Creep aging forming method and device based on digital twin technology

Abstract

The present invention discloses a creep aging forming method and device based on digital twin technology. The method includes collecting sensor data of aluminum alloy siding during creep aging forming; inputting the sensor data into a first prediction model to obtain the predicted springback and yield strength output by the first prediction model; comparing the predicted springback and yield strength with the target springback and yield strength to determine the first difference between the predicted springback and yield strength and the target springback and yield strength; based on the first difference, using a second prediction model, determining the first process parameter for updating the process of the aluminum alloy siding, sending the first process parameter to an autoclave, and controlling the autoclave to perform process operations with the first process parameter. The solution provided by the present invention can adjust the process based on real-time sensor data to obtain components with precise forming and target performance.

Description

Technical Field

[0001] The present invention relates to the technical field of creep aging forming, and in particular to a creep aging forming method and device based on digital twin technology. Background Art

[0002] Large thin-walled components are important components and key load-bearing structures of aerospace equipment, accounting for more than 40% of the weight of the fuselage / rocket body structure. Creep aging forming technology is often used in the forming of large thin-walled components. Its advantage is that it can simultaneously form and improve the strength of the components, so it is widely used in the manufacturing of large thin-walled components in aerospace.

[0003] For creep aging forming of large thin-walled components, the component needs to be placed on the mold, and the vacuum bag shrinks during the vacuuming process at room temperature to make the wall panel fit the mold as much as possible. Then the mold and component are placed in the autoclave and heated and pressurized until the component is attached to the mold, and then the creep aging forming stage begins. However, after entering the autoclave, due to the high temperature and high pressure environment, it is impossible to know in real time when the mold is attached in the actual situation and whether the component deflects during the die attach process. It can only be known after the creep aging forming process is completed and the autoclave is opened. Under certain conditions, the entire creep forming process takes more than ten hours. Therefore, once the actual die attach situation does not match the pre-simulation, the pre-set process will not be able to achieve the forming accuracy. Summary of the invention

[0004] In order to solve the technical problem that the existing creep aging forming process cannot adjust the pre-set process, resulting in the performance of large thin-walled components failing to reach the target accuracy, an embodiment of the present invention provides a creep aging forming method and device based on digital twin technology.

[0005] The technical solution of the embodiment of the present invention is achieved as follows:

[0006] The embodiment of the present invention provides a creep aging forming method based on digital twin technology, the method comprising: collecting sensor data of an aluminum alloy siding during creep aging forming; inputting the sensor data into a first prediction model to obtain a predicted springback and yield strength output by the first prediction model; the first prediction model is a prediction model obtained by using a first training method for an initial network model, and the initial network model is an LSTM algorithm model; the input of the first prediction model is time, temperature, heating rate and pressure, and the output of the first prediction model is the predicted springback and yield strength of the aluminum alloy siding; the predicted springback and yield strength are compared with the target springback and yield strength. The springback amount and yield strength are compared to determine the first difference between the predicted springback amount and yield strength and the target springback amount and yield strength; based on the first difference, a second prediction model is used to determine the first process parameter for process update of the aluminum alloy siding, the first process parameter is sent to the autoclave, and the autoclave is controlled to perform process operation with the first process parameter; the second prediction model is a prediction model obtained by using the second training method for the initial network model; the input of the second prediction model is the process update moment, time, temperature, heating rate and pressure, and the output of the second prediction model is the predicted springback amount and yield strength of the aluminum alloy siding.

[0007] In one embodiment, obtaining a first prediction model includes: conducting a specimen-level creep aging forming test of aluminum alloy to obtain a sample data set; constructing a material constitutive equation using the sample data set; based on the material constitutive equation, performing a finite element simulation on a creep aging forming process of a component-level aluminum alloy siding to obtain a correspondence between strain and displacement of characteristic points of the aluminum alloy siding at different times; determining initial process parameters of the aluminum alloy siding during the finite element simulation and the springback and yield strength at the time of mold application based on the correspondence between strain and displacement of the characteristic points of the aluminum alloy siding at different times; using the initial process parameters and the springback and yield strength at the time of mold application as the first training sample, training the initial network model, and obtaining a first prediction model.

[0008] In one embodiment, the material constitutive equation includes:

[0009] ε1=f(σ,T,t,γ)=f(σ)f(T)f(γ)f(t) Calculation formula (1)

[0010] Among them, ε1 is the creep strain; σ is the applied creep stress; T is the creep temperature; γ is the creep heating rate; t is the creep time, and f(σ), f(T), f(γ), and f(t) are the creep stress, temperature, heating rate, and time, respectively.

