Dynamic simulation method for ultrasonic welding process of multiple layers of thin plates

Abstract

The invention discloses a dynamic simulation method for the ultrasonic welding process of multiple layers of thin plates. According to the method, simulation model combinations of different dimensions are provided, so that the problem of simulation of the mechanical property non-uniformity of the welded joint is effectively solved. The entity unit detailed model can finely present the microstructure characteristics of the welding joint, including the interface reaction area and the welding seam fusion state of each layer of plate, and the shell unit simplified model greatly improves the calculation efficiency on the premise of ensuring the engineering precision; the part impact test simulation model further combines the material failure behavior with the actual working condition. The three models support one another through a unified material failure setting system, the performance difference between a welding seam area and base metal is captured, the dynamic response rule under the impact load is reflected, and the simulation result can truly reproduce the mechanical behavior of a welding joint in the complex stress state.

Description

Technical Field

[0001] This invention belongs to the field of automotive body welding technology, specifically a dynamic simulation method for ultrasonic welding process of multi-layer thin plates. Background Technology

[0002] Ultrasonic welding technology, as a highly efficient and environmentally friendly joining process, has found its application in the automotive manufacturing industry driven by the industry's continuous pursuit of safety, lightweighting, and production efficiency. Traditional welding methods (such as arc welding and resistance welding) suffer from problems such as large heat-affected zones, high material consumption, and severe pollution. Ultrasonic welding, however, achieves solid-state bonding of materials through high-frequency mechanical vibration, eliminating the need for additional solder or adhesives. This meets the stringent requirements of automotive parts for high strength, high precision, and low damage. Ultrasonic welding typically completes a single process within seconds, making it suitable for large-scale assembly line production in the automotive industry and shortening the production cycle of traditional processes. The welding process produces no sparks or harmful gas emissions, meeting the environmental standards of automotive factories and reducing the cost of air quality control in the workshop. It can be integrated with robots and vision inspection systems to achieve real-time monitoring and adjustment of welding parameters. With the development of new energy vehicles and intelligent vehicles, the demand for battery connections and lightweight material welding has further promoted the widespread adoption of this technology.

[0003] Existing methods for simulating ultrasonic welded joints mainly include the common-node method, rigid element connection method, beam element connection method, and solid element modeling method. Among them, the common-node method and rigid element connection method have the characteristic of not failing, while the beam element connection method and solid element modeling method, although able to simulate the failure of the connection point, cannot accurately simulate the non-uniformity of the mechanical properties of the ultrasonic welded joint. With the popularization of ultrasonic welding technology, developing a highly accurate joint simulation method is particularly important. Summary of the Invention

[0004] The purpose of this invention is to provide a dynamic simulation method for the ultrasonic welding process of multilayer thin plates in order to solve the problems mentioned above.

[0005] The technical solution adopted in this invention is as follows: A dynamic simulation method for ultrasonic welding process of multilayer thin plates, comprising the following steps:

[0006] S1: Conduct mechanical property tests on multi-layer thin plate lap joint samples. Obtain the basic mechanical parameters of the weld joint by measuring the load-displacement curve. The test data will serve as the core input for the detailed model S2 and the simplified model S3.

[0007] S2: Based on LS-DYNA software, a detailed model of the sample is established, 0.05mm solid elements are used to divide the upper plate, the lower plate, the aluminum interface interphase, the weld zone and the fusion line, the molten pool topography obtained from the S1 test is used to determine the penetration parameters, by adjusting the material failure curve coefficient and the weld zone size, the simulation and test curve COR fitting degree is more than 90%, the model calibration results will be directly transmitted to S3 for simplified model parameter initialization.

[0008] S3: A simplified shell element model (grid size 0.4mm) of the sample is constructed, the material failure curve coefficient and the weld zone parameters output by S2 detailed model are integrated, through secondary optimization, the COR fitting degree is more than 85%, the model results are used as the basis template for S4 component level simulation.

[0009] S4: Component level impact test is carried out, a simulation model is established based on the simplified model of S3, the weld zone and fusion line parameters are adjusted to COR fitting degree ≥85%, and the joint failure form in the test is fed back to S3 for optimizing the boundary condition setting of the model.

