Method for predicting fracture position of rotary friction welding joint and related device
By establishing a three-dimensional thermal coupling model and a BP neural network prediction model in Abaqus software, processing the temperature and plastic strain data of the rotating friction welding joint, and determining the weak zone, the problem of difficult to predict the fracture position of the welding joint in the prior art is solved, and welding production efficiency is improved.
Patent Information
- Application Number
- CN202411810909.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-10
- Publication Date
- 2025-03-07
AI Technical Summary
The prior art is difficult to quickly and effectively predict the fracture position of the rotating friction welding joint, resulting in a decrease in welding quality and low production efficiency.
By establishing a three-dimensional thermal coupling model in Abaqus software, obtaining the peak temperature and peak equivalent plastic strain of the rotary friction welding head, combined with the BP neural network prediction model, these data are processed to obtain microhardness values, thereby determining the weak zone.
It realizes rapid screening of weak areas of rotating friction welding joints, improves welding production efficiency, and provides guidance for actual production.
Smart Images
Figure CN119578253A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the field of rotary friction welding, and in particular relates to a method for predicting a fracture position of a rotary friction welding joint and a related device. Background Art
[0002] As a solid phase welding technology, rotary friction welding uses the heat generated by the friction between the surfaces of the weldments under the action of external force to make the contact surface of the weldments reach a thermoplastic state, and then applies an upsetting force to quickly upset and complete the welding. As it has no welding defects such as thermal cracks, pores, lack of fusion, and lack of penetration, it is widely used in the welding of rotating bodies of different materials.
[0003] During the welding process, each area of the joint is always subjected to the strong coupling of heat and force. The heat generation and pressure distribution along the interface are uneven in the radial direction, and the metal at the interface will produce uneven plastic deformation, which will lead to differences in mechanical properties in different areas of the joint and even a decrease in welding quality. At present, most of the research on weak areas of welded joints is based on the microstructure characterization and mechanical property test comparison of each area of the joint after the welding experiment. For friction welding of the same material with different welding parameters, similar experiments need to be repeated for characterization tests, which greatly reduces the research efficiency and actual production efficiency. In addition, the finite element simulation of friction welding can effectively simulate the temperature field and plastic flow field during the welding process, providing some reference for welding experimental research and production, but it cannot be further combined with reality to provide more effective help to significantly improve the experimental research and actual production efficiency. Summary of the invention
[0004] In view of the above-mentioned problems, the purpose of the present invention is to provide a method and related device for predicting the fracture position of a rotary friction welding joint, so as to quickly screen the weak areas of the joint, solve the problems in a targeted manner, provide guidance for actual welding production and greatly improve welding production efficiency.
[0005] To achieve the above purpose, the technical solution adopted by the present invention is as follows: A method for predicting the fracture position of a rotary friction welding joint includes the following steps: The three-dimensional thermal-mechanical coupling model established by Abaqus software is used to obtain the peak temperature and peak equivalent plastic strain of all grid points of the rotary friction welding joint at the fracture position of the rotary friction welding joint to be predicted under the preset rotary friction welding parameters; The peak temperature and peak equivalent plastic strain of all grid points of the rotary friction welding joint are processed by the BP neural network prediction model which has met the prediction accuracy, and the microhardness values corresponding to all grid points are obtained. According to the microhardness values corresponding to all grid points, the weak area on the rotary friction welding joint is determined, and the weak area is used as the predicted fracture position of the rotary friction welding joint.
[0006] Preferably, the material of the rotary friction welding head to be predicted is the same as the material of the rotary friction welding head used to establish the three-dimensional thermal-mechanical coupling model; the material size specification of the rotary friction welding head to be predicted is the same as the material size specification of the rotary friction welding head used to establish the three-dimensional thermal-mechanical coupling model.