[0011] In one embodiment, the obtaining of the sample data set includes: conducting a specimen-level creep aging forming test of aluminum alloy to obtain an initial sample data set; determining an optimal setting range of the K value in the K nearest neighbor algorithm; setting the K value in the K nearest neighbor algorithm based on the optimal setting range, and performing data expansion on the initial sample data set based on the K nearest neighbor algorithm after the K value is set to obtain a new sample data set.

[0012] In one embodiment, the corresponding relationship between the strain and displacement of the characteristic point of the aluminum alloy wall panel at different times includes:

[0013] Calculation formula (2)

[0014] in, is the strain of the characteristic points of the aluminum alloy siding at different times, is the distance between the initial position of the feature point and the mold, is the distance between the characteristic point of the aluminum alloy siding and the mold at different times.

[0015] In one embodiment, the training of the initial network model to obtain the first prediction model includes: determining that the algorithm optimizer of the LSTM algorithm model is SGD, Adam, Adamax or RMSpropadam, and the activation function of the LSTM algorithm model is Sigmiod, Tanh or ReLU; using the algorithm optimizer and the activation function as model parameters, training the initial network model to obtain the first prediction model.

[0016] In one embodiment, a second prediction model is obtained, including: in a finite element simulation process, taking the moment when room temperature loading is completed as the initial moment, and the moment when the aluminum alloy wall panel is fully fitted with the mold as the end moment, and according to a preset process update frequency, determining multiple process update moments between the initial moment and the end moment to obtain a process update moment sequence; obtaining finite element simulation process data corresponding to each process update moment in the process update moment sequence; using the process update moment sequence and the finite element simulation process data corresponding to each process update moment in the process update moment sequence as second training data, training the initial network model, and obtaining a second prediction model.

[0017] In one embodiment, the determining of multiple process update moments includes: assuming that the moment when room temperature loading is completed is the initial moment T1, the moment when the aluminum alloy wall panel is completely fitted with the mold after being placed in the autoclave for high temperature loading is the end moment t1, the time between T1 and t1 is T, and the process update frequency is f;

[0018] Then the process update time T2 is:

[0019] T2= T×f+T1;

[0020] The process update time T3 is:

[0021] T3= T×f+T2;

[0022] By analogy, the process update time Tn is:

[0023] Tn= T×f+Tn-1.

[0024] In one embodiment, after determining the first process parameter for updating the process of the aluminum alloy siding, the method further includes: collecting sensor data of the aluminum alloy siding during the creep aging forming process again after each preset time interval; inputting the sensor data into the first prediction model to obtain the predicted springback and yield strength output by the first prediction model; comparing the predicted springback and yield strength with the target springback and yield strength to determine a first difference between the predicted springback and yield strength and the target springback and yield strength; based on the first difference, using the second prediction model, determining the first process parameter for updating the process of the aluminum alloy siding, sending the first process parameter to the autoclave, and controlling the autoclave to perform the process operation with the first process parameter, and repeating the above operation until the predicted springback and yield strength are no different from the target springback and yield strength.

[0025] An embodiment of the present invention also provides a creep aging forming device based on digital twin technology, and the creep aging forming device based on digital twin technology includes: a sealed vacuum bag, an autoclave, a high-temperature strain gauge, a temperature sensor, a pressure sensor and a processor; the sealed vacuum bag is connected to the high-temperature strain gauge, the temperature sensor and the pressure sensor, and is used to shrink during the vacuuming process to make the component fit the mold; the autoclave is connected to the high-temperature strain gauge, the temperature sensor and the pressure sensor, and is used to make the component enter the creep aging forming stage after being fit to the mold under heating and pressurizing conditions; the high-temperature strain gauge is used to collect status data of the component; the temperature sensor is used to collect temperature data, and the pressure sensor is used to collect pressure data; the processor is used to execute the steps of the above-mentioned method.

[0026] The creep aging forming method and device based on digital twin technology provided in the embodiment of the present invention collects sensor data of aluminum alloy siding during creep aging forming; inputs the sensor data into a first prediction model to obtain predicted springback and yield strength output by the first prediction model; compares the predicted springback and yield strength with the target springback and yield strength to determine a first difference between the predicted springback and yield strength and the target springback and yield strength; based on the first difference, using a second prediction model, determines a first process parameter for updating the process of the aluminum alloy siding, sends the first process parameter to an autoclave, and controls the autoclave to perform process operations with the first process parameter. The solution provided by the present invention can adjust the process based on real-time sensor data to obtain components with precise forming and target performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 It is a flow chart of a creep aging forming method based on digital twin technology according to an embodiment of the present invention;

[0028] Figure 2 A schematic diagram of the digital twin system flow established for an embodiment of the present invention;

[0029] Figure 3 1 is a diagram showing the internal structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0030] The present invention will be described in further detail below with reference to the accompanying drawings and embodiments.

