A method and system for predicting penetration depth of power battery busbar point ring laser welding

Abstract

The present invention belongs to, but is not limited to, the field of laser processing technology. It discloses a method and system for predicting the penetration depth of a power battery busbar spot ring laser welding. The method obtains the keyhole depth point data from the height between the upper surface of the base material and the laser head and the height between the bottom of the keyhole and the laser head during the current welding process. The difference between the two is the keyhole depth point data, and the keyhole curve data from the spot ring laser welding experiment is obtained. Outliers and multiple reflection points are removed, and the keyhole depth curve is refitted. The keyhole depth point data is preprocessed, and a curve fitting of the keyhole depth signal is performed using a sliding window minimum algorithm. A keyhole depth dataset is constructed based on the obtained keyhole curve data to predict the busbar penetration depth. The input parameters of the busbar penetration prediction model are selected based on actual working conditions. The constructed neural network model is used to predict the busbar penetration depth, and the predicted results are compared and analyzed with the actual penetration results. The method of the present invention is computationally simple, highly reliable, and highly practical.

Description

Technical Field

[0001] The present invention belongs to but is not limited to the field of laser processing technology, and in particular relates to a method and system for predicting the penetration depth of point ring laser welding of a power battery bus. Background Art

[0002] New energy vehicles are listed as an emerging industry in my country, and the country has set higher standards for their range and safety performance. As a core component of new energy vehicles, power batteries play a decisive role in the vehicle's range and performance. Power battery manufacturing costs account for approximately 40% of the total cost of new energy vehicles. To enhance the power performance of new energy vehicles, ensuring high-quality and efficient power battery manufacturing is crucial.

[0003] The busbar is responsible for collecting and distributing the power battery's current, as well as sampling voltage and temperature, and plays a crucial role in power battery safety. The connection of the power battery busbar is one of the core manufacturing processes of the power battery module. Compared to traditional laser welding for cladding and the manufacturing of other new energy vehicle components, power battery busbar welding faces more challenges. It is necessary to ensure that the upper plate is fully welded through and the lower plate is partially welded through during lap welding. Otherwise, the busbar will suffer from cold welds or weld penetration, leading to electrolyte leakage. At the same time, power battery welding involves many steps and high speeds, and the quality requirements for busbar welding are extremely high. Since the material is highly reflective, a spot ring laser has been proposed to improve welding quality in response to defects such as cold welds and spatter that currently occur in busbar welding.

[0004] The depth of penetration in busbar lap welds not only affects the thermal diffusion of the battery but also plays a decisive role in the joint strength and electrical conductivity. Optical coherence imaging technology, due to its advantages in depth measurement, has begun to be used for laser welding penetration monitoring. However, issues such as keyhole collapse, multiple beam reflections within the keyhole, and spike interference noise hinder high-precision busbar penetration monitoring, hindering further quality assessment and process research on power battery busbar laser welding.

[0005] Therefore, it is necessary to conduct a more in-depth study on busbar melt depth monitoring and propose a suitable busbar keyhole data denoising and melt depth prediction method to address the problems of current optical coherence imaging technology. Summary of the Invention

[0006] To address the challenges of existing technologies, this invention provides a method and system for denoising keyhole data and predicting weld penetration for spot-ring laser welding of power battery busbars for new energy vehicles. This technology is crucial for evaluating weld penetration and quality during spot-ring laser welding of power battery busbars, and also provides a theoretical basis for optimizing subsequent laser welding processes for power battery busbars.

[0007] The present invention is implemented as follows: a method for predicting the penetration depth of a power battery busbar point ring laser welding, comprising the following steps:

[0008] S1, based on the optical coherence imaging system, obtain the height DH(t) from the upper surface of the base material to the laser head and the height DP(t) from the bottom of the keyhole to the laser head during the current welding process;

[0009] S2, the optical coherence imaging system calculates the difference between the height DH(t) from the upper surface of the base material to the laser head and the height DP(t) from the bottom of the keyhole to the laser head to obtain the keyhole depth point data DP0(t);

[0010] S3, collecting and processing data through an optical coherent imaging system to obtain keyhole curve data DP1(t) in the spot ring laser welding experiment;

[0011] S4, remove outliers and multiple reflection points, and refit the keyhole depth curve;