[0007] Preferably, the process of establishing the three-dimensional thermal-mechanical coupling model established by the Abaqus software includes: Select metal materials of preset material and specification, weld the metal materials by using rotary friction welding process, and measure the temperature at different positions during the welding process; Combined with the constitutive equation of the selected metal material, based on Abaqus software, the coupled CEL method and the strain-compensated Arrhenius material model are used to establish a three-dimensional thermal-mechanical coupling model. During the process of establishing the three-dimensional thermal-mechanical coupling model, meshing is performed on each area along the axial cross section with the welding joint surface as the center. The three-dimensional thermal-mechanical coupling model is used to simulate the temperature field and plastic flow field of each area of the joint during the welding process. The established three-dimensional thermal-mechanical coupling model is corrected in combination with the measured temperatures at different positions during the welding process, thereby obtaining the three-dimensional thermal-mechanical coupling model established by the Abaqus software.
[0008] Preferably, the predicted rotary friction welding joint is a round bar or a round tube.
[0009] Preferably, the parameters of the rotary friction welding include rotation speed, friction pressure and upset pressure.
[0010] Preferably, when the peak temperature and the peak equivalent plastic strain of all grid points of the rotary friction welding joint are processed by the BP neural network prediction model that has satisfied the prediction accuracy: The peak temperature and peak equivalent plastic strain of all grid points of the rotary friction welding joint are first normalized, and then the normalized results are input into the BP neural network prediction model that has met the prediction accuracy for processing. After that, the BP neural network prediction model that has met the prediction accuracy outputs the microhardness values corresponding to all grid points.
[0011] Preferably, the BP neural network prediction model adopts a multi-layer feedforward neural network, and the BP neural network prediction model includes an input layer, a hidden layer and an output layer. Neurons in each hidden layer and output layer are given specific thresholds and activation functions, and weight connections are used between layers; The output of the hidden layer is calculated using the following formula: ))
[0012] in: is the regularization correction, is the residual connection correction, is the weight matrix, X is the input vector; is the bias term; is the tanh activation function, defined as ; When training the BP neural network prediction model, the adjusted determination coefficient Adjusted R 2 Determine the prediction accuracy of the BP neural network prediction model and adjust the determination coefficient Adjusted R 2 The calculation formula is as follows:
[0013]
[0014] in: is the true value of the i-th sample, is the average of the true values, is the predicted value of the i-th sample, is the number of samples, is the number of independent variables.
[0015] The present invention also provides a system for predicting the fracture position of a rotary friction welding joint, which is similar to the method for predicting the fracture position of a rotary friction welding joint of the present invention, and comprises: Input data acquisition unit: used to obtain the peak temperature and peak equivalent plastic strain of all grid points in the region where the rotational friction welding joint fracture position to be predicted is located under preset rotational friction welding parameters using a three-dimensional thermal-mechanical coupling model established by Abaqus software; Calculation unit: used to process the peak temperature and peak equivalent plastic strain of all grid points of the rotary friction welding joint through the BP neural network prediction model that has met the prediction accuracy, and obtain the microhardness values corresponding to all grid points; Prediction unit: used to determine the weak area on the rotary friction welding joint according to the microhardness values corresponding to all grid points, and use the weak area as the predicted fracture position of the rotary friction welding joint.
[0016] The present invention also provides an electronic device, comprising: one or more processors; a storage device having one or more programs stored thereon; When the one or more programs are executed by the one or more processors, the one or more processors are enabled to implement the method for predicting the fracture position of a rotary friction welding joint as described above.
[0017] The present invention further provides a storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the method for predicting the fracture position of a rotary friction welding joint as described above.
[0018] The present invention has the following beneficial effects: In the method for predicting the fracture position of a rotary friction welding joint of the present invention, for the rotary friction welding joint whose fracture position is to be predicted, it is only necessary to set the parameters of the rotary friction welding joint whose fracture position is to be predicted in the three-dimensional thermal-mechanical coupling model established by the Abaqus software when performing rotary friction welding, so as to obtain the peak temperature and peak equivalent plastic strain of all grid points in the area where the rotary friction welding joint is located, and then use the BP neural network prediction model that has met the prediction accuracy to process the peak temperature and peak equivalent plastic strain of all grid points of the rotary friction welding joint, and obtain the measured microhardness values corresponding to all grid points. Since the microhardness value can be converted into the ultimate tensile strength of the grid point, the weak area on the rotary friction welding joint (i.e., the area with the minimum ultimate tensile strength) can be determined according to the microhardness values corresponding to all grid points. It can be seen that for the rotating bodies of the same material and specification, the present invention can quickly screen the weak areas of the rotary friction welding joint, solve the problem in a targeted manner, provide guidance for the actual welding production, and greatly improve the welding production efficiency. BRIEF DESCRIPTION OF THE DRAWINGS
[0019] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.