[0010] S5: Through the multi-physical field coupling adaptive optimization module, the mechanical parameters of S1, the temperature field and displacement field data of S2-S4 are integrated, the "sample-component-vehicle" three-level parameter mapping relationship is established, the material failure threshold is dynamically corrected through machine learning algorithm, the optimization parameters output by this module will guide the calibration process of S6 and the final model establishment of S8.

[0011] S6: Start the model calibration closed loop process, carry out sample level joint mechanical property test and component level impact test, after the impact test, establish a joint detailed model, compare the joint detailed model with the sample level joint mechanical property test, and determine whether they are consistent. If not, modify the establishment of the joint detailed model; if consistent, output the material parameter information.

[0012] S7: Use the output material parameter information to establish a joint simplified model, compare the joint simplified model with the sample level joint mechanical property test, and determine whether they are consistent. If not, modify the establishment of the joint simplified model; if consistent, preliminarily determine the joint simulation model.

[0013] S8: According to the preliminarily determined joint simulation model, a component impact test simulation model is established. Compare the component impact model with the component level impact test, and determine whether they are consistent. If not, modify the component impact model; if consistent, finally establish the joint simulation model.

[0014] In a preferred embodiment, in step S1, when carrying out the sample-level joint mechanical property test, a multi-layer sheet lap joint structure is used for loading test, and the loading rate is controlled to be consistent with the stress change rate in the actual welding process. The deformation data of the welded joint under different loads are collected by a high-precision sensor to obtain a complete load-displacement curve. The welding process parameters of the sample are matched with the ultrasonic frequency, amplitude and welding pressure in the actual production, so as to ensure that the test results can reflect the mechanical properties of the actual welded joint and provide reliable basic data support for subsequent model establishment.

[0015] In a preferred embodiment, in step S2, a detailed model of the joint sample test is established based on LS-DYNA software, and a solid element is used for modeling to capture the microstructure characteristics of the welded joint. The model includes key components such as the upper plate, the lower plate, the aluminum interfacial intermediate phase, the weld zone and the fusion line. The fusion depth size of each layer of plate is determined by analyzing the molten pool topography of the sample joint, and the boundary range of the weld zone and the fusion line is divided in combination with the hardness distribution curve. In the model debugging stage, the material failure curve coefficient and the width parameters of the weld zone and the fusion line are adjusted, and the correlation coefficient fitting degree of the simulation output temperature-time curve and height difference-distance curve and the test data is improved through multiple iterations, so as to ensure that the model can accurately reproduce the dynamic mechanical behavior in the welding process.

[0016] In a preferred embodiment, in step S3, a simplified model of the sample test is constructed based on the detailed model, a shell element is used for modeling to reduce the calculation complexity, and the model structure includes the upper plate base metal, the lower plate base metal, the weld zone and the fusion line. The material parameters of the simplified model are directly inherited from the calibration results of the detailed model, and the grid size and element type are adjusted to balance the calculation efficiency and simulation accuracy. The simulation results of the temperature field and the displacement field are monitored in real time during the model debugging process, the geometric parameters of the weld zone and the fusion line are fine-tuned, the correlation coefficient fitting degree of the simulation curve of the simplified model and the test data meets the engineering application requirements, and a high-efficiency and feasible model template is provided for subsequent component-level simulation.

[0017] In a preferred embodiment, in step S4, the test device needs to simulate the impact load conditions in the actual use scenario, and the deformation mode and failure time of the joint during the impact process are recorded by a high-speed camera and a dynamic data acquisition system. A component impact simulation model is established based on the calibrated simplified model, and the material failure parameters and the geometric size of the weld zone are adjusted during the simulation process, so that the correlation coefficient fitting degree of the simulation results and the test data meets the specified standard. The joint failure form observed in the test is compared and analyzed with the simulation results, the simulation accuracy of the model for the failure behavior of the welded joint is verified, and key basis is provided for the final model establishment.

[0018] In a preferred embodiment, in the step S5, the parameter dynamic optimization is realized by constructing a multi-scale data fusion framework. This module first extracts the features of the load-displacement curve obtained in S1, identifies key mechanical characteristic quantities such as the elastic segment, the yield point and the fracture threshold, and integrates the temperature field distribution cloud map and the displacement field time series data output by the detailed model and the simplified model in S2-S4. Through the space-time alignment algorithm, different physical field data are mapped to a unified coordinate system.