[0012] S5, preprocessing the keyhole depth point data DP0(t) collected along the length direction of the weld using the optical coherent imaging system; the preprocessing includes removing outliers, multiple reflection points and spike interference noise;

[0013] S6, after removing outliers and spike interference noise, the keyhole depth signal is curve fitted by the sliding window minimum algorithm;

[0014] S7, constructing a keyhole depth data set based on the obtained keyhole curve data for predicting busbar penetration, and selecting input parameters of a busbar penetration prediction model based on actual working conditions;

[0015] S8, for the selected test data set, the constructed BP neural network model, GA-BP neural network model and neural network model integrating process parameters are used to predict the busbar melting depth, and the predicted results are compared and analyzed with the actual melting depth results.

[0016] Furthermore, in step S5, outliers and spike interference noise are removed by using a moving average filtering algorithm; the moving average filtering is based on calculating the average value of data within a certain window in the keyhole signal.

[0017] Furthermore, given a keyhole signal sequence d[n] containing N data, where n is the sequence index of the keyhole data (from 0 to N-1), filtering is performed by moving a window of length M on the keyhole signal sequence and calculating the average of the samples in the window;

[0018] For each position k of the sliding window, the filtered output h[k] can be calculated by the following formula:

[0019]

[0020] Where d[k] represents the sample value at index k in the keyhole signal sequence. 1 / M in the formula is a normalization factor used to average the keyhole data values ​​within the window.

[0021] Furthermore, in step S6, given a window size s, it is necessary to find the minimum value in each window in the data, and use the minimum value in each window (i.e., the deepest keyhole depth) as the keyhole data at the current moment. All the obtained keyhole data are connected in sequence to obtain the corresponding keyhole depth curve DP2(t).

[0022] Furthermore, in step S7, laser parameters (laser power and welding speed) are selected as input parameters of the busbar spot ring laser welding penetration prediction model.

[0023] Furthermore, in step S8, the prediction accuracy of the three models is judged by comparing the maximum absolute error (MAE) and root mean square error (RMSE) predicted by different models, and finally the penetration depth prediction model with the smallest error is selected as the penetration depth prediction model of the power battery bus.

[0024] Another object of the present invention is to provide a method for predicting the penetration depth of a power battery busbar point ring laser welding, and a system for predicting the penetration depth of a power battery busbar point ring laser welding, comprising:

[0025] Height acquisition module, which acquires the height DH(t) from the upper surface of the base material to the laser head and the height DP(t) from the bottom of the keyhole to the laser head during the current welding process based on the optical coherence imaging system;

[0026] Keyhole depth data point acquisition module: The optical coherent imaging system obtains the keyhole depth point data DP0(t) by calculating the difference between the height DH(t) from the upper surface of the base material to the laser head and the height DP(t) from the bottom of the keyhole to the laser head;

[0027] The keyhole curve data acquisition module collects and processes data through the optical coherence imaging system to obtain the keyhole curve data DP1(t) in the spot ring laser welding experiment;

[0028] Keyhole depth curve fitting module, which removes outliers and multiple reflection points and refits the keyhole depth curve;

[0029] The keyhole depth point data preprocessing module preprocesses the keyhole depth point data DP0(t) collected along the length of the weld using the optical coherent imaging system; the preprocessing includes removing outliers, multiple reflection points, and spike interference noise;

[0030] The keyhole depth signal curve fitting module performs curve fitting on the keyhole depth signal through the sliding window minimum algorithm after removing outliers and spike interference noise;

[0031] The prediction model input parameter determination module constructs a keyhole depth data set based on the obtained keyhole curve data to predict the busbar penetration depth. The input parameters of the busbar penetration prediction model are selected based on the actual working conditions.

[0032] The busbar melting depth prediction module predicts the busbar melting depth for the selected test data set using the constructed BP neural network model, GA-BP neural network model and neural network model integrating process parameters, and compares and analyzes the predicted results with the actual melting depth results.

[0033] Another object of the present invention is to provide a computer device, which includes a memory and a processor. The memory stores a computer program. When the computer program is executed by the processor, the processor executes the steps of the method for predicting the penetration depth of point ring laser welding of a power battery bus.

[0034] Another object of the present invention is to provide a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, the processor executes the steps of the method for predicting the penetration depth of point ring laser welding of a power battery bus.

