Process parameter optimization method and equipment for ultrasonic metal welding

Abstract

The invention provides a technological parameter optimization method and equipment for ultrasonic metal welding. The method comprises the steps that multiple sets of to-be-optimized technological parameters of multiple technological stages in an ultrasonic metal welding technology are received; inputting the plurality of groups of to-be-optimized process parameters into a process parameter prediction layer to obtain a plurality of prediction process parameters in one-to-one correspondence with the plurality of process stages; when the plurality of prediction process parameters exceed a set parameter range, adjusting and optimizing the plurality of groups of to-be-optimized process parameters based on a parameter adjusting and optimizing module to obtain a plurality of groups of optimized process parameters; and when the multiple prediction process parameters are within the set parameter range, the multiple prediction process parameters are input into the welding quality prediction layer, and the predicted welding quality is obtained. The process parameter prediction layer can predict the process parameters of all the process stages, the parameter adjusting and optimizing module adjusts and optimizes the multiple sets of to-be-optimized process parameters based on the process parameter prediction result, the consistency of the optimized process parameters can be improved, and then the accuracy of welding quality prediction is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of ultrasonic metal welding, and in particular to a process parameter optimization method and equipment for ultrasonic metal welding. Background Art

[0002] Ultrasonic Metal Welding (UMW) is a solid-phase welding technology that utilizes high-frequency ultrasonic energy. Because UMW is less dependent on materials, it minimizes the formation of intermetallic compounds and energy loss in the contact zone. It has been widely used in the production of automotive parts such as lithium-ion battery tabs, wire terminals, and electronic devices. A major challenge facing UMW technology is how to accurately assess weld quality, which is closely related to the input or set process parameters. To ensure that weld quality meets requirements, appropriate process parameters must be input or set.

[0003] The method of ensuring welding quality in the prior art is: before actual welding, the process parameters corresponding to the expected welding quality are determined through the debugging stage. Specifically: different process parameters are manually input to obtain the welding quality corresponding to the process parameters. If the welding quality does not meet the requirements, the process parameters are manually adjusted until the optimized process parameters corresponding to the welding quality that meets the requirements are found. It has the following technical problems: 1. The ultrasonic metal welding process is a staged sequential process with many process parameters, and there is a coupling relationship between the process parameters in different stages, which makes the debugging process complicated, and thus causes the problem of low efficiency in process parameter optimization. 2. The ultrasonic metal welding process has process parameters between the process parameters and the final quality prediction. In the process of debugging the process parameters, the prior art only considers the final quality and does not consider the process parameters. In the production process, due to inconsistencies in production materials, environment and equipment, the same process parameters cannot guarantee the output of products of the same quality, that is, the optimized process parameters have poor adaptability and cannot meet the welding quality under any production conditions.

[0004] Therefore, there is an urgent need to provide a method and equipment for optimizing process parameters for ultrasonic metal welding to achieve automated optimization of process parameters and improve the efficiency of process parameter optimization. In addition, the process parameters should be considered during the process parameter optimization process to achieve the adaptability of the process parameters to factors such as production materials and the environment, thereby improving the quality of products produced according to the optimized process parameters. Summary of the Invention

[0005] In view of this, it is necessary to provide a method and equipment for optimizing the process parameters of ultrasonic metal welding to solve the technical problems in the prior art of manual optimization of process parameters, low optimization efficiency, and failure to consider the process parameters of ultrasonic metal welding, resulting in low quality of products produced by optimized process parameters.

[0006] In a first aspect, to solve the above-mentioned technical problems, the present invention provides a method for optimizing process parameters of ultrasonic metal welding, wherein the process parameters of ultrasonic metal welding are optimized based on a process parameter optimization model, wherein the process parameter optimization model includes a hierarchical analysis model and a parameter tuning module, wherein the hierarchical analysis model includes a process parameter prediction layer and a welding quality prediction layer connected in sequence, and the input and output of the parameter tuning module are respectively connected to the output and input of the process parameter prediction layer; The method comprises: receiving a plurality of sets of process parameters to be optimized at a plurality of process stages in an ultrasonic metal welding process; Inputting the multiple sets of process parameters to be optimized into the process parameter prediction layer to perform process parameter prediction, and obtaining multiple predicted process parameters corresponding to the multiple process stages; When the plurality of predicted process parameters exceed the set parameter range, the plurality of sets of process parameters to be optimized are tuned based on the parameter tuning module to obtain a plurality of sets of optimized process parameters; the plurality of predicted control parameters corresponding to the plurality of sets of optimized process parameters are within the set parameter range; When the plurality of prediction process parameters are within the set parameter range, the plurality of prediction process parameters are input into the welding quality prediction layer to perform quality prediction and obtain predicted welding quality.

[0007] In one possible implementation, the multiple process stages include a pressing stage, an ultrasonic vibration stage, a pressure holding stage, and a post-weld processing stage, and the multiple groups of process parameters to be optimized include pressing parameters to be optimized, ultrasonic vibration parameters to be optimized, pressure holding parameters to be optimized, and post-weld processing parameters to be optimized, and the multiple predicted process parameters include pressing process prediction parameters, ultrasonic vibration process prediction parameters, pressure holding process prediction parameters, and post-weld processing process prediction parameters.

[0008] In a possible implementation, the process parameter prediction layer includes a pressing process parameter prediction unit, an ultrasonic vibration process parameter prediction unit, a pressure holding process parameter prediction unit, and a post-weld processing process parameter prediction unit; The pressing process parameter prediction unit is used to predict the process parameters of the pressing stage based on the pressing parameters to be optimized to obtain the pressing process prediction parameters; The ultrasonic vibration process parameter prediction unit is used to predict the process parameters of the ultrasonic vibration stage based on the ultrasonic vibration parameters to be optimized and the downward pressure process prediction parameters to obtain the ultrasonic vibration process prediction parameters; The pressure holding process parameter prediction unit is used to predict the process parameters of the pressure holding stage based on the pressure holding parameters to be optimized and the ultrasonic vibration process prediction parameters to obtain the pressure holding process prediction parameters; The post-weld processing parameter prediction unit is used to predict the process parameters of the post-weld processing stage based on the post-weld processing parameters to be optimized and the pressure holding process prediction parameters to obtain the post-weld processing process prediction parameters.