[0019] An improved radial basis function neural network is used to establish a "coupon-component-vehicle" three-level parameter transfer mechanism, in which the coupon-level data is used as the basic features of the input layer of the network, the component-level impact response is used as the constraint condition of the intermediate layer, and the final output is the equivalent parameter set adapted to the vehicle crash simulation. The reinforcement learning agent built in the module takes the COR fitting degree deviation as the reward signal, and adjusts the correction coefficient of the material failure threshold through the Q-learning algorithm. When the simulation deviation between the coupon level and the component level exceeds the threshold, the parameter backtracking mechanism is automatically triggered to retrieve the historical optimal parameters from the S6 calibration process for secondary training until the parameter package that meets the accuracy requirements of the final model in S8 is generated.

[0020] In the step S5, the dynamic correction formula of the material failure threshold is:

[0021] ;

[0022] In the formula:

[0023] represents the material failure threshold after the kth iteration

[0024] represents the initial value of the failure threshold of the previous iteration

[0025] α represents the adaptive learning rate

[0026] ΔT represents the temperature field peak value deviation between the current simulation and the test

[0027] ΔT0 represents the reference temperature deviation

[0028] COR target represents the target fitting degree

[0029] COR current represents the correlation coefficient calculation value of the current iteration

[0030] The formula realizes nonlinear adjustment of the correction amount by a hyperbolic tangent function, and when the temperature deviation and the fitting degree deviation increase in the same direction, the correction amplitude of the failure threshold is gradient-enhanced, so that parameter adjustment can quickly converge and avoid overshoot, and the local optimal problem caused by traditional linear correction is effectively solved.

[0031] In a preferred embodiment, in the step S6, when starting the model calibration closed loop process, firstly, the simulation results of the detailed model are systematically compared with the sample-level mechanical property test data, and the consistency of the temperature field distribution, stress concentration area and load-displacement curve is mainly checked. If significant deviation is found between the simulation results and the test data, the grid partition accuracy or material constitutive parameters of the model need to be adjusted back; if the consistency meets the requirements, the verified material parameter information is output, and these parameters will be used as the basic input for the construction of the simplified model, ensuring the reliability of the model in the subsequent steps.

[0032] In a preferred embodiment, in the step S7, the joint simplified model is established by using the calibrated material parameters, and is again compared and verified with the sample-level test data in multiple dimensions, including dynamic response characteristics, failure mode and stress distribution at key positions. By adjusting the boundary conditions and contact settings of the model, the balance between the calculation efficiency and simulation accuracy of the model is optimized, ensuring that the mechanical behavior of the welded joint can be accurately reflected while reducing the calculation cost. After verification, the structure and parameter configuration of the joint simulation model are preliminarily determined, providing a standardized model template for component-level simulation.

[0033] In a preferred embodiment, in the step S8, based on the preliminarily determined joint simulation model, a complete simulation system for component-level impact test is constructed, and the simulation results are compared with the component-level impact test data to verify the failure mode, load-carrying capacity and energy absorption characteristics of the joint under impact load. According to the comparison results, the model parameters are finally optimized, including key variables such as material failure threshold and weld zone stiffness, to ensure that the model can accurately predict the joint performance under different working conditions. The finally established simulation model needs to be verified by multiple sets of test data to ensure its stability and reliability in engineering applications.

[0034] In summary, due to the adoption of the above technical solutions, the present application has the following advantages:

[0035] 1、In the present application, by providing different dimensions of simulation model combination, the simulation problem of unevenness of the mechanical properties of the welded joint is effectively solved. The entity unit detailed model can finely present the microstructure characteristics of the welded joint, including the interface reaction zone of each layer of plate and the weld fusion state, while the shell unit simplified model greatly improves the calculation efficiency under the premise of ensuring engineering accuracy, and the component impact test simulation model further combines the material failure behavior with the actual working condition. The three models support each other through a unified material failure setting system, which not only captures the performance difference between the weld zone and the base material, but also reflects the dynamic response law under impact load, so that the simulation results can truly reproduce the mechanical behavior of the welded joint under complex stress state.

