of the base mold, thereby completing the electromagnetic forming of the workpiece. The solution provided by the present invention establishes the relationship between multiple parameters through finite element simulation and neural network architecture, realizes intelligent control of the distance between the coil and the workpiece and stress prediction, and realizes multiple step-by-step forming through multiple adjustments.
[0150] In order to implement the method of an embodiment of the present invention, an embodiment of the present invention also provides a step-by-step electromagnetic forming system for composite materials, wherein the step-by-step electromagnetic forming system for composite materials includes a movable platform, a composite network model and a processor; the movable platform includes a base mold, a discharge coil and a screw lift; the base mold is used to place the workpiece for processing; the discharge coil is used to generate electromagnetic force so that the workpiece is formed under the action of the electromagnetic force; the screw lift is used to adjust the distance from the discharge coil to the workpiece for processing; the composite network model includes a first network model, a second network model and a third network model; the processor is used to execute the steps of any of the above methods.
[0151] To implement the method of an embodiment of the present invention, an embodiment of the present invention further provides a computer program product. The computer program product includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the steps of the above method.
[0152] Based on the hardware implementation of the above program modules, and in order to implement the method of the embodiment of the present invention, the embodiment of the present invention further provides an electronic device (computer device). Specifically, in one embodiment, the computer device can be a terminal, and its internal structure diagram can be as follows: Figure 7 As shown. The computer device includes a processor A01, a network interface A02, a display screen A04, an input device A05 and a memory (not shown in the figure) connected via a system bus. Among them, the processor A01 of the computer device is used to provide computing and control capabilities. The memory of the computer device includes an internal memory A03 and a non-volatile storage medium A06. The non-volatile storage medium A06 stores an operating system B01 and a computer program B02. The internal memory A03 provides an environment for the operation of the operating system B01 and the computer program B02 in the non-volatile storage medium A06. The network interface A02 of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor A01, the method of any one of the above embodiments is implemented. The display screen A04 of the computer device can be a liquid crystal display or an electronic ink display screen, and the input device A05 of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad provided on the computer device housing, or an external keyboard, touchpad or mouse, etc.
[0153] Those skilled in the art will understand that Figure 7 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.
[0154] The device provided by an embodiment of the present invention includes a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, the method of any one of the above embodiments is implemented.
[0155] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.
[0156] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes 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 steps in the process. 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.
[0157] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work 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 The function specified in one or more boxes.
[0158] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.
[0159] In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.
[0160] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.
[0161] Computer-readable media includes permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. The information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory computer-readable media (transitory media), such as modulated data signals and carrier waves.
[0162] It is understood that the memory of the embodiments of the present invention can be volatile memory or non-volatile memory, or can include both volatile and non-volatile memory. Among them, the non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), magnetic random access memory (FRAM), flash memory, magnetic surface memory, optical disk, or compact disc read-only memory (CD-ROM); magnetic surface memory can be magnetic disk memory or tape memory. Volatile memory can be random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of RAM are available, such as static random access memory (SRAM), synchronous static random access memory (SSRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDRSDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM), and direct rambus random access memory (DRRAM).The memories described in the embodiments of the present invention are intended to include, but are not limited to, these and any other suitable types of memories.
[0163] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity, or apparatus that includes the element.
[0164] The above are merely embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should all be included within the scope of the claims of the present application.
Claims
1. A composite material step-by-step electromagnetic forming method, characterized in that: The method is applied to a composite material step-by-step electromagnetic forming system, which comprises a movable platform and a composite network model; the movable platform comprises a base mold, a discharge coil and a screw lifter; the base mold is used to provide a forming shape for a workpiece; the discharge coil is used to generate an electromagnetic force so that the workpiece is formed under the action of the electromagnetic force; The screw lift is used to adjust the distance from the discharge coil to the workpiece; The composite network model includes a first network model, a second network model and a third network model; the method includes: Using the first network model, finite element simulation is performed on the electromagnetic forming process of the workpiece to obtain multiple sets of simulation data; each set of simulation data includes electromagnetic parameters, workpiece parameters, process parameters, adjustment parameters, and the corresponding maximum deformation and maximum stress of the workpiece; Inputting the plurality of sets of simulation data into the second network model to obtain similar simulation data generated by the second network model; The simulation data and the similar simulation data are spliced and then processed in data format to obtain training samples; The third network model is trained by using the training samples, and the adjusted parameters and the maximum stress of the processed workpiece under the conditions of electromagnetic parameters, workpiece parameters, process parameters and maximum deformation are predicted by using the trained third network model; the trained third network model includes a mapping relationship between electromagnetic parameters, workpiece parameters, process parameters, maximum deformation and adjusted parameters, and a mapping relationship between electromagnetic parameters, workpiece parameters, process parameters, maximum deformation and maximum stress; Judging whether the maximum stress meets the permitted processing range of the workpiece according to the maximum stress; if the maximum stress meets the permitted processing range of the workpiece, controlling the screw lifter to adjust the distance from the discharge coil to the workpiece so that the distance from the discharge coil to the workpiece is the adjustment parameter; The above operation is repeated for multiple times until the workpiece reaches the bottom of the base mold, and the electromagnetic forming of the workpiece is completed.