[0035] Another object of the present invention is to provide an information data processing terminal, which includes the power battery busbar point ring laser welding penetration prediction system.

[0036] In combination with the above technical solutions and the technical problems solved, the advantages and positive effects of the technical solutions to be protected by the present invention are as follows:

[0037] First, the present invention provides a method and system for predicting the penetration depth of power battery bus laser welding, which is based on optical coherence imaging technology and related data processing methods, and effectively records the keyhole depth during the point-ring laser welding process of new energy vehicle bus, and can obtain the penetration depth result of power battery bus laser welding. This solution is simple to calculate, highly reliable, and highly practical. In the subsequent operation process, the penetration depth data results can be obtained without cutting the sample, avoiding the destruction of the bus cutting and tedious post-weld inspection process, greatly improving efficiency and saving costs. In addition, this method is suitable for penetration depth measurement under different welding process parameters (such as different point-ring ratios, welding speeds, defocus amounts, etc.), with high robustness and stability, and provides a new and reliable solution for process quality detection of power battery bus point-ring laser welding.

[0038] The method and system for calculating the penetration depth of point-ring laser welding for power battery busbars provided by this invention are simple and fast in actual operation, effectively improving the intelligent level of laser welding monitoring. In actual production, this method enables rapid measurement of the penetration depth of power battery busbars, thereby verifying the consistency of internal penetration and the quality of surface quality of the workpiece, evaluating weld quality, and providing strong guarantees for product reliability. Furthermore, the high-quality and efficient fitting of penetration depth provides a theoretical basis for research related to power battery busbar laser welding processes.

[0039] Second, the technical effects of the present invention are also reflected in the following important aspects:

[0040] (1) The technical solution of the present invention fills the technical gap in this field:

[0041] Currently, OCT has a good measurement effect on the normal keyhole morphology at low welding speeds. However, due to the high welding speed and spot-ring power difference of the new energy vehicle power battery busbar spot ring laser, errors are caused in the busbar penetration monitoring. Currently, there is little research in this field, which is a measurement technology problem. Therefore, the solution proposed in this patent fills the technical gap in this field and provides a strong guarantee for the high-quality and efficient manufacturing of power batteries.

[0042] (2) The technical solution of the present invention solves the technical problems that people have been eager to solve but have never been able to solve successfully:

[0043] my country's new energy vehicle sales have ranked first in the world for nine consecutive years, with sales of 9.495 million in 2023, a year-on-year increase of 40.4%. The current manufacturing cost of power batteries accounts for about 40% of the cost of new energy vehicles. However, the current domestic solutions for power battery penetration measurement mainly use vision and process control, and there are few related OCT studies in China. At the same time, the high welding speed of power batteries and the complex changes in keyhole morphology also affect penetration monitoring. At the same time, in order to solve the problems of keyhole oscillation and welding spatter, the solution of spot ring laser welding is also a new process that has just been proposed in China in recent years. The monitoring research on point ring laser buses is currently relatively blank and has always been eager to be solved. Therefore, the power battery bus spot ring laser welding penetration measurement method and system proposed in the present invention have put forward new ideas for the process and monitoring solutions in this field. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 This is a flow chart of a method for predicting penetration depth of a power battery busbar point ring laser welding provided by an embodiment of the present invention;

[0045] Figure 2 1 is a diagram showing the structure and measurement principle of an optical coherent imaging system provided by an embodiment of the present invention;

[0046] Figure 3 Schematic diagram of the tailing phenomenon occurring during busbar spot ring laser welding according to an embodiment of the present invention;

[0047] Figure 4 Figure a shows the keyhole morphology and partial reflection path of the measurement beam during busbar spot ring laser welding at low welding speeds provided by an embodiment of the present invention; and figure b shows the keyhole morphology and multiple reflections of the measurement beam in the keyhole during busbar spot ring laser welding at low welding speeds.