[0020] Figure 1 is the BP neural network prediction model used in the embodiment of the present invention; Figure 2 A comparison chart of tensile strength prediction results of the test data set used in the embodiment of the present invention; Figure 3 This is the prediction result of the joint hardness at 1500 rpm used in the embodiment of the present invention. DETAILED DESCRIPTION
[0021] In order to explain the present invention more intuitively and demonstrate the advantages of the present invention, the following embodiments are used to specifically illustrate the implementation mode and related implementation details of the present invention.
[0022] The method for predicting the fracture position of a rotary friction welding joint of the present invention comprises the following steps: Step 1: Select a rotating body of a metal material, perform rotational friction welding using a rotational friction welding process, and measure the temperature of the rotational friction welding joint area during the welding process.
[0023] Step 2. Combined with the constitutive equation of the metal material selected in step 1, based on Abaqus software, a three-dimensional thermomechanical coupling model is established by using the coupled CEL (coupled Euler-Lagrange) method and the strain-compensated Arrhenius material model. The three-dimensional thermomechanical coupling model is used to simulate the temperature field and plastic flow field of each region of the rotary friction welding joint during the rotary welding process. The three-dimensional thermomechanical coupling model is corrected in combination with the welding temperature measured in the experiment in step 1 to ensure the accuracy of the numerical simulation of the three-dimensional thermomechanical coupling model.
[0024] Step 3: From the three-dimensional thermal-mechanical coupling model established in step 2, the peak temperature (T) and peak equivalent plastic strain (ε) of each grid point in different regions of the rotary friction welding joint are extracted along the axial direction of the rotating body. p ).
[0025] Step 4: Based on the position of each grid point in step 3, the microhardness value of the corresponding point of the rotational friction welding joint is measured in the experiment.
[0026] Step 5: Compare the peak temperature (T) and peak equivalent plastic strain (ε p ) as the input parameter and the measured microhardness value of the corresponding point of the rotary friction welding joint as the output parameter to establish a BP neural network prediction model.
[0027] Step six: train and learn the established BP neural network prediction model, and repeat the training and verification process until a high prediction accuracy is achieved.
[0028] Step 7: For a friction welded joint that has been subjected to friction welding (the rotating metal body of friction welding is made of the same material and has the same size as the rotating metal body of friction welding in step 1), the process parameters of the friction welded joint are input into the three-dimensional thermal-mechanical coupling model established in step 2. The peak temperature (T) and the peak equivalent plastic strain (ε) of each grid point in the friction welded joint region are obtained through the three-dimensional thermal-mechanical coupling model. p ), the peak temperature (T) and the peak equivalent plastic strain (ε p ) is input into the trained BP neural network prediction model, and after being processed by the BP neural network prediction model, the microhardness values of the grid points corresponding to the rotary friction welding joint are obtained, and the weak areas on the rotary friction welding joint are determined according to the microhardness values of the grid points.
[0029] In the above solution of the present invention, the rotating body includes a round bar or a round tube. The rotation friction welding parameters include rotation speed, friction pressure and upsetting pressure.
[0030] In step 1, the experimental temperature measurement method is to drill holes at specific locations of the welding material and embed thermocouples to complete the temperature collection during the welding process.
[0031] In step 2, the specific calculation formula of the constitutive equation of the selected metal material is as follows:
[0032]
[0033] in: is the flow stress; is a specific coefficient; is the thermal deformation activation energy of the material; is the material constant; is the gas constant; is the deformation temperature of the material; Z is the Zener-Hollomon parameter.