[0031] The embodiment of the present invention provides a creep aging forming method based on digital twin technology, such as Figure 1 As shown, the method includes:

[0032] Step 101: collecting sensor data of the aluminum alloy siding during creep aging forming;

[0033] Step 102: input the sensor data into a first prediction model to obtain predicted springback and yield strength output by the first prediction model; the first prediction model is a prediction model obtained by using a first training method for an initial network model, and the initial network model is an LSTM algorithm model; the input of the first prediction model is time, temperature, heating rate and pressure, and the output of the first prediction model is predicted springback and yield strength of the aluminum alloy siding;

[0034] Step 103: comparing the predicted springback amount and yield strength with the target springback amount and yield strength to determine a first difference value between the predicted springback amount and yield strength and the target springback amount and yield strength;

[0035] Step 104: Based on the first difference value, a second prediction model is used to determine a first process parameter for updating the process of the aluminum alloy siding, the first process parameter is sent to the autoclave, and the autoclave is controlled to perform the process operation with the first process parameter; the second prediction model is a prediction model obtained by using a second training method for the initial network model; the input of the second prediction model is the process update moment, time, temperature, heating rate and pressure, and the output of the second prediction model is the predicted springback and yield strength of the aluminum alloy siding.

[0036] The method proposed in this embodiment collects the process data of the component in the autoclave by introducing high-temperature strain gauges, temperature sensors, pressure sensors and other collection devices during the sealing of the vacuum bag. The process data can reflect the real-time state of the component during the forming process in the autoclave, and is used to construct a prediction model with strain data, actual heating rate, die-attachment time, pressure data, etc. as inputs and the rebound amount and mechanical properties of the component after forming as outputs. Since the prediction data in this embodiment is obtained based on the real-time state data of the component, the prediction result is relatively accurate. And by adjusting the process using the prediction data, the precise forming and target performance of the component can be guaranteed.

[0037] Specifically, in one embodiment, the process of acquiring the first prediction model is: conducting a specimen-level creep aging forming test of aluminum alloy to obtain a sample data set; using the sample data set to construct a material constitutive equation; based on the material constitutive equation, performing a finite element simulation on the creep aging forming process of the component-level aluminum alloy siding to obtain the corresponding relationship between the strain and displacement of the characteristic points of the aluminum alloy siding at different times; based on the corresponding relationship between the strain and displacement of the characteristic points of the aluminum alloy siding at different times, determining the initial process parameters of the aluminum alloy siding during the finite element simulation process and the rebound and yield strength at the mold attaching moment; using the initial process parameters and the rebound and yield strength at the mold attaching moment as the first training sample, training the initial network model to obtain the first prediction model.

[0038] The sample data in this embodiment are the time, temperature, heating rate, stress, creep amount, springback amount and yield strength data obtained by performing creep aging forming test on standard creep specimens made of aluminum alloy.

[0039] In this embodiment, a creep aging forming test is conducted on a standard creep specimen made of aluminum alloy, with time, temperature, heating rate, and stress as inputs and creep amount as output. After expanding the sample data using machine learning methods, a material constitutive equation is established. Based on the constitutive equation, finite element simulation at the component level is performed later.

[0040] Here, the material constitutive equation is:

[0041] ε1=f(σ,T,t,γ)=f(σ)f(T)f(γ)f(t) Calculation formula (1)

[0042] Among them, ε1 is the creep strain; σ is the applied creep stress; T is the creep temperature; γ is the creep heating rate; t is the creep time, and f(σ), f(T), f(γ), and f(t) are the creep stress, temperature, heating rate, and time, respectively.

[0043] Here, the machine learning method is specifically to use the K nearest neighbor algorithm. That is, this embodiment specifically uses the K nearest neighbor algorithm to expand the sample data. When using the K nearest neighbor algorithm to expand the sample data, a creep aging forming test of aluminum alloy at the sample level is first performed to obtain an initial sample data set; the optimal setting range of the K value in the K nearest neighbor algorithm is determined; the K value in the K nearest neighbor algorithm is set based on the optimal setting range, and then the initial sample data set is expanded based on the K nearest neighbor algorithm after the K value is set to obtain a new sample data set.