[0048] Figure 5 This is a measurement diagram of the original keyhole data of the power battery bus provided by an embodiment of the present invention;

[0049] Figure 6 3. This is a comparative schematic diagram of removing outliers, multiple reflection points, and tip interference noise from a power battery bus provided by an embodiment of the present invention;

[0050] Figure 7 2. It is a schematic diagram comparing the actual keyhole depth curve fitted by the power battery busbar provided by the embodiment of the present invention;

[0051] Figure 8 Schematic diagram of a neural network model for integrating process parameters provided by an embodiment of the present invention;

[0052] Figure 9 This is a structural diagram of a power battery busbar point ring laser welding penetration prediction system provided by an embodiment of the present invention.

[0053] Figure 10 This is a comparison diagram of the keyhole measurement and actual penetration depth of the power battery bus spot ring laser welding at a low welding speed provided by an embodiment of the present invention.

[0054] Figure 11 This is a comparison diagram of keyhole measurement and actual penetration depth of power battery busbar spot ring laser welding at high welding speed provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0055] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0056] Example 1: Real-time penetration prediction in power battery busbar welding

[0057] In the production of power batteries, the penetration depth of busbar welding is crucial to weld strength and conductivity. The system of the present invention enables real-time prediction and monitoring of penetration depth, improving welding quality.

[0058] 1. Use the optical coherent imaging system to obtain real-time welding height data DH(t) and DP(t) and calculate the keyhole depth point data DP0(t).

[0059] 2. Remove noise through curve fitting to generate a smooth keyhole depth signal curve DP1(t).

[0060] 3. Construct a prediction model input dataset containing laser power, welding speed, and keyhole depth curves.

[0061] 4. Use the GA-BP neural network model to predict the penetration depth and compare the predicted value with the actual measured value in real time.

[0062] 5. If the predicted value deviates from the set range, the system automatically adjusts the laser power and welding speed to achieve closed-loop control.

[0063] Example 2: Parameter optimization in spot ring laser welding experiment

[0064] In a laboratory environment, the performance of the penetration prediction model is optimized by controlling welding parameters and used to formulate production process standards.

[0065] 1. Set different laser powers (such as 1000W, 2000W, 3000W) and welding speeds (such as 60mm / s, 700mm / s, 80mm / s) and conduct spot ring laser welding experiments respectively.

[0066] 2. Use the optical coherence imaging system to collect the keyhole depth point data DP0(t) and curve data DP1(t).

[0067] 3. Preprocess the experimental data, remove outliers and spike noise, and fit the keyhole depth curve.

[0068] 4. Construct a data set containing multiple sets of experimental parameters, and use the BP neural network model and the neural network model fused with process parameters for training and prediction respectively.

[0069] 5. Compare the prediction results of different models and select the model with the smallest error as the production standard.

[0070] Effect:

[0071] The influence of laser power and welding speed on penetration depth is clarified.

[0072] It provides an accurate model input parameter configuration solution for industrial applications.

[0073] Example 3: Welding penetration assessment during power battery maintenance

[0074] During the maintenance of power batteries, it is necessary to evaluate whether the penetration depth of the welded busbar meets the standard requirements. The system of the present invention can achieve a rapid evaluation of the penetration depth.

[0075] 1. Install the optical coherence imaging device on the maintenance workbench, rescan the busbar with a laser, and collect keyhole depth data in the weld area.

[0076] 2. Use the preprocessing module to analyze the historical welding data, remove interference data, and fit the keyhole depth curve DP1(t).

[0077] 3. Input the actual scanning data into the penetration prediction module, and combine it with the repair process parameters (such as repair welding laser power and speed), and use the neural network model that integrates the process parameters to predict the penetration depth.

[0078] 4. The system automatically generates an evaluation report, indicating which welds have a penetration depth below the standard and recommending optimized parameters for repair welding.

[0079] Effect:

[0080] Quickly assess weld quality and reduce repair time.

[0081] Ensure that the busbar welding performance after repair meets the use requirements and improve the reliability of the power battery.

[0082] like Figure 1 As shown, an embodiment of the present invention provides a method for predicting the penetration depth of a power battery busbar point ring laser welding, comprising the following steps:

[0083] S1, based on the optical coherence imaging system, obtain the height DH(t) from the upper surface of the base material to the laser head and the height DP(t) from the bottom of the keyhole to the laser head during the current welding process;

[0084] S2, the optical coherence imaging system calculates the difference between the height DH(t) from the upper surface of the base material to the laser head and the height DP(t) from the bottom of the keyhole to the laser head to obtain the keyhole depth point data DP0(t);

[0085] Under the condition of ensuring good busbar lap weld formation, the spot ring laser welding process experiment of 1060 aluminum alloy was carried out using appropriate process parameters (spot ring laser power ratio, welding speed, defocus amount, etc.).