[0009] In a possible implementation, the welding quality prediction layer includes a pressing quality prediction unit, an ultrasonic vibration quality prediction unit, a pressure holding quality prediction unit, and a post-weld treatment quality prediction unit; The pressing quality prediction unit is used to predict the quality of the pressing stage based on the pressing process prediction parameters to obtain implicit information on the pressing quality; The ultrasonic vibration quality prediction unit is used to predict the quality of the ultrasonic vibration stage based on the ultrasonic vibration process prediction parameters and the downward pressure quality implicit information to obtain the ultrasonic vibration quality implicit information; The pressure holding quality prediction unit is used to predict the quality of the pressure holding stage based on the pressure holding process prediction parameters and the ultrasonic vibration quality implicit information to obtain the pressure holding quality implicit information; The post-weld processing quality prediction unit is used to predict the quality of the post-weld processing stage based on the post-weld processing process prediction parameters and the pressure holding quality implicit information to obtain the predicted welding quality.

[0010] In a possible implementation, the parameter tuning module includes a downward pressure parameter tuning unit, an ultrasonic vibration parameter tuning unit, a pressure holding parameter tuning unit, and a post-weld processing parameter tuning unit; The pressing parameter tuning unit is used to optimize the pressing parameter to be optimized to obtain a target pressing parameter when the pressing process prediction parameter exceeds a set pressing parameter range; the pressing process prediction parameter corresponding to the target pressing parameter is within the set pressing parameter range; The ultrasonic vibration parameter tuning unit is used to optimize the ultrasonic vibration parameter to be optimized to obtain a target ultrasonic vibration parameter when the ultrasonic vibration process prediction parameter exceeds a set ultrasonic vibration parameter range; the target ultrasonic vibration parameter is within the set ultrasonic vibration parameter range; The pressure holding parameter tuning unit is used to optimize the pressure holding parameter to be optimized to obtain a target pressure holding parameter when the pressure holding process prediction parameter exceeds a set pressure holding parameter range; the target pressure holding parameter is within the set pressure holding parameter range; The post-weld processing parameter tuning unit is used to optimize the post-weld processing parameters to be optimized to obtain target post-weld processing parameters when the post-weld processing process prediction parameters exceed the set post-weld processing parameter range; the target post-weld processing parameters are within the set post-weld processing parameter range.

[0011] In a possible implementation, the hierarchical analysis model further includes a process parameter input layer, which includes a holding pressure parameter input unit, an ultrasonic vibration parameter input unit, a holding pressure parameter input unit, and a post-weld processing parameter input unit; The holding pressure parameter input unit is used to input the pressing parameter to be optimized into the pressing process parameter prediction unit; The ultrasonic vibration parameter input unit is used to input the ultrasonic vibration parameter to be optimized into the ultrasonic vibration process parameter prediction unit; The pressure holding parameter input unit is used to input the pressure holding parameter to be optimized into the pressure holding process parameter prediction unit; The post-weld processing parameter input unit is used to input the post-weld processing parameter to be optimized into the post-weld processing process parameter prediction unit.

[0012] In a possible implementation, the hierarchical analysis model further includes a first multi-head self-attention layer provided between the process parameter input layer and the process parameter prediction layer; The first multi-head self-attention layer is used to perform self-attention learning on the multiple groups of process parameters to be optimized to obtain attention process parameter features; The process parameter prediction layer is used to perform process parameter prediction on the attention process parameter features to obtain the multiple process prediction parameters.

[0013] In a possible implementation, the hierarchical analysis model further includes a second multi-head self-attention layer provided between the process parameter prediction layer and the welding quality prediction layer; The second multi-head self-attention layer is used to perform self-attention learning on the multiple process prediction parameters to obtain attention process parameter features; The welding quality prediction layer is used to perform quality prediction on the attention process parameters to obtain the predicted welding quality.

[0014] In one possible implementation, the pressing parameters to be optimized include pressing pressure, the ultrasonic vibration parameters to be optimized include a first variable amplitude of the horn, a first ultrasonic vibration power, and a welding time, the holding pressure parameters to be optimized include a holding pressure time, and the post-weld processing parameters to be optimized include a second variable amplitude of the horn, a second ultrasonic vibration frequency, and a post-processing time. The prediction parameters of the pressing process include the peak value of the welding head depth and the average value of the weldment temperature in the pressing stage; the prediction parameters of the ultrasonic vibration process include the peak value of the welding head depth, the average value of the welding head depth, the standard deviation of the welding head depth, the peak value of the weldment temperature, the average value of the weldment temperature, the standard deviation of the weldment temperature, the peak value of the welding head amplitude, the average value of the weldment temperature and the standard deviation of the welding head amplitude in the ultrasonic vibration stage; the prediction parameters of the holding pressure process include the peak value of the welding head depth and the average value of the weldment temperature in the holding pressure stage; the prediction parameters of the post-weld treatment process include the average value of the weldment temperature, the peak value of the welding head amplitude, the average value of the welding head amplitude and the standard deviation of the welding head amplitude in the post-weld treatment stage; The predicted welding quality includes predicted joint tensile strength and predicted contact resistance.

[0015] In a second aspect, the present invention further provides a process parameter optimization device for ultrasonic metal welding, comprising a process parameter collector and a process parameter optimization module; The process parameter collector is used to collect multiple groups of process parameters to be optimized in multiple process stages of the ultrasonic metal welding process; The process parameter optimization module includes a memory and a processor, wherein: The memory is used to store programs; The processor is coupled to the memory and is configured to execute the program stored in the memory to implement the steps of the method for optimizing process parameters of ultrasonic metal welding described in any one of the possible implementations above.