[0034] In step 2, when the three-dimensional thermal-mechanical coupling model is corrected in combination with the welding temperature measured in the experiment of step 1, after temperature comparison and parameter adjustment, when the correlation coefficient R>0.99, the changes in temperature field and plastic flow field during friction welding can be accurately simulated.
[0035] In the step 2, during the process of establishing the three-dimensional thermal-mechanical coupling model, when meshing is performed, the welding joint surface is taken as the center and meshes are divided into each area along the axial cross section.
[0036] The peak temperature (T) and the peak equivalent plastic strain (ε p ) is the highest temperature and the highest equivalent plastic strain reached at this point from the beginning to the end of welding. It can represent the plastic deformation state of this point under the action of thermal-mechanical coupling during the welding process, and directly determines the microstructure and mechanical properties of this point.
[0037] The microhardness value measurement is firstly to divide the weld joint into a grid according to step three, and then to use a microhardness tester to measure the Vickers hardness, Rockwell hardness and Brinell hardness.
[0038] like Figure 1The BP neural network prediction model adopts a multi-layer feedforward neural network, which includes an input layer, a hidden layer and an output layer. The neurons in each hidden layer and the output layer are given specific thresholds and activation functions, and weights are used to connect the layers. The BP neural network prediction model uses Z-score to normalize the input value. The output of the hidden layer of the BP neural network prediction model is calculated using the following formula: ))
[0039] in: is regularization correction; Correction for residual connection; is the weight matrix, which determines how the input affects the output of the hidden layer; X is the input vector, which contains the input data passed to the hidden layer; It is a bias term that helps the model fit the data better; is the tanh activation function. It is defined as .
[0040] The BP neural network prediction model uses mean square error (MSE) as the error function of the model, and its calculation method is as follows:
[0041] in: is the number of training samples, is the number of outputs, is the predicted value of the jth output of the ith sample, is the true value of the jth output of the ith sample.
[0042] The BP neural network prediction model adopts a gradient descent optimization algorithm to update the network weights.
[0043] The training data set, validation data set, and test data set account for 70%, 15%, and 15% respectively. 2 ) determines the prediction accuracy of the prediction model, and its calculation formula is as follows: Adjusted R 2 When it is >0.95, the prediction model is considered accurate and can achieve the prediction target.
[0044]
[0045]
[0046] in: is the true value of the i-th sample, is the average of the true values, is the predicted value of the i-th sample, is the number of samples, is the number of independent variables.
[0047] Example In this embodiment, 7E33 high-strength aluminum alloy is taken as an example, and rotary friction welding and finite element simulation are combined to provide a sample database for the BP neural network prediction model to achieve the prediction of the fracture position of the welded joint. The specific steps are as follows: Step (1), select 7E33 high-strength aluminum alloy tube as welding material, its size is φ12mm×φ22mm×45mm (inner diameter×outer diameter×length), grind its end face and clean it with anhydrous ethanol to remove the surface oxide layer. Use a continuous drive friction welding machine with equipment model ZCC-10TxDJ. The welding parameters are rotation speed 1300rpm, friction pressure 15kN, and upsetting pressure 20kN. Punch holes along the axial direction of the welding rod to embed the thermocouple, and the punching positions are 2mm, 4mm, 6mm, 8mm, and 12mm away from the welding joint surface, and measure the welding temperature at each point.
[0048] Step (2), based on the constitutive equation of 7E33 aluminum alloy, a three-dimensional thermomechanical coupling model was established using Abaqus software, and the temperature field and plastic flow field of the joint during welding were simulated using the three-dimensional thermomechanical coupling model. The simulated temperature was compared with the measured temperature, and the three-dimensional thermomechanical coupling model was adjusted, so that the simulation error was finally reduced to only 1.2%.
[0049] Step (3), with the joint surface after welding as the center, divide the grid along the axial direction with a row and column interval of 0.5 mm, and divide it into 200 points in total. Extract the peak temperature (T) and peak equivalent plastic strain (ε) of each point in the welding process from the three-dimensional thermal-mechanical coupling model. p ).