[0044] Through comparison, this embodiment found that when the K value of the K-nearest neighbor algorithm is set to 3, 4, and 5, the root mean square error RMSE of the predicted 2219 aluminum alloy at different times, temperatures, heating rates, and stress magnitudes is smaller. Therefore, this embodiment can select a K value range of 3, 4, and 5. Using this range to expand the sample data can greatly reduce the amount of creep specimen testing, and can form a new sample data set with sufficient data.

[0045] The finite element simulation process will take the points of the aluminum alloy siding that are attached to the mold under normal temperature loading as characteristic points, establish a mathematical model for mapping the characteristic point strain and the siding forming state, and obtain the overall deformation state of the siding under different characteristic point strain conditions. That is, the corresponding relationship between the strain and displacement of the aluminum alloy siding characteristic points at different times will be obtained.

[0046] Here, the corresponding relationship between the strain and displacement of the characteristic point of the aluminum alloy wall panel at different times is:

[0047] Calculation formula (2)

[0048] in, is the strain of the characteristic points of the aluminum alloy siding at different times, is the distance between the initial position of the feature point and the mold, is the distance between the characteristic point of the aluminum alloy siding and the mold at different times.

[0049] After the finite element simulation process, a database can be established with time, temperature, heating rate, autoclave pressure and characteristic point strain as input, and the springback and yield strength of the aluminum alloy siding after creep forming as output, and a prediction model can be established through a deep learning algorithm. The prediction model includes a first prediction model and a second prediction model.

[0050] In one embodiment, the initial network model is trained to obtain a first prediction model, including: determining that the algorithm optimizer of the LSTM algorithm model is SGD, Adam, Adamax or RMSpropadam, and the activation function of the LSTM algorithm model is Sigmiod, Tanh or ReLU; using the algorithm optimizer and the activation function as model parameters, the initial network model is trained to obtain the first prediction model.

[0051] The process of establishing the second prediction model includes: in the finite element simulation process, the moment when room temperature loading is completed is taken as the initial moment, and the moment when the aluminum alloy wall panel is fully fitted with the mold is taken as the end moment, and according to the preset process update frequency, multiple process update moments are determined between the initial moment and the end moment to obtain a process update moment sequence; the finite element simulation process data corresponding to each process update moment in the process update moment sequence is obtained; the process update moment sequence and the finite element simulation process data corresponding to each process update moment in the process update moment sequence are used as the second training data to train the initial network model to obtain the second prediction model.

[0052] Here, multiple process update moments are determined, including: the moment when room temperature loading is completed is set as the initial moment T1, the moment when the aluminum alloy wall panel is placed in the autoclave for high-temperature loading and the aluminum alloy wall panel is completely fitted with the mold is set as the end moment t1, the time between T1 and t1 is T, and the process update frequency is f;

[0053] Then the process update time T2 is:

[0054] T2= T×f+T1;

[0055] The process update time T3 is:

[0056] T3= T×f+T2;

[0057] By analogy, the process update time Tn is:

[0058] Tn= T×f+Tn-1.

[0059] After completing the construction of the above prediction model, the process parameters in the creep aging forming process of the aluminum alloy wall panel can be adjusted according to the actual collected data of the aluminum alloy wall panel. That is, in this embodiment, the high-temperature strain gauge can be attached to the un-molded position of the wall panel component, and then loaded by compressing the vacuum bag, inputting the temperature sensor and pressure sensor attached to the mold at the same time, first loading at room temperature outside the autoclave, and then inputting the previously determined initial process parameters, and then putting it into the autoclave for high-temperature loading, and starting the creep aging forming test of the wall newspaper component. Then, the autoclave, mold and wall panel model are established through the digital twin software Unity3d on the computer side, and the strain gauge collection data, temperature sensor, and pressure sensor data are connected to the computer side to construct a digital twin model of creep aging forming of aluminum alloy wall panels. Then, through the data collected in real time by the sensor, the rebound amount and yield strength after the creep forming of the wall panel are predicted, and the process is adjusted in real time according to the target rebound amount and yield strength.

[0060] In the actual adjustment process, multiple adjustments can be made. That is, in one embodiment, after determining the first process parameter for updating the process of the aluminum alloy siding, the sensor data of the aluminum alloy siding during the creep aging forming process can be collected again after each preset time interval; the sensor data is input into the first prediction model to obtain the predicted springback and yield strength output by the first prediction model; the predicted springback and yield strength are compared with the target springback and yield strength to determine the first difference between the predicted springback and yield strength and the target springback and yield strength; based on the first difference, the second prediction model is used to determine the first process parameter for updating the process of the aluminum alloy siding, the first process parameter is sent to the autoclave, and the autoclave is controlled to perform the process operation with the first process parameter, and the above operation is repeated until the predicted springback and yield strength are no different from the target springback and yield strength.