[0016] The beneficial effects of the present invention are as follows: the process parameter optimization method for ultrasonic metal welding provided by the present invention realizes automatic optimization of the process parameters of ultrasonic metal welding based on the process parameter optimization model, without manual parameter adjustment, thereby improving the optimization efficiency of the process parameters. In addition, the present invention sets a hierarchical analysis model including a process parameter prediction layer, which realizes the prediction of process parameters of each process stage through the process parameter prediction layer, and sets a parameter tuning module. When multiple predicted process parameters exceed the set parameter range, multiple groups of process parameters to be optimized are tuned based on the parameter tuning module to obtain optimized process parameters. The matching degree between the process parameters and production materials, environment and equipment can be optimized, that is, the optimized process parameters will be optimized in real time as the production materials, environment and equipment change, reducing the adverse effects of factors such as materials and environment on production quality, and further improving the rationality of the determined optimized process parameters to ensure the accuracy of the welding quality prediction of ultrasonic metal welding.

[0017] Furthermore, the hierarchical analysis model in the present invention uses the idea of ​​hierarchical analysis to characterize the nonlinear relationship between process parameters, process parameters and welding quality, and decouples process parameters and process parameters based on process stages. For example, the process parameters and process parameters of the current process stage will not affect the previous process stage, thereby reducing the model solution speed and further improving the optimization efficiency of the process parameters. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0019] Figure 1 A schematic flow chart of an embodiment of a method for optimizing process parameters of ultrasonic metal welding provided by the present invention; Figure 2 A schematic diagram of an embodiment of the process parameter optimization model provided by the present invention; Figure 3 This is a schematic structural diagram of an embodiment of the process parameter optimization equipment for ultrasonic metal welding provided by the present invention. DETAILED DESCRIPTION

[0020] The following will provide a clear and complete description of the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0021] It should be understood that the schematic drawings are not drawn to scale. The flowcharts used in the present invention illustrate operations implemented according to some embodiments of the present invention. It should be understood that the operations of the flowcharts can be implemented out of sequence, and steps that have no logical contextual relationship can be reversed in order or implemented simultaneously. In addition, those skilled in the art, guided by the content of the present invention, can add one or more other operations to the flowcharts or remove one or more operations from the flowcharts. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in the form of software, or in one or more hardware modules or integrated circuits, or in different networks and / or processor systems and / or microcontroller systems.

[0022] References herein to "embodiments" mean that a particular feature, structure, or characteristic described in connection with the embodiments may be included in at least one embodiment of the present invention. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it constitute a separate or alternative embodiment that is mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0023] The present invention provides a method and device for optimizing process parameters of ultrasonic metal welding, which are described below.

[0024] Figure 1 This is a flow chart of an embodiment of the method for optimizing process parameters of ultrasonic metal welding provided by the present invention. Figure 2 A schematic diagram of an embodiment of the process parameter optimization model provided by the present invention is shown in FIG. Figure 1 and Figure 2 As shown, the process parameter optimization model includes a hierarchical analysis model and a parameter tuning module ( Figure 2 The hierarchical analysis model includes the process parameter prediction layers ( Figure 2 The module composed of dot-dashed line units in the figure) and the welding quality prediction layer, the input and output of the parameter tuning module are connected to the output and input of the process parameter prediction layer respectively; The process parameter optimization methods for ultrasonic metal welding include: S101, receiving multiple sets of process parameters to be optimized in multiple process stages of an ultrasonic metal welding process.

[0025] The number of the multiple sets of process parameters to be optimized corresponds to the number of the multiple process stages. For example, if the ultrasonic metal welding process includes four stages, the multiple sets of process parameters to be optimized include four sets, which correspond to the process stages one by one.

[0026] S102 , inputting multiple sets of process parameters to be optimized into a process parameter prediction layer to perform process parameter prediction, and obtaining multiple predicted process parameters corresponding to multiple process stages.

[0027] It should be noted that: since the ultrasonic metal welding process is time-series, the process parameter prediction layer should be a time-series prediction model. Specifically, the process parameter prediction layer is a long short-term memory network (LSTM) model.

[0028] S103. When multiple predicted process parameters exceed the set parameter range, multiple groups of process parameters to be optimized are tuned based on the parameter tuning module to obtain multiple groups of optimized process parameters; multiple predicted control parameters corresponding to the multiple groups of optimized process parameters are within the set parameter range.

[0029] Among them, the parameter tuning module has built-in optimization algorithms, such as particle swarm optimization algorithm, ant colony optimization algorithm, gray wolf optimization algorithm and other group optimization algorithms, which are used to optimize the process parameters to be optimized within the set parameter range and obtain the optimized process parameters.

[0030] In a specific embodiment of the present invention, the process of obtaining multiple sets of optimized process parameters in step S103 will go through at least one iteration, and the results of each iteration are multiple sets of iterative process parameters. The corresponding prediction control parameters are determined based on the iterative process parameters, and it is judged whether the prediction control parameters are within the set parameter range. If so, the current iterative process parameters are the optimized process parameters.

[0031] S104. When the multiple prediction process parameters are within the set parameter range, the multiple prediction process parameters are input into the welding quality prediction layer to perform quality prediction and obtain predicted welding quality.

[0032] It should be understood that the set parameter range can be set or adjusted according to the actual application scenario and is not specifically limited here.

[0033] Among them, the welding quality prediction layer should also be a time series prediction model, specifically, the welding quality prediction layer is also an LSTM model.

[0034] The memory unit in the LSTM model uses an input gate, forget gate, and output gate structure to memorize long-term information between time series and discard useless information. This makes it effective for learning time series relationships between high-dimensional data. The specific structure of the LSTM model is not described here.

[0035] It should be noted that the process parameter optimization model needs to be trained, verified and tested based on actual data before it can be used.