[0050] In step (4), a Vickers hardness micrometer is used to characterize the microhardness of 200 points of the experimental joint corresponding to the three-dimensional thermal-mechanical coupling. The measured microhardness values of some points are shown in Table 1: Table 1
[0051] Step (5) is to use the peak temperature (T) and the peak equivalent plastic strain (ε p ) as the input value and the ultimate tensile strength of the corresponding point as the output value, 140 groups of corresponding values are selected as the training data set, 30 groups of corresponding values are selected as the verification data set, and 30 groups of corresponding values are selected as the test data set.
[0052] Step (6) establishes a BP neural network prediction model, trains it with the sample training set constructed in step (5), verifies it with the validation data set, tests it with the sample test set, repeats the training and verification, and finally, achieves the required prediction accuracy requirement and obtains a trained BP neural network prediction model. The comparison of the prediction results of the test data set in this embodiment is shown in the figure below: Figure 2 As shown in the figure, the error between the measured value and the predicted value at each point is within 5%, which meets the prediction accuracy requirement. Step (7), relative to step (1), changes the rotation speed to 1500rpm, and keeps the other parameters unchanged. The peak temperature (T) and peak equivalent plastic strain (ε) of each point in the welding process in the three-dimensional thermal-mechanical coupling model are extracted. p ). The BP neural network prediction model is used to calculate and output the microhardness value corresponding to each grid point, and the microhardness value of each grid point is made into Figure 3 The microhardness value cloud diagram of the joint is shown, and the location of the weak area of the joint is determined on this cloud diagram (see Figure 3 ).
[0053] It can be seen from the above scheme that the present invention can quickly screen the weak areas of the rotary friction welding joint.
[0054] The present invention also provides a system for predicting the fracture position of a rotary friction welding joint, which is similar to the method for predicting the fracture position of a rotary friction welding joint of the present invention, and comprises: Input data acquisition unit: used to obtain the peak temperature and peak equivalent plastic strain of all grid points in the region where the rotational friction welding joint fracture position to be predicted is located under preset rotational friction welding parameters using a three-dimensional thermal-mechanical coupling model established by Abaqus software; Calculation unit: used to process the peak temperature and peak equivalent plastic strain of all grid points of the rotary friction welding joint through the BP neural network prediction model that has met the prediction accuracy, and obtain the microhardness values corresponding to all grid points; Prediction unit: used to determine the weak area on the rotary friction welding joint according to the microhardness values corresponding to all grid points, and use the weak area as the predicted fracture position of the rotary friction welding joint.
[0055] The embodiments of the present invention also provide corresponding electronic devices and computer-readable storage media for implementing the solutions provided by the embodiments of the present invention.
[0056] The device includes a memory and a processor, the memory is used to store instructions or codes, and the processor is used to execute the instructions or codes so that the device executes the method for predicting the fracture position of a rotary friction welding joint described in any embodiment of the present application.
[0057] The storage medium stores a computer program, wherein the computer program, when executed by the processor, implements the method for predicting the fracture position of a rotary friction welding joint as described in any embodiment of the present application.
[0058] Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.
[0059] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.
Claims
1. A method for predicting the fracture position of a rotary friction welding joint, characterized in that: The process includes the following: The three-dimensional thermal-mechanical coupling model established by Abaqus software is used to obtain the peak temperature and peak equivalent plastic strain of all grid points of the rotary friction welding joint at the fracture position of the rotary friction welding joint to be predicted under the preset rotary friction welding parameters; The peak temperature and peak equivalent plastic strain of all grid points of the rotary friction welding joint are processed by the BP neural network prediction model which has met the prediction accuracy, and the microhardness values corresponding to all grid points are obtained. According to the microhardness values corresponding to all grid points, the weak area on the rotary friction welding joint is determined, and the weak area is used as the predicted fracture position of the rotary friction welding joint.
2. The method for predicting the fracture position of a rotary friction welding joint according to claim 1, characterized in that: The material of the rotational friction welding head to be predicted is the same as the material of the rotational friction welding head used to establish the three-dimensional thermal-mechanical coupling model; the material size specification of the rotational friction welding head to be predicted is the same as the material size specification of the rotational friction welding head used to establish the three-dimensional thermal-mechanical coupling model.