[0086] Specifically, in the laser welding experiment, plates of the same thickness as the power battery busbars were used and welded under the same process environment.

[0087] In order to reduce the interference of impurities on the workpiece surface and avoid affecting the joint quality during the welding process, the 1060 aluminum alloy plate was polished before welding, and the plate surface was cleaned with anhydrous alcohol to remove the oxide film and stains on the plate surface to prevent affecting the subsequent quality and measurement results.

[0088] S3, collecting and processing data through an optical coherent imaging system to obtain keyhole curve data DP1(t) in the spot ring laser welding experiment;

[0089] The keyhole depth data point DP0(t) and the original keyhole curve data DP1(t) in the spot ring laser welding experiment are as follows: Figure 5 shown.

[0090] S4, because the welding speed of the power battery busbar is faster than that of traditional welding, the tailing phenomenon of the keyhole is more serious during the welding process, so the first point where the laser and measuring beam enter the keyhole is not the lowest point of the keyhole, which will cause errors in the keyhole depth measurement, such as Figure 3 In addition, due to the keyhole appearance and the tailing phenomenon, the measuring beam of the optical coherence measurement system undergoes multiple reflections inside the keyhole during keyhole depth measurement, which makes the error of the keyhole depth value measured under high-speed welding even greater, as shown in Figure 2. Figure 4 Therefore, it is necessary to remove outliers and multiple reflection points and refit the keyhole depth curve.

[0091] S5, preprocessing the keyhole depth point data DP0(t) collected along the length direction of the weld using the optical coherent imaging system; the preprocessing includes removing outliers, multiple reflection points and spike interference noise.

[0092] Among them, the removal of outliers and spike interference noise is achieved through the moving average filtering algorithm.

[0093] Preferably, a moving average filter can be used to reduce noise in the keyhole depth signal or remove high-frequency components, thereby smoothing the signal. It is based on calculating the average value of data within a certain window in the keyhole signal.

[0094] Preferably, given a keyhole signal sequence d[n] containing N data, where n is the sequence index of the keyhole data (from 0 to N-1), filtering is performed by moving a window of length M on the keyhole signal sequence and calculating the average of the samples in the window.

[0095] For each position k of the sliding window, the filtered output h[k] can be calculated by the following formula:

[0096]

[0097] Where d[k] represents the sample value at index k in the keyhole signal sequence. 1 / M in the formula is a normalization factor used to average the keyhole data values ​​within the window.

[0098] Preferably, the data obtained after removing outliers and spike interference noise by moving average filtering algorithm is as follows: Figure 6 shown.

[0099] S6, after removing outliers and spike interference noise, the keyhole depth signal is curve fitted using the sliding window minimum algorithm.

[0100] Preferably, given a window size s, it is necessary to find the minimum value in each window in the data, and use the minimum value in each window (i.e., the deepest keyhole depth value) as the keyhole data at the current moment. All the keyhole data obtained are connected in sequence to obtain the corresponding keyhole depth curve DP2(t). The keyhole curve comparison finally obtained is Figure 7 The middle grey curve shows that compared with the original curve of the system, Figure 7 As shown by the black line in the middle, this method effectively removes outlier multiple reflection points and spike interference noise, and the fitting accuracy is higher.

[0101] Preferably, the power battery busbar material is 1 series aluminum alloy.

[0102] S7, constructing a keyhole depth data set based on the obtained keyhole curve data for predicting the busbar penetration depth, and selecting input parameters of the busbar penetration prediction model in combination with actual working conditions.

[0103] Preferably, in the actual laser process, keyhole depth data, welding speed, laser power, etc. will affect the penetration depth, among which welding speed is the most important influencing parameter. Especially for high-speed welding such as busbar, the keyhole tailing phenomenon will affect the penetration depth measurement accuracy, making the traditional single reliance on keyhole data input prediction not universal. Therefore, in addition to the traditional BPN and GA-BPN using keyhole data as input data, the present invention also selects laser parameters (laser power and welding speed) as input parameters of the penetration depth prediction model for busbar point ring laser welding.