[0036] Specifically, to mitigate the large fluctuations in validation scores caused by a small number of validation data points, a k-fold cross-validation method was used to train the model. During training, the root mean square error (RMSE) was used as the model's loss function. During parameter optimization, gradient descent was used to adjust model parameters until the loop termination criteria were met. The final model parameters were then saved to obtain a usable process parameter optimization model.

[0037] It should be understood that the ultrasonic metal welding process parameter optimization method of the embodiments of the present invention can be implemented in any device based on the ultrasonic metal welding process parameter optimization method, such as an ultrasonic metal welding control device. Specifically, the ultrasonic metal welding process parameter optimization method is stored in the aforementioned device as a pre-programmed program. When the device is powered on, the program is invoked and the ultrasonic metal welding process parameter optimization method is implemented.

[0038] Compared with the prior art, the process parameter optimization method for ultrasonic metal welding provided by the embodiment of the present invention realizes automatic optimization of the process parameters of ultrasonic metal welding based on the process parameter optimization model, without manual parameter adjustment, thereby improving the optimization efficiency of the process parameters. In addition, the embodiment of the present invention sets a hierarchical analysis model including a process parameter prediction layer, which realizes the prediction of process parameters of each process stage through the process parameter prediction layer, and sets a parameter tuning module. When multiple predicted process parameters exceed the set parameter range, the parameter tuning module is used to adjust multiple groups of process parameters to be optimized to obtain optimized process parameters. The matching degree between the process parameters and production materials, environment and equipment can be optimized. That is, the optimized process parameters will be optimized in real time as the production materials, environment and equipment change, reducing the adverse effects of factors such as materials and environment on production quality, thereby further improving the rationality of the determined optimized process parameters to ensure the accuracy of the welding quality prediction of ultrasonic metal welding.

[0039] Furthermore, the hierarchical analysis model in the embodiment of the present invention uses the idea of ​​the hierarchical analysis method (AHP) to characterize the nonlinear relationship between process parameters, process parameters and welding quality, and decouples the process parameters and process parameters based on the process stage. For example, the process parameters and process parameters of the current process stage will not affect the previous process stage, thereby reducing the model solution speed and further improving the optimization efficiency of the process parameters.

[0040] In some embodiments of the present invention, the multiple process stages include a pressing stage, an ultrasonic vibration stage, a pressure holding stage, and a post-weld processing stage.

[0041] (1) Downward pressure stage: In this stage, the welding head of the welding equipment, under the action of the cylinder or other driving device, applies a certain pressure to the metal workpiece to be welded placed below, so that the upper and lower metal workpieces are in close contact, and the pressure is maintained until the end of the pressure holding stage. During this period, the key process parameter is the downward pressure P.

[0042] (2) Ultrasonic vibration stage: The ultrasonic generator generates a high-frequency electrical signal, which is converted into mechanical vibration energy through the transducer. The amplitude is then amplified by the amplitude transformer and transmitted to the welding head, causing the welding head to generate high-frequency ultrasonic vibration. The parameters that affect the welding quality in this process mainly include the first amplitude of the amplitude transformer C1, the first ultrasonic vibration power W1, and the welding time T1.

[0043] (3) Pressure holding stage: When entering the pressure holding stage, the ultrasonic vibration stops, but the welding head still maintains the same pressure on the workpiece, so that the weld joint gradually cools and solidifies under the action of pressure to form a stable solid-state connection. In this stage, P remains unchanged. Another important parameter is the pressure holding time T2.

[0044] (4) Post-processing stage: To prevent the weldment from adhering to the welding head, a slight vibration is often applied to the welding head after the pressure holding period. This vibration can destroy any microscopic connection between the weldment and the welding head, thereby separating them. This stage is similar to the ultrasonic vibration stage. The main parameters are the second variable amplitude C2, the second ultrasonic vibration power W2, and the post-processing time T3.

[0045] The process parameters to be optimized in the above stages can be expressed as:

[0046] In the formula X 1=P, X 2=(C1,W1,T2) T , X 3=T2, X 4=(C2,W2,T3) T .

[0047] The multiple sets of process parameters to be optimized include the pressing parameters to be optimized X 1. Ultrasonic vibration parameters to be optimized X 2. Pressure holding parameters to be optimized X 3 and post-weld processing parameters to be optimized X 4.

[0048] Likewise, the plurality of predicted process parameters include a pressing process prediction parameter, an ultrasonic vibration process prediction parameter, a pressure holding process prediction parameter, and a post-weld treatment process prediction parameter.

[0049] In a specific embodiment of the present invention, the prediction parameters of the pressing process include the peak depth D1 of the welding head and the average temperature E1 of the weldment in the pressing stage; the prediction parameters of the ultrasonic vibration process include the peak depth D2 of the welding head, the average depth D3 of the welding head, the standard deviation D4 of the welding head depth, the peak temperature E2 of the weldment, the average temperature E3 of the weldment, the standard deviation E4 of the weldment temperature, the peak amplitude S1 of the welding head, the average amplitude S2 of the welding head and the standard deviation S3 of the welding head amplitude in the ultrasonic vibration stage; the prediction parameters of the holding pressure process include the peak depth D5 of the welding head and the average temperature E5 of the weldment in the holding pressure stage; the prediction parameters of the post-weld processing process include the average temperature E6 of the weldment in the post-weld processing stage, the peak amplitude S4 of the welding head, the average amplitude S5 of the welding head and the standard deviation S6 of the welding head amplitude.

[0050] Specifically, the prediction process parameter is represented by Z, which is:

[0051] Where Z1=(D1,E1) T , Z2=(D2,D3,D4,E2,E3,E4,S1,S2,S3) T ,Z3=(D5, E5)T ,Z3=(E6, S4,S5,S6) T .

[0052] In a specific embodiment of the present invention, the predicted welding quality includes predicting the joint tensile strength and predicting the contact resistance. The predicted welding quality is represented by Y, which is specifically:

[0053] Where, F To predict the tensile strength of joints; R To predict the contact resistance.