[0036] 2、In the present application, key parameters are obtained from the closed-loop verification process of sample test, component test and simulation calibration, avoiding the deviation caused by theoretical assumption, and ensuring that the method has a solid engineering application foundation. By establishing the mapping relationship between core parameters and process conditions, the technical personnel only need to adjust a small number of key variables to quickly adapt to the joint simulation requirements under different welding process parameters, without the need to repeatedly build complete models. This parameterized design not only reduces the simulation threshold, but also realizes the rapid iteration and optimization of process parameters in new product development, significantly shortening the technical verification period.

[0037] 3、In the present application, the failure evolution path of the ultrasonic welded joint in the steel-aluminum hybrid vehicle body during the whole vehicle collision process can be accurately simulated, including the whole process behavior from initial deformation to complete fracture. By correlating the simulation results with the vehicle body safety performance indicators, the welding process parameters and joint structure design can be optimized, avoiding potential safety hazards in traditional experience design. This simulation-based forward development mode provides scientific guidance for the connection reliability of new energy vehicle lightweight body, and helps to reduce the weight of the vehicle body while ensuring the crash safety performance. BRIEF DESCRIPTION OF DRAWINGS

[0038] Figure 1 is the flow principle schematic diagram of the present application;

[0039] Figure 2 is the physical sample schematic diagram in the present application;

[0040] Figure 3 is the three-dimensional vibration model schematic diagram of the lap joint in the present application;

[0041] Figure 4 is the thermal-mechanical coupling shell unit model schematic diagram of the lap joint in the present application;

[0042] Figure 5 is the comparison schematic diagram of the simulation temperature and the measured temperature of the steel / aluminum / steel ultrasonic welding in the present application;

[0043] Figure 6 A comparison diagram of the height difference of the upper workpiece surface simulated and measured in the application. DETAILED DESCRIPTION

[0044] In order to make the purpose, technical scheme and advantages of the application more clear, the application is further described in detail below in combination with the drawings and examples. It should be understood that the specific examples described herein are only used to explain the application and do not limit the application.

[0045] Reference Figures 1-6 A multi-layer sheet ultrasonic welding process dynamics simulation method, the method comprising the following steps:

[0046] S1: Carry out multi-layer sheet lap joint sample level mechanical property test, obtain the basic mechanical parameters of the welded joint by measuring the load-displacement curve, and the test data will be the core input basis for S2 detailed model and S3 simplified model.

[0047] S2: Establish a sample detailed model based on LS-DYNA software, divide the upper plate, lower plate, aluminum interfacial intermediate phase, weld zone and fusion line by 0.05mm solid element, determine the penetration depth parameters by using the molten pool topography obtained from S1 test, adjust the material failure curve coefficient and weld zone size, so that the simulation and test curve COR fitting degree is more than 90%, and the model calibration result will be directly transmitted to S3 for simplified model parameter initialization.

[0048] S3: Construct a sample simplified shell element model (grid size 0.4mm), integrate the material failure curve coefficient and weld zone parameters output by S2 detailed model, through secondary optimization, the COR fitting degree is more than 85%, and the model result is used as the basis template for S4 component level simulation.

[0049] S4: Carry out component level impact test, establish a simulation model based on the simplified model of S3, adjust the weld zone and fusion line parameters to COR fitting degree ≥85% at the same time, and feed back the joint failure form in the test to S3 for optimizing the boundary condition setting of the model.

[0050] S5: Introduce a multi-physical field coupling adaptive optimization module, integrate the mechanical parameters of S1, the temperature field and displacement field data of S2-S4, establish a "sample-component-vehicle" three-level parameter mapping relationship, dynamically correct the material failure threshold value through machine learning algorithm, and the optimized parameters output by the module will guide the calibration process of S6 and the final model establishment of S8 at the same time.

[0051] S6: Start the model calibration closed loop process, perform the coupon level joint mechanical property test and the component level impact test, after the impact test, establish the joint detailed model, compare the joint detailed model with the coupon level joint mechanical property test, determine whether to match. If not, modify the establishment of the joint detailed model; if match, output the material parameter information.

[0052] S7: Establish the joint simplified model using the output material parameter information, compare the joint simplified model with the coupon level joint mechanical property test, determine whether to match. If not, modify the establishment of the joint simplified model; if match, preliminarily determine the joint simulation model.