[0104] Preferably, if Figure 8 As shown, a three-layer neural network structure is built, in which the input layer units are the input factors determined in step S7, namely keyhole depth, laser power and welding speed, and the output layer units are the busbar penetration data.

[0105] S8, for the selected test data set, the constructed BP neural network model, GA-BP neural network model and neural network model integrating process parameters are used to predict the busbar melting depth, and the predicted results are compared and analyzed with the actual melting depth results.

[0106] By comparing the maximum absolute error (MAE) and root mean square error (RMSE) of the predictions of different models, the prediction accuracy of the three models was judged, and finally the melting depth prediction model with the smallest error was selected as the melting depth prediction model for the power battery bus.

[0107] The method proposed in this invention relies on a coherent optical sensing system to measure the penetration depth of laser welding of power battery busbars. It consists of a light source, a spectroscope, an interferometer, an OCT optical fiber, a power supply and data transmission cable, a signal processing unit, an imaging unit, etc. It can measure the keyhole depth in real time during laser welding. The schematic diagram is shown in FIG. Figure 2 shown.

[0108] It should be noted that the optical coherence imaging system needs to calibrate the measuring beam before operation to ensure that it is coaxial with the keyhole centroid in order to accurately measure the busbar keyhole depth.

[0109] The optical coherence imaging system's imaging module is integrated into the laser head, and the system's sampling frequency is set to 70kHz. The galvanometer's oscillating reference point on the workpiece surface is set to 2mm directly in front of the keyhole. During the welding process, the OCT scans the workpiece surface at a rate of 70 measurement points per second.

[0110] Specifically, before conducting the laser welding experiment, an optical coherence imaging system is used to record the relative position of the center of the measuring beam and the centroid of the keyhole, and the center of the measuring beam is calibrated.

[0111] This method for predicting the penetration depth of spot-ring laser welding for power battery busbars does not require sample destruction, is simple and quick to operate, and exhibits good robustness and reliability. This method can also be applied to calculating the penetration depth of laser welding for power battery busbars under other processing environments and process parameters.

[0112] like Figure 9 As shown, an embodiment of the present invention provides a method for predicting the penetration depth of a power battery busbar point ring laser welding and a system for predicting the penetration depth of a power battery busbar point ring laser welding, comprising:

[0113] Height acquisition module, which acquires the height DH(t) from the upper surface of the base material to the laser head and the height DP(t) from the bottom of the keyhole to the laser head during the current welding process based on the optical coherence imaging system;

[0114] Keyhole depth data point acquisition module: The optical coherent imaging system obtains the keyhole depth point data DP0(t) by calculating the difference between the height DH(t) from the upper surface of the base material to the laser head and the height DP(t) from the bottom of the keyhole to the laser head;

[0115] The keyhole curve data acquisition module collects and processes data through the optical coherence imaging system to obtain the keyhole curve data DP1(t) in the spot ring laser welding experiment;

[0116] Keyhole depth curve fitting module, which removes outliers and multiple reflection points and refits the keyhole depth curve;

[0117] The keyhole depth point data preprocessing module preprocesses the keyhole depth point data DP0(t) collected along the length of the weld using the optical coherent imaging system; the preprocessing includes removing outliers, multiple reflection points, and spike interference noise;

[0118] The keyhole depth signal curve fitting module performs curve fitting on the keyhole depth signal through the sliding window minimum algorithm after removing outliers and spike interference noise;

[0119] The prediction model input parameter determination module constructs a keyhole depth data set based on the obtained keyhole curve data to predict the busbar penetration depth. The input parameters of the busbar penetration prediction model are selected based on the actual working conditions.

[0120] The busbar melting depth prediction module predicts the busbar melting depth for the selected test data set using the constructed BP neural network model, GA-BP neural network model and neural network model integrating process parameters, and compares and analyzes the predicted results with the actual melting depth results.

[0121] The power battery busbar point ring laser welding penetration prediction system uses an optical coherence imaging system to obtain height information during the welding process in real time.

[0122] The height DH(t) from the upper surface of the base material to the laser head and the height DP(t) from the bottom of the keyhole to the laser head were recorded.

[0123] By calculating the difference between the two, the keyhole depth point data DP0(t) is obtained, providing basic data for subsequent depth analysis.