[0054] Since in the actual production process, the previous process parameters will affect the next process parameters, in order to improve the accuracy of the overall process parameter prediction, in some embodiments of the present invention, such as Figure 2 As shown, the process parameter prediction layer includes a pressing process parameter prediction unit, an ultrasonic vibration process parameter prediction unit, a pressure holding process parameter prediction unit and a post-weld treatment process parameter prediction unit; The pressing process parameter prediction unit is used to predict the process parameters of the pressing stage based on the pressing parameters to be optimized, and obtain the pressing process prediction parameters; The ultrasonic vibration process parameter prediction unit is used to predict the process parameters of the ultrasonic vibration stage based on the ultrasonic vibration parameters to be optimized and the downward pressure process prediction parameters to obtain the ultrasonic vibration process prediction parameters; The pressure holding process parameter prediction unit is used to predict the process parameters of the pressure holding stage based on the pressure holding parameters to be optimized and the ultrasonic vibration process prediction parameters to obtain the pressure holding process prediction parameters; The post-weld processing parameter prediction unit is used to predict the process parameters of the post-weld processing stage based on the post-weld processing parameters to be optimized and the pressure holding process prediction parameters to obtain the post-weld processing process prediction parameters.

[0055] When predicting process parameters, the latter parameter prediction unit of the embodiment of the present invention takes into account the influence of the predicted process parameters predicted by the previous parameter prediction unit, thereby further improving the accuracy of the predicted process parameters in each process stage. In addition, orderly prediction can also improve the prediction efficiency of the predicted process parameters.

[0056] In a specific embodiment of the present invention, Figure 2 As shown, the welding quality prediction layer includes a pressing quality prediction unit, an ultrasonic vibration quality prediction unit, a pressure holding quality prediction unit and a post-weld treatment quality prediction unit; The pressing quality prediction unit is used to predict the quality of the pressing stage based on the pressing process prediction parameters to obtain the pressing quality implicit information; The ultrasonic vibration quality prediction unit is used to predict the quality of the ultrasonic vibration stage based on the ultrasonic vibration process prediction parameters and the implicit information of the downward pressure quality, and obtain the implicit information of the ultrasonic vibration quality; The pressure holding quality prediction unit is used to predict the quality of the pressure holding stage based on the pressure holding process prediction parameters and the ultrasonic vibration quality implicit information, and obtain the pressure holding quality implicit information; The post-weld treatment quality prediction unit is used to predict the quality of the post-weld treatment stage based on the post-weld treatment process prediction parameters and the implicit information of the pressure holding quality to obtain the predicted welding quality.

[0057] When predicting process parameters, the latter quality prediction unit of the embodiment of the present invention takes into account the influence predicted by the previous quality prediction unit, thereby further improving the accuracy of the quality implicit information of each process stage, and thus improving the accuracy of the final predicted welding quality.

[0058] In a specific embodiment of the present invention, Figure 2 As shown, the parameter tuning module includes a downward pressure parameter tuning unit, an ultrasonic vibration parameter tuning unit, a pressure holding parameter tuning unit and a post-weld processing parameter tuning unit; The pressing parameter tuning unit is used to optimize the pressing parameters to be optimized to obtain target pressing parameters when the predicted parameters of the pressing process exceed the set pressing parameter range; the pressing process predicted parameters corresponding to the target pressing parameters are within the set pressing parameter range; The ultrasonic vibration parameter tuning unit is used to optimize the ultrasonic vibration parameters to be optimized to obtain target ultrasonic vibration parameters when the predicted parameters of the ultrasonic vibration process exceed the set ultrasonic vibration parameter range; the target ultrasonic vibration parameters are within the set ultrasonic vibration parameter range; The pressure holding parameter tuning unit is used to optimize the pressure holding parameters to be optimized and obtain the target pressure holding parameters when the predicted parameters of the pressure holding process exceed the set pressure holding parameter range; the target pressure holding parameters are within the set pressure holding parameter range; The post-weld processing parameter tuning unit is used to optimize the post-weld processing parameters to be optimized when the predicted parameters of the post-weld processing process exceed the set post-weld processing parameter range to obtain the target post-weld processing parameters; the target post-weld processing parameters are within the set post-weld processing parameter range.

[0059] Because the predicted parameters of the previous process are not within the set parameter range, the prediction of the parameters of the subsequent process will be inaccurate. Therefore, in the preferred embodiment of the present invention, the ultrasonic vibration parameter tuning unit is tuned based on the fact that the down-press parameter tuning unit has already obtained the target down-press parameter. That is, the parameters of the subsequent process stage are tuned only after the accuracy of the previous process stage is ensured. Similarly, the holding pressure parameter tuning unit is tuned based on the fact that the target ultrasonic vibration signal has been obtained, and the post-weld processing parameter tuning unit is tuned based on the fact that the target holding pressure parameter has been obtained.

[0060] Since the input process parameters to be optimized are diverse, in order to further improve the optimization efficiency of the process parameters, in some embodiments of the present invention, for example Figure 1 As shown, the hierarchical analysis model also includes a process parameter input layer, which includes a holding pressure parameter input unit, an ultrasonic vibration parameter input unit, a holding pressure parameter input unit, and a post-weld processing parameter input unit; The holding pressure parameter input unit is used to input the pressing parameters to be optimized into the pressing process parameter prediction unit; The ultrasonic vibration parameter input unit is used to input the ultrasonic vibration parameters to be optimized into the ultrasonic vibration process parameter prediction unit; The pressure holding parameter input unit is used to input the pressure holding parameters to be optimized into the pressure holding process parameter prediction unit; The post-weld processing parameter input unit is used to input the post-weld processing parameters to be optimized into the post-weld processing process parameter prediction unit.

[0061] The embodiment of the present invention sets up a parameter input unit for each group of process parameters to be optimized, so as to achieve a one-to-one correspondence between the input process parameters and the process parameter prediction unit, thereby avoiding the problem of parameter selection or inputting wrong parameters in the process parameter prediction unit, and ensuring the accuracy of process parameter optimization while improving the efficiency of process parameter optimization.