[0053] S8: Establish the component impact test simulation model according to the preliminarily determined joint simulation model. Compare the component impact model with the component level impact test, determine whether to match. If not, modify the component impact model; if match, finally determine the joint simulation model.

[0054] In the present application, the method obtains the mechanical property parameters of the welded joint through the coupon level test, and establishes a solid element detailed model (element size 0.05 mm) and a shell element simplified model (grid size 0.4 mm) based on LS-DYNA software. The detailed model includes five parts of the upper plate, the lower plate, the aluminum interfacial intermediate phase, the weld zone and the fusion line, and the simplified model adopts shell element modeling, which is convenient for engineering application.

[0055] In the present application, in the model calibration process, by adjusting the material failure curve coefficient, the weld zone and the fusion line width, the correlation coefficient (COR) fitting degree of the simulation results and the test data (such as temperature-time, height difference-distance curve) is respectively reached above 90% (detailed model) and 85% (simplified model and component level model), which ensures the simulation accuracy.

[0056] In the present application, the component level impact test simulation model is established on the basis of the coupon model, the joint failure mode is verified by combining the test and simulation results, and finally the simulation model of the ultrasonic welded joint suitable for the whole vehicle crash analysis is established. The method effectively improves the accuracy and stability of the steel-aluminum hybrid vehicle body ultrasonic welded joint in CAE simulation, and is especially suitable for the safety performance development of new energy vehicle lightweight structure.

[0057] In step S1, when carrying out the joint mechanical property test of the sample, a multi-layer sheet lap joint structure is used for loading test. The loading rate is controlled to be consistent with the stress change rate in the actual welding process. The deformation data of the welded joint under different loads are collected by a high-precision sensor to obtain a complete load-displacement curve. The welding process parameters of the sample are matched with the ultrasonic frequency, amplitude and welding pressure in the actual production, so as to ensure that the test results can reflect the mechanical properties of the actual welded joint and provide reliable basic data support for subsequent model establishment.

[0058] In step S2, a detailed model of the joint sample test is established based on the LS-DYNA software. The modeling is carried out by using solid elements to capture the microstructure characteristics of the welded joint. The model includes the upper plate, the lower plate, the aluminum interfacial intermediate phase, the weld zone and the fusion line and other key components. The fusion depth size of each layer of plate is determined by analyzing the molten pool morphology of the sample joint, and the boundary range of the weld zone and the fusion line is divided in combination with the hardness distribution curve. In the model debugging stage, the material failure curve coefficient and the width parameters of the weld zone and the fusion line are adjusted. Through multiple iterations, the correlation coefficient fitting degree of the temperature-time curve and the height difference-distance curve output by the simulation and the test data is improved, so as to ensure that the model can accurately reproduce the dynamic mechanical behavior in the welding process.

[0059] In step S3, a simplified model of the sample test is constructed based on the detailed model. The modeling is carried out by using shell elements to reduce the calculation complexity. The model structure includes the upper plate base metal, the lower plate base metal, the weld zone and the fusion line. The material parameters of the simplified model are directly inherited from the calibration results of the detailed model. The grid size and element type are adjusted to balance the calculation efficiency and simulation accuracy. In the model debugging process, the simulation results of the temperature field and the displacement field are monitored in real time. The geometric parameters of the weld zone and the fusion line are fine-tuned to make the correlation coefficient fitting degree of the simulation curve of the simplified model and the test data meet the engineering application requirements, thereby providing an efficient and feasible model template for subsequent component-level simulation.

[0060] In step S4, the test device needs to simulate the impact load conditions in the actual use scenario. The deformation mode and failure time of the joint during the impact process are recorded by a high-speed camera and a dynamic data acquisition system. A component impact simulation model is established based on the calibrated simplified model. The material failure parameters and the geometric size of the weld zone are adjusted during the simulation process to make the correlation coefficient fitting degree of the simulation results and the test data meet the specified standard. The joint failure form observed in the test is compared and analyzed with the simulation results to verify the simulation accuracy of the model for the failure behavior of the welded joint, thereby providing a key basis for establishing the final model.