2. The composite material step-by-step electromagnetic forming method according to claim 1, characterized in that: The first network model includes a first script and a first solver; the first network model is used to perform finite element simulation on the electromagnetic forming process of the workpiece to obtain multiple sets of simulation data, including: Establishing an initial simulation process through the first software to obtain a first script; Running the first script in the first solver, setting parameters and cycle conditions, inputting multiple sets of electromagnetic parameters, workpiece parameters, process parameters and adjustment parameters, and obtaining the corresponding output maximum stress and maximum deformation; Based on multiple sets of electromagnetic parameters, workpiece parameters, process parameters, adjustment parameters, corresponding maximum stresses and maximum deformations, multiple sets of simulation data are obtained.
3. The composite material step-by-step electromagnetic forming method according to claim 1, characterized in that: The second network model includes an encoder, a generator, and a discriminator. The plurality of sets of simulation data are input into the second network model to obtain similar simulation data generated by the second network model, including: Inputting the plurality of sets of simulation data into the encoder to obtain intermediate variables output by the encoder; Input the intermediate variable into the generator to obtain similar simulation data output by the generator; The simulation data and the similar simulation data are input into the discriminator to obtain the identification probability output by the discriminator; the identification probability is used to identify whether the similar simulation data meets the preset requirements, and the similar simulation data that meets the preset requirements is output.
4. The composite material step-by-step electromagnetic forming method according to claim 3, characterized in that: The encoder, the generator and the discriminator are all deep neural networks, the input layer of the encoder contains 8 neurons or 6 neurons, and the output layer of the encoder contains 2 neurons; the input layer of the generator contains 2 neurons, and the output layer of the generator contains 8 neurons or 6 neurons; the input layer of the discriminator contains 16 neurons or 12 neurons, and the output layer of the discriminator contains one neuron; the hidden layers of the encoder, the generator and the discriminator and the number of neurons in the hidden layers are positively correlated with the data volume of the simulation data.
5. The composite material step-by-step electromagnetic forming method according to claim 3, characterized in that: The loss function of the second network model during training includes a first reconstruction loss part and a second part loss: The first reconstruction loss part includes: Loss reconstruction →||Output Generator -G(E(Input Encoder ))|| Among them, G() is the generator, E() is the encoder, Output Generator is the output of the generator, Input Encoder is the input of the encoder; The second part of the loss includes: Loss GAN →log(D(Input Encoder ))+log(1-D(G(E(Output Encoder ))))+log(1-D(G(Z p1 ,Z p2 ))) Among them, D() is the discriminator, Z p1 ,Z p2 is an intermediate variable, obtained by resampling in the normal distribution space, Input Encoder is the input of the encoder, Output Encoder is the output of the encoder, G() is the generator, and E() is the encoder.
6. The composite material step-by-step electromagnetic forming method according to claim 1, characterized in that: The simulation data and the similar simulation data are spliced and then processed in data format to obtain training samples, including: splicing the simulation data and the similar simulation data row by row to obtain spliced data; The input-output relationship of the spliced data is adjusted to obtain a training sample; the training sample includes a first training sample and a second training sample; the input items of the first training sample are electromagnetic parameters, workpiece parameters, process parameters and maximum deformation, and the output item of the first training sample is an adjustment parameter; the input items of the second training sample are electromagnetic parameters, workpiece parameters, process parameters and maximum deformation, and the output item of the second training sample is maximum stress.
7. The composite material step-by-step electromagnetic forming method according to claim 6, characterized in that: The third network model includes a first sub-model and a second sub-model; and using the training sample to train the third network model includes: Training the first sub-model using the first training sample; The second sub-model is trained using the second training sample; the trained first sub-model includes a mapping relationship between electromagnetic parameters, workpiece parameters, process parameters, and maximum deformation to adjustment parameters; the trained second sub-model includes a mapping relationship between electromagnetic parameters, workpiece parameters, process parameters, and maximum deformation to maximum stress.
8. The composite material step-by-step electromagnetic forming method according to claim 7, characterized in that: The loss function of the first sub-model during training is: Among them, H i To adjust the parameters, that is, the distance from the coil to the workpiece (the distance from the coil to the lowest point of the mold), H' i Generate parameters for the corresponding generator, H″ i is the prediction parameter of LSTM1, and m is the amount of data of LSTM; The loss function of the second sub-model during training is: Among them, SMAX i is the maximum stress of the workpiece, SMAX′ i Generate stress for generator, SMAX″ i The stress prediction of LSTM2, m is the amount of data of LSTM.
9. The composite material step-by-step electromagnetic forming method according to claim 1, characterized in that: Controlling the screw lifter to adjust the distance from the discharge coil to the workpiece includes: The linear quadratic regulator is used to control the screw lift to adjust the distance from the discharge coil to the workpiece.
10. A composite material step-by-step electromagnetic forming system, characterized in that: The composite material step-by-step electromagnetic forming system includes a movable platform, a composite network model and a processor; the movable platform includes a base mold, a discharge coil and a screw lift; the base mold is used to place a workpiece for processing; the discharge coil is used to generate electromagnetic force so that the workpiece is formed under the action of the electromagnetic force; the screw lift is used to adjust the distance from the discharge coil to the workpiece for processing; the composite network model includes a first network model, a second network model and a third network model; the processor is used to execute the steps of the method described in any one of claims 1 to 9.
Citation Information
Patent Citations
Workpiece forming control method for separating workpiece electromagnetic force based on simulating optimization
CN111468588A
Method for realizing gradual electromagnetic forming sequence simulation based on LS-DYNA
CN116562056A
Cited By
Electromagnetic forming numerical simulation method and system based on multi-fidelity data
CN121145628A
A multi-fidelity data based electromagnetic forming numerical simulation method and system
CN121145628B
Dynamic PBR material generation method and system based on state space model
CN122244271A
A dynamic PBR material generation method and system based on a state space model
CN122244271B