[0062] In order to further improve the prediction accuracy of process prediction parameters and provide optimization direction for process parameter optimization, in some embodiments of the present invention, Figure 1 As shown, the hierarchical analysis model also includes a first multi-head self-attention (MHSA) layer arranged between the process parameter input layer and the process parameter prediction layer; The first multi-head self-attention layer is used to perform self-attention learning on multiple sets of process parameters to be optimized to obtain attention process parameter features; The process parameter prediction layer is used to predict process parameters based on the attention process parameter features to obtain multiple process prediction parameters.

[0063] By implementing a first multi-head self-attention layer, the present invention can evaluate the contribution of different historical inputs to the output based on multiple sets of parameters to be optimized, adding more weight to key feature information to improve the accuracy of process prediction parameters. Furthermore, because the first multi-head self-attention layer assigns different weights to different inputs, the weights represent the degree of influence of each input process parameter on the process prediction parameter. Therefore, they provide guidance for subsequent process parameter optimization, thereby improving the efficiency and accuracy of process parameter optimization.

[0064] Among them, the process of the multi-head attention layer is: the query vector, key vector and value vector are obtained based on the input through the parameter matrix mapping of the three science departments, and then the similarity between the features is measured by the dot product of the query vector and the key vector. The attention score is then weighted by the softmax function. Finally, the outputs of all heads are spliced ​​and the final feature representation is obtained by changing the learnable output weight matrix.

[0065] Furthermore, in order to improve the prediction accuracy of welding quality, in some embodiments of the present invention, as Figure 2 As shown, the hierarchical analysis model also includes a second multi-head self-attention layer arranged between the process parameter prediction layer and the welding quality prediction layer; The second multi-head self-attention layer is used to perform self-attention learning on multiple process prediction parameters to obtain attention process parameter features; The welding quality prediction layer is used to perform quality prediction on the attention process parameters and obtain the predicted welding quality.

[0066] By setting a second multi-head self-attention layer, the embodiment of the present invention can assign different weights according to the input of multiple different process prediction parameters, add more weights to the key process prediction parameters, and thus improve the accuracy of the final predicted welding quality.

[0067] To verify the effectiveness of the process parameter optimization model proposed in the embodiments of the present invention, in some embodiments of the present invention, 10 test sets are used to predict welding quality. The error between the prediction results and the experimental results is shown in Table 1: Table 1 Prediction and experimental results of tensile strength and contact resistance

[0068] Table 1 shows that the average relative percentage errors for tensile strength and contact resistance obtained by the process parameter optimization model proposed in this embodiment of the present invention are 3.21% and 3.7%, respectively. This demonstrates that the model's predictions for the test set samples are very close to the actual measured values, and that the model has high predictive and generalization capabilities. Furthermore, while the predicted error for contact resistance is also very small, it is still higher than the error for tensile strength. This result is primarily due to measurement error in contact resistance.

[0069] Furthermore, to verify whether the introduction of the multi-head attention mechanism into the process parameter optimization model effectively improves the prediction ability of welding quality and to verify the efficiency of the proposed model, the same test set samples were used to compare the quality prediction of three trained models: MHSA-LSTM-AHP (a process parameter optimization model that includes the first and second multi-head self-attention layers), MHSA-AHP (a process parameter optimization model that does not include the first and second multi-head self-attention layers), and GA-BP. The comparison indicators used were mean square error (MSE), RMSE, and mean absolute percentage error (MAPE). The comparison results are shown in Table 2: Table 2 Performance comparison of different models

[0070] Table 2 shows that for tensile strength prediction, the MHSA-LSTM-AHP model achieved MAE, RMSE, and MAPE of 15.1, 16.16, and 0.032, respectively. These results represent significant reductions compared to the LSTM-AHP and GA-BP algorithms. Similarly, the error reduction for resistance prediction is also significant, demonstrating that the MHSA-LSTM-AHP model possesses superior predictive performance and generalization capabilities. Compared to the MHSA-LSTM algorithm, the MAE, RMSE, and MAPE of the MHSA-LSTM-AHP are reduced by 17%, 13.4%, and 16.3%, respectively. For resistance prediction, the reductions are even greater, at 24%, 21.5%, and 24.7%. This demonstrates that the self-attention mechanism introduced in the LSTM-AHP significantly improves the model's predictive capabilities. The errors of GA-BP, LSTM-AHP, and MHSA-LSTM-AHP in resistance prediction are all higher than those in tensile strength prediction, but the error increase of GA-BP is greater. This shows that LSTM-AHP and MHSA-LSTM-AHP, which can learn temporal dependencies, can mine more information and show better robustness, while the performance of traditional GA-BP is somewhat weak.

[0071] In summary, compared with the traditional GA-BP model and LSTM-AHP model, MHSA-LSTM-AHP has better prediction ability and better robustness. It is superior to the other two models in all indicators. When predicting resistance, the error is much smaller than that of the GA-BP and LSTM-AHP models.

[0072] like Figure 3 As shown, the present invention also provides a process parameter optimization device 300 for ultrasonic metal welding. The process parameter optimization device 300 for ultrasonic metal welding includes a process parameter collector 310 and a process parameter optimization module 320; The process parameter collector 310 is used to collect multiple sets of process parameters to be optimized in multiple process stages of the ultrasonic metal welding process; The process parameter optimization module 320 includes a processor 321 , a memory 322 , and a display 323 . Figure 3 Only some components of the ultrasonic metal welding process parameter optimization apparatus 300 are shown, but it should be understood that it is not required to implement all of the shown components, and more or fewer components may be implemented instead.

[0073] In some embodiments, the processor 321 may be a central processing unit (CPU), a microprocessor, or other data processing chip, configured to execute program codes or process data stored in the memory 322 , such as the process parameter optimization method for ultrasonic metal welding of the present invention.