[0124] This stage ensures accurate capture of key geometric parameters of the weld area, providing reliable input data for penetration prediction.

[0125] The keyhole depth curve data DP1(t) in the spot ring laser welding experiment was collected by an optical coherent imaging system.

[0126] After data collection is completed, outliers and multiple reflection points are removed.

[0127] Curve fitting technology is used to optimize the keyhole depth signal to ensure that the fitting result can truly reflect the variation law of the keyhole depth.

[0128] This process significantly improves the accuracy and applicability of the data, providing high-quality characteristic curves for the penetration prediction model.

[0129] To ensure data quality, the collected keyhole depth point data DP0(t) is preprocessed.

[0130] The algorithm removes outliers, multiple reflection points and spike interference noise to minimize experimental errors.

[0131] The sliding window minimum algorithm is applied to further fit the keyhole depth signal curve to make it smooth and physically meaningful.

[0132] The preprocessing stage solved the interference problem in the experimental data and laid a solid foundation for the selection of model input parameters.

[0133] After obtaining high-quality keyhole depth curve data, a complete data set is constructed for penetration depth prediction.

[0134] According to the experimental conditions and the characteristics of the keyhole depth data, the input parameters that have a significant impact on the prediction of the penetration depth are selected, such as the keyhole depth change rate, laser power, welding speed, etc.

[0135] These parameters serve as the input of the model to ensure that the prediction model can effectively capture the mapping relationship between welding process and penetration depth.

[0136] The system uses a variety of neural network models to predict busbar penetration, including:

[0137] BP neural network model: used to preliminarily establish a nonlinear mapping relationship for penetration depth prediction;

[0138] GA-BP neural network model: optimize the parameters of the BP model through genetic algorithm to improve prediction accuracy;

[0139] Neural network model integrating process parameters: This model comprehensively considers process parameters such as laser power and welding speed to optimize the penetration depth prediction in multiple dimensions.

[0140] The model generates a network structure that can accurately predict the melting depth by learning the training data set and the test data set.

[0141] The system verifies the performance of the prediction model through a test data set and compares and analyzes the predicted penetration depth with the actual measured penetration depth results.

[0142] The comparison indicators include prediction error, goodness of fit and generalization ability of the model.

[0143] By repeatedly optimizing the model structure and input parameters, we were able to achieve high-precision prediction of the penetration depth of point-ring laser welding of power battery busbars, significantly improving the automation level of welding process control and product quality stability.

[0144] This invention is primarily used in the laser manufacturing of new energy vehicles, such as power battery busbars and battery casings. my country has led the world in new energy vehicle sales for nine consecutive years, with sales expected to reach 9.495 million units in 2023, a year-on-year increase of 40.4%. Currently, power battery manufacturing costs account for approximately 40% of the total cost of new energy vehicles. Therefore, this invention has broad application prospects and considerable economic value in the new energy vehicle manufacturing sector. It can also be extended to other laser welding applications, providing support for manufacturing industries with stringent requirements for penetration depth and quality.

[0145] The present invention performs filtering and denoising on the noise points appearing in the measurement and extracts the keyhole depth. For example, Figure 5 、 Figure 6 、 Figure 7 As shown in the figure, the effectiveness of the filtering and keyhole curve extraction algorithm proposed in the present invention can be verified by comparison. At the same time, in the data obtained from the example welding, the keyhole depth curve measured by OCT in the 40mm / s example is shallower than the actual weld penetration curve, while the keyhole depth measured by OCT in the 100mm / s example is deeper than the actual weld penetration curve. This is because in spot ring laser welding, the laser power difference between the center beam and the outer ring beam is greater at low welding speed than at high welding speed within the same distance, resulting in Figure 4 There are more wine glass-shaped keyholes in a. At high welding speeds, due to inertia, the tailing phenomenon is more serious, and Figure 4 The keyhole morphology in b is more. The actual measurement results are as follows Figure 10 、 Figure 11 shown.