[0061] In step S5, dynamic parameter optimization is achieved by constructing a multi-scale data fusion framework. This module first extracts features from the load-displacement curves obtained in S1, identifying key mechanical features such as the elastic segment, yield point, and fracture threshold. Simultaneously, it integrates the temperature field distribution cloud maps and displacement field time series data output from the detailed and simplified models in S2-S4, and maps different physical field data to a unified coordinate system through a spatiotemporal alignment algorithm.

[0062] An improved radial basis function neural network is used to establish a three-level parameter transfer mechanism of "sample-component-vehicle". Sample-level data serves as the basic features of the network input layer, component-level impact response serves as the constraint condition of the intermediate layer, and finally outputs an equivalent parameter set adapted to the vehicle collision simulation. The module's built-in reinforcement learning agent uses the COR fit deviation as the reward signal and iteratively adjusts the correction coefficient of the material failure threshold through the Q-learning algorithm. When the simulation deviation between the sample level and the component level exceeds the threshold, the parameter backtracking mechanism is automatically triggered to retrieve the historical best parameters from the S6 calibration process for secondary training until a parameter package that meets the accuracy requirements of the final model in S8 is generated.

[0063] In step S5, the dynamic correction formula for the material failure threshold is:

[0064] ;

[0065] In the formula:

[0066] Represents the material failure threshold after the k-th iteration.

[0067] This represents the initial value of the failure threshold from the previous iteration.

[0068] α represents the adaptive learning rate (ranging from 0.02 to 0.15, determined by the variance of the sample-level experimental data).

[0069] ΔT represents the peak deviation of the temperature field between the current simulation and the experiment.

[0070] ΔT0 represents the reference temperature deviation (1.2 times the standard deviation of the sample-level test).

[0071] COR target This represents the target goodness of fit (0.9 for the detailed model and 0.85 for the simplified model).

[0072] COR current This represents the calculated correlation coefficient value for the current iteration;

[0073] This formula achieves nonlinear adjustment of the correction amount through the hyperbolic tangent function. When the temperature deviation and the fit deviation increase in the same direction, the correction magnitude of the failure threshold increases with a gradient, ensuring that the parameter adjustment can converge quickly and avoid overshoot, effectively solving the local optimum problem caused by traditional linear correction.

[0074] In step S6, when initiating the model calibration closed-loop process, the simulation results of the detailed model are first systematically compared with the experimental data of the sample-level mechanical properties, with a focus on checking the consistency of temperature field distribution, stress concentration areas, and load-displacement curves. If a significant deviation is found between the simulation results and the experimental data, the mesh generation accuracy or material constitutive parameters of the model need to be adjusted retrospectively. If the agreement meets the requirements, the verified material parameter information is output. These parameters will serve as the basic input for simplifying model construction and ensuring the reliability of the model in subsequent steps.

[0075] In step S7, a simplified joint model is established using the calibrated material parameters, and then compared and verified again with sample-level test data from multiple dimensions, including dynamic response characteristics, failure modes, and stress distribution at key locations. By adjusting the model's boundary conditions and contact settings, the balance between computational efficiency and simulation accuracy is optimized, ensuring that the mechanical behavior of the welded joint can still be accurately reflected while reducing computational costs. After successful verification, the structure and parameter configuration of the joint simulation model are preliminarily determined, providing a standardized model template for component-level simulation.

[0076] In step S8, based on the initially determined joint simulation model, a complete simulation system for component-level impact testing is constructed. The simulation results are comprehensively compared with component-level impact test data to verify the joint's failure modes, load-bearing capacity, and energy absorption characteristics under impact loads. Based on the comparison results, the model parameters are finally optimized, including key variables such as material failure threshold and weld zone stiffness, to ensure the model can accurately predict joint performance under different working conditions. The final simulation model needs to be validated through multiple sets of experimental data to ensure its stability and reliability in engineering applications.

[0077] From the above, we can conclude that:

[0078] This invention effectively solves the simulation problem of non-uniform mechanical properties of welded joints by providing a combination of simulation models from different dimensions. The detailed solid element model can accurately represent the microstructural characteristics of the welded joint, including the interface reaction zone of each plate layer and the weld fusion state, while the simplified shell element model significantly improves computational efficiency while ensuring engineering accuracy. The component impact test simulation model further combines material failure behavior with actual working conditions. The three models support each other through a unified material failure setting system, capturing both the performance differences between the weld zone and the base material and reflecting the dynamic response law under impact load, enabling the simulation results to realistically reproduce the mechanical behavior of the welded joint under complex stress conditions.