[0074] In some embodiments, the memory 322 may be an internal storage unit of the ultrasonic metal welding process parameter optimization device 300, such as a hard disk or memory of the ultrasonic metal welding process parameter optimization device 300. In other embodiments, the memory 322 may also be an external storage device of the ultrasonic metal welding process parameter optimization device 300, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) card, a flash card, etc., equipped in the ultrasonic metal welding process parameter optimization device 300.

[0075] Furthermore, the memory 322 may include both an internal storage unit of the ultrasonic metal welding process parameter optimization device 300 and an external storage device. The memory 322 is used to store application software and various data of the ultrasonic metal welding process parameter optimization device 300.

[0076] In some embodiments, the display 323 can be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen. The display 323 is used to display information about the ultrasonic metal welding process parameter optimization device 300 and to display a visual user interface. The components 321-323 of the ultrasonic metal welding process parameter optimization device 300 communicate with each other via a system bus.

[0077] In some embodiments of the present invention, when the processor 321 executes the ultrasonic metal welding process parameter optimization program in the memory 322, the following steps may be implemented: receiving a plurality of sets of process parameters to be optimized at a plurality of process stages in an ultrasonic metal welding process; Inputting multiple sets of process parameters to be optimized into the process parameter prediction layer to perform process parameter prediction, and obtaining multiple predicted process parameters corresponding to multiple process stages; When multiple predicted process parameters exceed the set parameter range, multiple groups of process parameters to be optimized are tuned based on the parameter tuning module to obtain multiple groups of optimized process parameters; multiple predicted control parameters corresponding to the multiple groups of optimized process parameters are within the set parameter range; When the multiple prediction process parameters are within the set parameter range, the multiple prediction process parameters are input into the welding quality prediction layer to perform quality prediction and obtain the predicted welding quality.

[0078] It should be understood that, when the processor 321 executes the ultrasonic metal welding process parameter optimization program in the memory 322 , in addition to the above functions, it can also implement other functions. For details, please refer to the description of the corresponding method embodiment above.

[0079] The ultrasonic metal welding process parameter optimization device 300 in the embodiment of the present invention can be installed in passenger cars and commercial vehicles.

[0080] Accordingly, an embodiment of the present invention also provides a computer-readable storage medium, which is used to store computer-readable programs or instructions. When the program or instructions are executed by a processor, it can implement the steps or functions in the process parameter optimization method for ultrasonic metal welding provided in the above-mentioned method embodiments.

[0081] Those skilled in the art will appreciate that all or part of the process flow of the above-described method embodiment can be implemented by instructing related hardware (such as a processor, controller, etc.) through a computer program, and the computer program can be stored in a computer-readable storage medium. The computer-readable storage medium may be a magnetic disk, an optical disk, a read-only memory, or a random access memory.

[0082] The above is a detailed introduction to the process parameter optimization method and equipment for ultrasonic metal welding provided by the present invention. Specific examples are used in this article to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea; at the same time, for technical personnel in this field, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A method for optimizing process parameters of ultrasonic metal welding, characterized in that: Optimizing process parameters for ultrasonic metal welding based on a process parameter optimization model, the process parameter optimization model comprising a hierarchical analysis model and a parameter tuning module, the hierarchical analysis model comprising a process parameter prediction layer and a welding quality prediction layer connected in sequence, the input and output of the parameter tuning module being connected to the output and input of the process parameter prediction layer, respectively; The method comprises: receiving a plurality of sets of process parameters to be optimized at a plurality of process stages in an ultrasonic metal welding process; Inputting the multiple sets of process parameters to be optimized into the process parameter prediction layer to perform process parameter prediction, and obtaining multiple predicted process parameters corresponding to the multiple process stages; When the plurality of predicted process parameters exceed the set parameter range, the plurality of sets of process parameters to be optimized are tuned based on the parameter tuning module to obtain a plurality of sets of optimized process parameters; the plurality of predicted control parameters corresponding to the plurality of sets of optimized process parameters are within the set parameter range; When the plurality of prediction process parameters are within the set parameter range, the plurality of prediction process parameters are input into the welding quality prediction layer to perform quality prediction and obtain predicted welding quality.

2. The process parameter optimization method for ultrasonic metal welding according to claim 1, characterized in that: The multiple process stages include a pressing stage, an ultrasonic vibration stage, a pressure holding stage, and a post-weld processing stage. The multiple groups of process parameters to be optimized include pressing parameters to be optimized, ultrasonic vibration parameters to be optimized, pressure holding parameters to be optimized, and post-weld processing parameters to be optimized. The multiple predicted process parameters include pressing process prediction parameters, ultrasonic vibration process prediction parameters, pressure holding process prediction parameters, and post-weld processing process prediction parameters.

3. The process parameter optimization method for ultrasonic metal welding according to claim 2, characterized in that: The process parameter prediction layer includes a pressing process parameter prediction unit, an ultrasonic vibration process parameter prediction unit, a pressure holding process parameter prediction unit and a post-weld processing process parameter prediction unit; The pressing process parameter prediction unit is used to predict the process parameters of the pressing stage based on the pressing parameters to be optimized to obtain the pressing process prediction parameters; The ultrasonic vibration process parameter prediction unit is used to predict the process parameters of the ultrasonic vibration stage based on the ultrasonic vibration parameters to be optimized and the downward pressure process prediction parameters to obtain the ultrasonic vibration process prediction parameters; The pressure holding process parameter prediction unit is used to predict the process parameters of the pressure holding stage based on the pressure holding parameters to be optimized and the ultrasonic vibration process prediction parameters to obtain the pressure holding process prediction parameters; The post-weld processing parameter prediction unit is used to predict the process parameters of the post-weld processing stage based on the post-weld processing parameters to be optimized and the pressure holding process prediction parameters to obtain the post-weld processing process prediction parameters.