[0061] Specifically, see Figure 2 , the creep aging forming process of 2219 aluminum alloy siding will be explained below.

[0062] The creep aging forming test of 2219 aluminum alloy at the sample level was carried out. With different time, temperature, heating rate, and stress as the input of the test, and the creep amount as the output of the test, the creep aging forming test of 2219 aluminum alloy at the sample level was carried out to obtain the initial sample-level database (i.e., the initial sample data set). The data was expanded by machine learning (such as the K nearest neighbor algorithm). After comparison, it was found that when the K value of the K nearest neighbor algorithm was set in the range of 3, 4, and 5, the root mean square error RMSE of the prediction of 2219 aluminum alloy at different times, temperatures, heating rates, and stress sizes was smaller. Selecting the K nearest neighbor algorithm with the optimal K value setting range (i.e., the K value setting range is 3, 4, and 5) to expand the sample data volume can greatly reduce the amount of creep aging forming tests, and can form a new sample data set with sufficient data volume.

[0063] Based on the sample data set, the material constitutive equation for creep quantity associated with time, temperature, heating rate, and stress magnitude is established:

[0064] ε1=f(σ,T,t,γ)=f(σ)f(T)f(γ)f(t);

[0065] In the formula, ε1 is the creep strain; σ is the applied creep stress; T is the creep temperature; γ is the creep heating rate; t is the creep time, and f(σ), f(T), f(γ), and f(t) are called the creep stress, temperature, heating rate, and time, respectively.

[0066] By introducing the material constitutive equation, the finite element simulation of the creep aging forming process of 2219 aluminum alloy at the wall panel component level is carried out to obtain the strain and displacement values ​​of the characteristic points at the same time. Through the finite element simulation, it is also possible to determine which characteristic points failed to be attached to the mold when the loading was completed at room temperature. Since the unattached points can still be displaced, the relationship between the strain ε2 and the real-time displacement L of these characteristic points can be established through the finite element simulation results:

[0067] ;

[0068] in, is the distance between the initial position of the feature point and the mold, and L is the distance between the feature point of the aluminum alloy siding and the mold at different times. Here, the real-time strain condition can be collected through the subsequent strain gauge to obtain the real-time displacement of the feature point before the mold is attached. When the real-time displacement L is no longer changing, it is the mold attachment moment.

[0069] In addition, here, the initial process parameters can be determined through finite element simulation according to the target springback amount and yield strength of the component, and the initial process parameters can be set as the initial process parameters of the autoclave creep aging forming process.

[0070] The time, temperature, heating rate, and autoclave pressure (i.e., initial process parameters) in the above finite element simulation process are used as the input of the initial network model (i.e., LSTM algorithm model), and the yield strength and springback of the component after final forming (i.e., the springback and yield strength at the time of die attachment) are used as the output of the initial network model. After learning and training, the first prediction model is obtained that can quickly predict the output (i.e., the yield strength and springback of the aluminum alloy siding) by inputting process parameters (i.e., time, temperature, heating rate, and autoclave pressure). Here, before learning and training, the data in the new sample data set can be normalized.

[0071] Here, the algorithm optimizer of the initial network model can be selected as any one of SGD, Adam, Adamax and RMSpropadam, and the activation function can be selected as any one of Sigmiod, Tanh and ReLU. Selecting the best algorithm optimizer and activation function can obtain the first prediction model with the highest accuracy.

[0072] Next, the moment when the room temperature loading is completed during the finite element simulation can be set as the initial moment T1, and the moment when the wall panel and the mold are completely fitted can be set as the end moment t1. The time from T1 to t1 is defined as T, and the process update frequency is set to f. According to the process update frequency, the time T can be divided into multiple process update moments T1, T2, T3...Tn (that is, a process update moment sequence is obtained).

[0073] The process update time T2 is:

[0074] T2= T×f+T1;

[0075] The process update time T3 is:

[0076] T3= T×f+T2;

[0077] By analogy, the process update time Tn is:

[0078] Tn= T×f+Tn-1.

[0079] The finite element simulation process data under these different process update times T1, T2, T3...Tn are obtained as the second training samples. After learning the initial network model, a second prediction model is obtained, which can quickly predict the output (i.e., the rebound and yield strength of the aluminum alloy siding) by inputting process parameters (i.e., process update time, time, temperature, heating rate and pressure).