[0079] In this invention, key parameters are derived from a closed-loop verification process involving sample testing, component testing, and simulation calibration. This avoids deviations caused by theoretical assumptions and ensures the method has a solid foundation for engineering applications. By establishing a mapping relationship between core parameters and process conditions, technicians only need to adjust a few key variables to quickly adapt to joint simulation requirements under different welding process parameters, without having to repeatedly build a complete model. This parametric design not only lowers the simulation threshold but also enables rapid iterative optimization of process parameters in new product development, significantly shortening the technology verification cycle.

[0080] This invention accurately simulates the failure evolution path of ultrasonically welded joints in steel-aluminum hybrid vehicle bodies during a vehicle collision, including the entire process from initial deformation to complete fracture. By correlating the simulation results with vehicle safety performance indicators, welding process parameters and joint structure design can be optimized in a targeted manner, avoiding potential safety hazards that may arise in traditional empirical designs. This simulation-based forward development model provides scientific guidance for the connection reliability of lightweight vehicle bodies in new energy vehicles, helping to ensure collision safety performance while reducing vehicle weight.

[0081] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the term "comprising" or any other variations thereof is intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the element.

[0082] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A dynamic simulation method for ultrasonic welding process of multilayer thin plates, characterized in that: The method includes the following steps: S1: Conduct mechanical property tests on multi-layer thin plate lap joint samples, and obtain the basic mechanical parameters of the welded joint by measuring the load-displacement curve; S2: A detailed model of the sample was established based on LS-DYNA software. Solid elements were used to divide the upper plate, lower plate, aluminum interface intermediate phase, weld zone and fusion line. The melt pool morphology obtained from the S1 test was used to determine the melt depth parameters. The material failure curve coefficient and weld zone size were adjusted. S3: Construct a simplified shell element model of the sample, and integrate the material failure curve coefficients and weld zone parameters output from the detailed model in S2; S4: Conduct component-level impact tests, establish a simulation model based on the simplified model in S3, simultaneously adjust the parameters of the weld zone and fusion line to improve the COR fitting degree, and feed back the joint failure mode in the test to S3; S5: Through the multi-physics field coupling adaptive optimization module, the mechanical parameters of S1 and the temperature and displacement field data of S2-S4 are integrated to establish a three-level parameter mapping relationship between the sample, component and whole vehicle, and the material failure threshold is dynamically corrected through machine learning algorithm. S6: Initiate the model calibration closed-loop process, conduct sample-level joint mechanical property tests and component-level impact tests. After the impact tests are completed, establish a detailed joint model, compare the detailed joint model with the sample-level joint mechanical property test results, and determine whether they match. If they do not match, modify and establish the detailed joint model. If they match, output the material parameter information. S7: Use the output material parameter information to establish a simplified joint model, and compare the simplified joint model with the mechanical property test of the sample-level joint to determine whether they match. S8: Based on the preliminary joint simulation model, establish a component impact test simulation model; compare the component impact model with the component-level impact test to determine whether they match.

2. The dynamic simulation method for ultrasonic welding process of multilayer thin plates as described in claim 1, characterized in that: In step S1, when conducting mechanical property tests on the sample-level joints, a multi-layer thin plate lap structure is used for loading tests. The loading rate is controlled to be consistent with the stress change rate in the actual welding process. Deformation data of the welded joint under different loads are collected by a high-precision sensor to obtain a complete load-displacement curve. The welding process parameters of the sample are matched with the ultrasonic frequency, amplitude and welding pressure in actual production.

3. The dynamic simulation method for ultrasonic welding process of multilayer thin plates as described in claim 2, characterized in that: In step S2, a detailed model of the joint sample test is established based on LS-DYNA software. Solid elements are used for modeling to capture the microstructural features of the welded joint. The model includes key components such as the upper plate, lower plate, aluminum interface intermediate phase, weld zone, and fusion line. The weld depth of each plate layer is determined by analyzing the molten pool morphology of the sample joint. The boundary range of the weld zone and fusion line is divided by combining the hardness distribution curve. During the model debugging stage, the material failure curve coefficient and the width parameters of the weld zone and fusion line are adjusted. Through multiple iterations, the correlation coefficient fitting degree between the temperature-time curve and the height difference-distance curve output by the simulation and the experimental data is improved, so that the model can accurately reproduce the dynamic mechanical behavior during the welding process.