4. The process parameter optimization method for ultrasonic metal welding according to claim 2, characterized in that: The welding quality prediction layer includes a pressing quality prediction unit, an ultrasonic vibration quality prediction unit, a pressure holding quality prediction unit and a post-weld treatment quality prediction unit; The pressing quality prediction unit is used to predict the quality of the pressing stage based on the pressing process prediction parameters to obtain implicit information on the pressing quality; The ultrasonic vibration quality prediction unit is used to predict the quality of the ultrasonic vibration stage based on the ultrasonic vibration process prediction parameters and the downward pressure quality implicit information to obtain the ultrasonic vibration quality implicit information; The pressure holding quality prediction unit is used to predict the quality of the pressure holding stage based on the pressure holding process prediction parameters and the ultrasonic vibration quality implicit information to obtain the pressure holding quality implicit information; The post-weld processing quality prediction unit is used to predict the quality of the post-weld processing stage based on the post-weld processing process prediction parameters and the pressure holding quality implicit information to obtain the predicted welding quality.

5. The process parameter optimization method for ultrasonic metal welding according to claim 3, characterized in that: The parameter tuning module includes a downward pressure parameter tuning unit, an ultrasonic vibration parameter tuning unit, a pressure holding parameter tuning unit and a post-weld processing parameter tuning unit; The pressing parameter tuning unit is used to optimize the pressing parameter to be optimized to obtain a target pressing parameter when the pressing process prediction parameter exceeds a set pressing parameter range; the pressing process prediction parameter corresponding to the target pressing parameter is within the set pressing parameter range; The ultrasonic vibration parameter tuning unit is used to optimize the ultrasonic vibration parameter to be optimized to obtain a target ultrasonic vibration parameter when the ultrasonic vibration process prediction parameter exceeds a set ultrasonic vibration parameter range; the target ultrasonic vibration parameter is within the set ultrasonic vibration parameter range; The pressure holding parameter tuning unit is used to optimize the pressure holding parameter to be optimized to obtain a target pressure holding parameter when the pressure holding process prediction parameter exceeds a set pressure holding parameter range; the target pressure holding parameter is within the set pressure holding parameter range; The post-weld processing parameter tuning unit is used to optimize the post-weld processing parameters to be optimized to obtain target post-weld processing parameters when the post-weld processing process prediction parameters exceed the set post-weld processing parameter range; the target post-weld processing parameters are within the set post-weld processing parameter range.

6. The process parameter optimization method for ultrasonic metal welding according to claim 3, characterized in that: The hierarchical analysis model further includes a process parameter input layer, which includes a pressure holding parameter input unit, an ultrasonic vibration parameter input unit, a pressure holding parameter input unit, and a post-weld processing parameter input unit; The holding pressure parameter input unit is used to input the pressing parameter to be optimized into the pressing process parameter prediction unit; The ultrasonic vibration parameter input unit is used to input the ultrasonic vibration parameter to be optimized into the ultrasonic vibration process parameter prediction unit; The pressure holding parameter input unit is used to input the pressure holding parameter to be optimized into the pressure holding process parameter prediction unit; The post-weld processing parameter input unit is used to input the post-weld processing parameter to be optimized into the post-weld processing process parameter prediction unit.

7. The process parameter optimization method for ultrasonic metal welding according to claim 6, characterized in that: The hierarchical analysis model further includes a first multi-head self-attention layer disposed between the process parameter input layer and the process parameter prediction layer; The first multi-head self-attention layer is used to perform self-attention learning on the multiple groups of process parameters to be optimized to obtain attention process parameter features; The process parameter prediction layer is used to perform process parameter prediction on the attention process parameter features to obtain the multiple process prediction parameters.

8. The process parameter optimization method for ultrasonic metal welding according to any one of claims 1 to 6, characterized in that: The hierarchical analysis model further includes a second multi-head self-attention layer disposed between the process parameter prediction layer and the welding quality prediction layer; The second multi-head self-attention layer is used to perform self-attention learning on the multiple process prediction parameters to obtain attention process parameter features; The welding quality prediction layer is used to perform quality prediction on the attention process parameters to obtain the predicted welding quality.

9. The process parameter optimization method for ultrasonic metal welding according to claim 2, characterized in that: The pressing parameters to be optimized include the pressing pressure, the ultrasonic vibration parameters to be optimized include the first variable amplitude of the horn, the first ultrasonic vibration power and the welding time, the holding pressure parameters to be optimized include the holding pressure time, and the post-weld processing parameters to be optimized include the second variable amplitude of the horn, the second ultrasonic vibration frequency and the post-processing time; The prediction parameters of the pressing process include the peak value of the welding head depth and the average value of the weldment temperature in the pressing stage; the prediction parameters of the ultrasonic vibration process include the peak value of the welding head depth, the average value of the welding head depth, the standard deviation of the welding head depth, the peak value of the weldment temperature, the average value of the weldment temperature, the standard deviation of the weldment temperature, the peak value of the welding head amplitude, the average value of the weldment temperature and the standard deviation of the welding head amplitude in the ultrasonic vibration stage; the prediction parameters of the holding pressure process include the peak value of the welding head depth and the average value of the weldment temperature in the holding pressure stage; the prediction parameters of the post-weld treatment process include the average value of the weldment temperature, the peak value of the welding head amplitude, the average value of the welding head amplitude and the standard deviation of the welding head amplitude in the post-weld treatment stage; The predicted welding quality includes predicted joint tensile strength and predicted contact resistance.

10. A process parameter optimization device for ultrasonic metal welding, characterized in that: Including process parameter collector and process parameter optimization module; The process parameter collector is used to collect multiple sets of process parameters to be optimized in multiple process stages of the ultrasonic metal welding process; The process parameter optimization module includes a memory and a processor, wherein: The memory is used to store programs; The processor is coupled to the memory and is configured to execute the program stored in the memory to implement the steps of the method for optimizing process parameters of ultrasonic metal welding as described in any one of claims 1 to 9.

Images