3. The method for predicting the fracture position of a rotary friction welding joint according to claim 2, characterized in that: The process of establishing the three-dimensional thermal-mechanical coupling model established by the Abaqus software includes: Select metal materials of preset material and specification, weld the metal materials by using rotary friction welding process, and measure the temperature at different positions during the welding process; Combined with the constitutive equation of the selected metal material, based on Abaqus software, the coupled CEL method and the strain-compensated Arrhenius material model are used to establish a three-dimensional thermal-mechanical coupling model. During the process of establishing the three-dimensional thermal-mechanical coupling model, meshing is performed on each area along the axial cross section with the welding joint surface as the center. The three-dimensional thermal-mechanical coupling model is used to simulate the temperature field and plastic flow field of each area of the joint during the welding process. The established three-dimensional thermal-mechanical coupling model is corrected in combination with the measured temperatures at different positions during the welding process, thereby obtaining the three-dimensional thermal-mechanical coupling model established by the Abaqus software.
4. The method for predicting the fracture position of a rotary friction welding joint according to claim 1, characterized in that: The predicted rotary friction welding joint is a round bar or a round tube.
5. The method for predicting the fracture position of a rotary friction welding joint according to claim 1, characterized in that: The parameters of the rotary friction welding include rotation speed, friction pressure and upset pressure.
6. The method for predicting the fracture position of a rotary friction welding joint according to claim 1, characterized in that: When the peak temperature and peak equivalent plastic strain of all grid points of the rotary friction welding joint are processed by the BP neural network prediction model that has met the prediction accuracy: The peak temperature and peak equivalent plastic strain of all grid points of the rotary friction welding joint are first normalized, and then the normalized results are input into the BP neural network prediction model that has met the prediction accuracy for processing. After that, the BP neural network prediction model that has met the prediction accuracy outputs the microhardness values corresponding to all grid points.
7. The method for predicting the fracture position of a rotary friction welding joint according to claim 1, characterized in that: The BP neural network prediction model adopts a multi-layer feedforward neural network. The BP neural network prediction model includes an input layer, a hidden layer and an output layer. Neurons in each hidden layer and output layer are given specific thresholds and activation functions, and weight connections are used between layers. The output of the hidden layer is calculated using the following formula: )) in: is the regularization correction, is the residual connection correction, is the weight matrix, X is the input vector; is the bias term; is the tanh activation function, defined as ; When training the BP neural network prediction model, the adjusted determination coefficient Adjusted R 2 Determine the prediction accuracy of the BP neural network prediction model and adjust the determination coefficient Adjusted R 2 The calculation formula is as follows: in: is the true value of the i-th sample, is the average of the true values, is the predicted value of the i-th sample, is the number of samples, is the number of independent variables.
8. A system for predicting the fracture position of a rotary friction welding joint, characterized in that: include: Input data acquisition unit: used to obtain the peak temperature and peak equivalent plastic strain of all grid points in the region where the rotational friction welding joint fracture position to be predicted is located under preset rotational friction welding parameters using a three-dimensional thermal-mechanical coupling model established by Abaqus software; Calculation unit: used to process the peak temperature and peak equivalent plastic strain of all grid points of the rotary friction welding joint through the BP neural network prediction model that has met the prediction accuracy, and obtain the microhardness values corresponding to all grid points; Prediction unit: used to determine the weak area on the rotary friction welding joint according to the microhardness values corresponding to all grid points, and use the weak area as the predicted fracture position of the rotary friction welding joint.
9. An electronic device, characterized in that: include: one or more processors; a storage device having one or more programs stored thereon; When the one or more programs are executed by the one or more processors, the one or more processors are enabled to implement the method for predicting a fracture position of a rotational friction welding joint as claimed in any one of claims 1 to 7.
10. A storage medium, characterized in that: A computer program is stored thereon, wherein when the computer program is executed by a processor, the method for predicting the fracture position of a rotary friction welding joint as claimed in any one of claims 1 to 7 is implemented.