4. The dynamic simulation method for ultrasonic welding process of multilayer thin plates as described in claim 1, characterized in that: In step S3, a simplified model for the sample test is constructed based on the detailed model. Shell elements are used for modeling, and the model structure includes the upper plate base material, the lower plate base material, the weld zone, and the fusion line. The material parameters of the simplified model are directly inherited from the calibration results of the detailed model. The computational efficiency and simulation accuracy are balanced by adjusting the mesh size and element type.

5. The dynamic simulation method for ultrasonic welding process of multilayer thin plates as described in claim 4, characterized in that: In step S4, the test device needs to simulate the impact load conditions in the actual use scenario, and record the deformation mode and failure time of the joint during the impact process through a high-speed camera and dynamic data acquisition system; at the same time, a component impact simulation model is established based on the calibrated simplified model, and the material failure parameters and weld zone geometry are adjusted during the simulation process, and the joint failure mode observed in the test is compared and analyzed with the simulation results.

6. The method for kinetic simulation of ultrasonic welding process of multilayer thin plates as described in any one of claims 2-4, characterized in that: In step S5, dynamic parameter optimization is achieved by constructing a multi-scale data fusion framework. This module first extracts features from the load-displacement curve obtained in S1 and identifies key mechanical features such as elastic segment, yield point, and fracture threshold. At the same time, it integrates the temperature field distribution cloud map and displacement field time series data output by the detailed model and simplified model in S2-S4, and maps different physical field data to a unified coordinate system through a spatiotemporal alignment algorithm. An improved radial basis function neural network is used to establish a three-level parameter transfer mechanism of sample-component-vehicle. Sample-level data serves as the basic features of the network input layer, component-level impact response serves as the constraint condition of the intermediate layer, and finally outputs an equivalent parameter set adapted to the vehicle collision simulation. The module's built-in reinforcement learning agent uses the COR fit deviation as the reward signal and iteratively adjusts the correction coefficient of the material failure threshold through the Q-learning algorithm. When the simulation deviation between the sample level and the component level exceeds the threshold, the parameter backtracking mechanism is automatically triggered to retrieve the historical best parameters from the S6 calibration process for secondary training until a parameter package that meets the accuracy requirements of the final model in S8 is generated.

7. The dynamic simulation method for ultrasonic welding process of multilayer thin plates as described in claim 6, characterized in that: In step S5, the dynamic correction formula for the material failure threshold is: ； In the formula: This represents the material failure threshold after the k-th iteration; This represents the initial value of the failure threshold from the previous iteration; α represents the adaptive learning rate; ΔT represents the peak deviation of the temperature field between the current simulation and the experiment; ΔT0 represents the reference temperature deviation; COR target Indicates the target fit; COR current This represents the calculated correlation coefficient value for the current iteration.

8. The dynamic simulation method for ultrasonic welding process of multilayer thin plates as described in claim 1, characterized in that: In step S6, the model calibration closed-loop process is initiated, and the simulation results of the detailed model are systematically compared with the mechanical performance test data of the sample level to check the consistency of temperature field distribution, stress concentration area and load-displacement curve. If a significant deviation is found between the simulation results and the experimental data, the mesh generation accuracy of the model or the material constitutive parameters need to be adjusted retrospectively; if the consistency meets the requirements, the verified material parameter information is output.

9. The dynamic simulation method for ultrasonic welding process of multilayer thin plates as described in claim 8, characterized in that: In step S7, a simplified joint model is established using the calibrated material parameters, and then compared and verified with the sample-level test data in multiple dimensions, including dynamic response characteristics, failure modes, and stress distribution at key locations.

10. The dynamic simulation method for ultrasonic welding process of multilayer thin plates as described in claim 9, characterized in that: In step S8, based on the initially determined joint simulation model, a complete simulation system for component-level impact testing is constructed. The simulation results are then compared with the component-level impact test data to verify the failure mode, load-bearing capacity, and energy absorption characteristics of the joint under impact load.

Images