[0080] The above is the process of establishing the first prediction model and the second prediction model. Based on the established first prediction model and the second prediction model, the creep aging forming process of the aluminum alloy wall panel can be controlled.

[0081] That is, first, a high-temperature strain gauge is attached to the position of the wall panel component that is not attached to the mold to monitor the mold attachment situation, and loading is performed by compressing the vacuum bag. At the same time, a temperature sensor and a pressure sensor are attached to the mold. Room temperature loading is first performed outside the autoclave, and the initial process parameters are input. The wall panel component is placed in the autoclave for high-temperature loading, and the creep aging forming test of the wall panel component is started.

[0082] Import the 3D CAD model of the autoclave, mold and wall panel model into Unity3d software, and integrate the strain gauge data, temperature sensor and pressure sensor data into Unity3d software through serial communication or intranet TCP communication. In Unity3d software, the actual pressure change, time change and heating rate in the tank can be observed in real time, and the real-time 3D state and stress distribution of the wall panel components can be dynamically displayed. The pseudo code is as follows:

[0083] float pressure = float.Parse(values[0]);

[0084] float temperature = float.Parse(values[1]);

[0085] float strain = float.Parse(values[2]);

[0086] float currentTime = Time.time - startTime;

[0087] float heatingRate = (temperature - lastTemperature) / (currentTime -lastTime);

[0088] lastTemperature = temperature;

[0089] lastTime = currentTime;

[0090] UpdateDisplay(pressure, temperature, strain, currentTime, heatingRate).

[0091] The data actually collected by the current sensor, i.e., the mold attachment situation, the pressure change in the tank, the time change, and the heating rate in the actual process, are displayed in the form of a table. At the same time, the first prediction model is called, and the mold attachment situation, the pressure change in the tank, the time change, and the heating rate in the actual process are input into the first prediction model to predict the rebound amount and yield strength of the formed component, and displayed in Unity3d in the form of a cloud map.

[0092] The predicted springback and yield strength are compared with the target springback and yield strength, and the comparison results are used to update the process parameters at different process update times T1, T2, T3...Tn. The process parameters at different process update times T1, T2, T3...Tn are updated repeatedly and iteratively until the updated process parameters at different process update times T1, T2, T3...Tn are input into the second prediction model, and the springback and yield strength of the formed component predicted by the second prediction model meet the target springback and yield strength, and the autoclave is controlled to perform process operations at different process update times according to the updated process parameters at the process update time. The predicted dynamic final target springback and yield strength are displayed in the form of a cloud chart. The pseudo code is as follows:

[0093] SelectNewProcess(actualSpringback, actualYieldStrength);

[0094] SendProcessUpdateToAutoclave(currentProcess);

[0095] float[,] predictionData = GetPredictionData();

[0096] cloudMap.UpdateCloudMap(predictionData);

[0097] string finalProcessName = "Process" + (iterationCount + 1);

[0098] finalProcessText.text = "Final actual process: " + finalProcessName.

[0099] The digital twin forming monitoring of 2219 aluminum alloy siding component-level creep aging forming can be completed.

[0100] The creep aging forming method based on digital twin technology provided in the embodiment of the present invention collects sensor data of the aluminum alloy wall panel during the creep aging forming process; inputs the sensor data into the first prediction model to obtain the predicted springback and yield strength output by the first prediction model; compares the predicted springback and yield strength with the target springback and yield strength to determine the first difference between the predicted springback and yield strength and the target springback and yield strength; based on the first difference, using the second prediction model, determines the first process parameter for updating the process of the aluminum alloy wall panel, sends the first process parameter to the autoclave, and controls the autoclave to perform the process operation with the first process parameter. The solution provided by the present invention can adjust the process based on real-time sensor data to obtain components with precise forming and target performance.

[0101] The embodiment of the present invention provides a creep aging forming device based on digital twin technology, and the creep aging forming device based on digital twin technology includes:

[0102] Sealed vacuum bags, autoclaves, high temperature strain gauges, temperature sensors, pressure sensors and processors;

[0103] The sealed vacuum bag is connected to the high-temperature strain gauge, the temperature sensor and the pressure sensor, and is used for shrinking during the vacuuming process to make the component fit the mold; the autoclave is connected to the high-temperature strain gauge, the temperature sensor and the pressure sensor, and is used for entering the creep aging forming stage after the component is fit to the mold under heating and pressurizing conditions; the high-temperature strain gauge is used to collect status data of the component; the temperature sensor is used to collect temperature data, and the pressure sensor is used to collect pressure data; the processor is used to execute the steps of the above-mentioned method.

[0104] It can be seen from the above embodiments that the solution of this embodiment has the following advantages:

[0105] 1. In the past, creep aging forming became a "black box problem" after the mold and the wall panel were placed in the autoclave, and the final state of the wall panel could only be known after the process (often taking more than 10 hours) was completed. The present invention can monitor the state of the wall panel in real time and predict the final state of the wall panel, thereby improving the forming efficiency.

[0106] 2. Compared with the previous creep aging forming, which can only use offline finite element simulation technology to formulate the process in advance, it is impossible to consider the influence of the actual die-sticking time and state, temperature, heating rate, and pressure errors of the equipment. The real-time data collected by sensors can be used to adjust the process in real time to achieve precise forming.

[0107] 3. The digital twin system is trained through big data and the database is expanded through real test data, which makes the prediction accuracy of the prediction model higher and higher. That is, it can organically learn from completed experiments and often cannot establish connections with previously completed experiments, thus providing guidance for subsequent experiments.

[0108] 4. Since the autoclave is a special high-temperature and high-pressure equipment, the operating tables in previous tests were all next to the autoclave, which posed a certain risk to the operators. The digital twin system can remotely operate the autoclave in the cloud, thereby reducing the risk to the operators.

[0109] The solution provided by the present invention can adjust the process based on real-time sensor data to obtain components with precise forming and target performance.

[0110] In order to implement the method of the embodiment of the present invention, the embodiment of the present invention also provides a computer program product, which includes computer instructions, and the computer instructions are stored in a computer-readable storage medium. The processor of the computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device performs the steps of the above method.

[0111] 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 may be a terminal, and its internal structure diagram may be as follows: Figure 3 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 through 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 screen or an electronic ink display screen, and the input device A05 of the computer device can be a touch layer covered on the display screen, or a key, trackball or touchpad set on the computer device housing, or an external keyboard, touchpad or mouse, etc.

[0112] Those skilled in the art will understand that Figure 3The 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 those shown in the figure, or combine certain components, or have a different arrangement of components.

[0113] The device provided by the 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.

[0114] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0115] 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 generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0116] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0117] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1A step that specifies a function in one or more boxes.

[0118] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0119] 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.

[0120] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. 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 disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape 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 temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0121] It can be understood that the memory of the embodiment of the present invention can be a volatile memory or a non-volatile memory, and can also include both volatile and non-volatile memories. Among them, the non-volatile memory can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a magnetic random access memory (FRAM), a flash memory, a magnetic surface memory, an optical disc, or a compact disc read-only memory (CD-ROM); the magnetic surface memory can be a disk memory or a tape memory. The volatile memory can be a 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, SyncLink Dynamic Random Access Memory), and direct RAMbus random access memory (DRRAM, Direct Rambus Random Access Memory).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.

[0122] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0123] The above are only 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 modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.

Claims

1. A creep aging forming method based on digital twin technology, characterized in that: The method comprises: Collect sensor data of aluminum alloy siding during creep aging forming; The sensor data is input into a first prediction model to obtain predicted springback and yield strength output by the first prediction model; the first prediction model is a prediction model obtained by using a first training method for an initial network model, and the initial network model is an LSTM algorithm model; the input of the first prediction model is time, temperature, heating rate and pressure, and the output of the first prediction model is predicted springback and yield strength of the aluminum alloy siding; Comparing the predicted springback amount and yield strength with the target springback amount and yield strength to determine a first difference value between the predicted springback amount and yield strength and the target springback amount and yield strength; Based on the first difference value, a first process parameter for updating the process of the aluminum alloy siding is determined by using a second prediction model, the first process parameter is sent to the autoclave, and the autoclave is controlled to perform the process operation with the first process parameter; the second prediction model is a prediction model obtained by using a second training method for the initial network model; the input of the second prediction model is the process update moment, time, temperature, heating rate and pressure, and the output of the second prediction model is the predicted springback and yield strength of the aluminum alloy siding; Wherein, obtaining the first prediction model includes: Conducting a creep aging forming test of aluminum alloy at the sample level to obtain a sample data set; and constructing a material constitutive equation using the sample data set; Based on the material constitutive equation, a finite element simulation is performed on the creep aging forming process of the component-level aluminum alloy siding to obtain the corresponding relationship between the strain and displacement of the characteristic points of the aluminum alloy siding at different times; Based on the corresponding relationship between the strain and displacement of the characteristic points of the aluminum alloy wall panel at different times, the initial process parameters of the aluminum alloy wall panel and the springback and yield strength at the mold attachment time during the finite element simulation process are determined; The initial process parameters and the springback and yield strength at the die attach moment are used as first training samples to train the initial network model to obtain a first prediction model; Wherein, obtaining the second prediction model includes: In the finite element simulation process, the moment when the room temperature loading is completed is taken as the initial moment, and the moment when the aluminum alloy wall panel is completely fitted with the mold is taken as the end moment. According to the preset process update frequency, multiple process update moments are determined between the initial moment and the end moment to obtain a process update moment sequence; Acquire finite element simulation process data corresponding to each process update time in the process update time sequence; The process update time sequence and the finite element simulation process data corresponding to each process update time in the process update time sequence are used as second training data to train the initial network model to obtain a second prediction model.

2. The creep aging forming method based on digital twin technology according to claim 1 is characterized in that: The material constitutive equation includes: ε1=f(σ,T,t,γ)=f(σ)f(T)f(γ)f(t) Calculation formula (1) Among them, ε1 is the creep strain; σ is the applied creep stress; T is the creep temperature; γ is the creep heating rate; t is the creep time, and f(σ), f(T), f(γ), and f(t) are the creep stress, temperature, heating rate, and time, respectively.

3. The creep aging forming method based on digital twin technology according to claim 1 is characterized in that: The obtaining of the sample data set comprises: Conduct creep aging forming tests on aluminum alloys at the specimen level to obtain an initial sample data set; Determine the optimal setting range of K value in K nearest neighbor algorithm; The K value in the K nearest neighbor algorithm is set based on the optimal setting range, and the initial sample data set is expanded based on the K nearest neighbor algorithm after the K value is set to obtain a new sample data set.

4. The creep aging forming method based on digital twin technology according to claim 1 is characterized in that: The corresponding relationship between the strain and displacement of the characteristic points of the aluminum alloy wall panel at different times includes: Calculation formula (2) in, is the strain of the characteristic points of the aluminum alloy siding at different times, is the distance between the initial position of the feature point and the mold, is the distance between the characteristic point of the aluminum alloy siding and the mold at different times.

5. The creep aging forming method based on digital twin technology according to claim 4 is characterized in that: The training of the initial network model to obtain the first prediction model includes: Determine that the algorithm optimizer of the LSTM algorithm model is SGD, Adam, Adamax or RMSpropadam, and the activation function of the LSTM algorithm model is Sigmiod, Tanh or ReLU; The algorithm optimizer and the activation function are used as model parameters to train the initial network model to obtain a first prediction model.

6. The creep aging forming method based on digital twin technology according to claim 5 is characterized in that: The determining of multiple process update moments comprises: Assume that the moment when the room temperature loading is completed is the initial moment T1, and the moment when the aluminum alloy siding is completely fitted with the mold after being placed in the autoclave for high temperature loading is the ending moment t1, the time between T1 and t1 is T, and the process update frequency is f; Then the process update time T2 is: T2=T×f+T1; The process update time T3 is: T3=T×f+T2; By analogy, the process update time Tn is: Tn=T×f+Tn-1.

7. The creep aging forming method based on digital twin technology according to claim 1 is characterized in that: After determining the first process parameter for updating the aluminum alloy wall panel, the method further includes: After each preset time interval, the sensor data of the aluminum alloy siding during the creep aging forming process is collected again; the sensor data is input into the first prediction model to obtain the predicted springback and yield strength output by the first prediction model; the predicted springback and yield strength are compared with the target springback and yield strength to determine the first difference between the predicted springback and yield strength and the target springback and yield strength; based on the first difference, the second prediction model is used to determine the first process parameter for process update of the aluminum alloy siding, the first process parameter is sent to the autoclave, and the autoclave is controlled to perform the process operation with the first process parameter, and the above operation is repeated until the predicted springback and yield strength are no different from the target springback and yield strength.

8. A creep aging forming device based on digital twin technology, characterized in that: The creep aging forming device based on digital twin technology includes: Sealed vacuum bags, autoclaves, high temperature strain gauges, temperature sensors, pressure sensors and processors; The sealed vacuum bag is connected to the high-temperature strain gauge, the temperature sensor and the pressure sensor, and is used to shrink during the vacuuming process to make the component fit the mold; the autoclave is connected to the high-temperature strain gauge, the temperature sensor and the pressure sensor, and is used to make the component enter the creep aging forming stage after being fit to the mold under heating and pressurizing conditions; the high-temperature strain gauge is used to collect status data of the component; the temperature sensor is used to collect temperature data, and the pressure sensor is used to collect pressure data; the processor is used to execute the steps of the method described in any one of claims 1 to 7.

Images