[0146] It should be noted that the embodiments of the present invention can be implemented by hardware, software, or a combination of software and hardware. The hardware portion can be implemented using dedicated logic; the software portion can be stored in a memory and executed by an appropriate instruction execution system, such as a microprocessor or dedicated design hardware. Those skilled in the art will appreciate that the above-mentioned devices and methods can be implemented using computer-executable instructions and / or contained in processor control code, for example, such as a carrier medium such as a disk, CD or DVD-ROM, a programmable memory such as a read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. The devices and modules of the present invention can be implemented by hardware circuits such as very large-scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, or programmable hardware devices such as field programmable gate arrays, programmable logic devices, etc., can also be implemented by software executed by various types of processors, or can be implemented by a combination of the above-mentioned hardware circuits and software, such as firmware.

[0147] The above description is only a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications, equivalent substitutions and improvements made by any technician familiar with this technical field within the technical scope disclosed by the present invention and within the spirit and principles of the present invention should be covered by the scope of protection of the present invention.

Claims

1. A method for predicting penetration depth of power battery busbar point ring laser welding, characterized in that: The following steps are involved: S1, based on the optical coherence imaging system, obtain the height DH(t) from the upper surface of the base material to the laser head and the height DP(t) from the bottom of the keyhole to the laser head during the current welding process; S2, the optical coherence imaging system calculates the difference between the height DH(t) from the upper surface of the base material to the laser head and the height DP(t) from the bottom of the keyhole to the laser head to obtain the keyhole depth point data DP0(t); S3, collecting and processing data through an optical coherent imaging system to obtain keyhole curve data DP1(t) in the spot ring laser welding experiment; S4, remove outliers and multiple reflection points, and refit the keyhole depth curve; S5, preprocessing the keyhole depth point data DP0(t) collected along the length direction of the weld using the optical coherent imaging system; the preprocessing includes removing outliers, multiple reflection points and spike interference noise; S6, after removing outliers, multiple reflection points and spike interference noise, the keyhole depth signal is curve fitted using the sliding window minimum algorithm; S7, constructing a keyhole depth data set based on the obtained keyhole curve data for predicting busbar penetration, and selecting input parameters of a busbar penetration prediction model based on actual working conditions; S8, for the selected test data set, the constructed BP neural network model, GA-BP neural network model and neural network model integrating process parameters are used to predict the busbar penetration depth, and the predicted results are compared and analyzed with the actual penetration depth results; In step S7, laser power and welding speed are selected as input parameters of the busbar spot ring laser welding penetration prediction model.

2. The method for predicting penetration depth of power battery busbar spot ring laser welding according to claim 1, characterized in that: In step S5, outliers and spike interference noise are removed by using a moving average filtering algorithm; the moving average filtering is based on calculating the average value of data within a certain window in the keyhole signal.

3. The method for predicting penetration depth of power battery busbar spot ring laser welding according to claim 2, characterized in that: Given a keyhole signal sequence d[n] containing N data, where n is the sequence index of the keyhole data, from 0 to N-1; filtering is performed by moving a window of length M on the keyhole signal sequence and calculating the average of the samples in the window; For each position k of the sliding window, the filtered output h[k] is calculated by the following formula: Where d[k] represents the sample value with index k in the keyhole signal sequence; 1 / M is the normalization factor used to average the keyhole data values ​​within the window.

4. The method for predicting penetration depth of power battery busbar spot ring laser welding according to claim 1, characterized in that: In step S6, a window size s is given, the minimum value in each window in the data is found, and the minimum value in each window is used as the keyhole data at the current moment. All the obtained keyhole data are sequentially connected to obtain the corresponding keyhole depth curve DP2(t).

5. The method for predicting penetration depth of power battery busbar spot ring laser welding according to claim 1, characterized in that: In step S8, the prediction accuracy of the three models is determined by comparing the maximum absolute error (MAE) and root mean square error (RMSE) predicted by different models, and the penetration depth prediction model with the smallest error is finally selected as the penetration depth prediction model for the power battery bus.

6. A power battery busbar spot ring laser welding penetration prediction system according to any one of claims 1 to 5, characterized in that: include: Height acquisition module, which acquires the height DH(t) from the upper surface of the base material to the laser head and the height DP(t) from the bottom of the keyhole to the laser head during the current welding process based on the optical coherence imaging system; Keyhole depth data point acquisition module: The optical coherent imaging system obtains the keyhole depth point data DP0(t) by calculating the difference between the height DH(t) from the upper surface of the base material to the laser head and the height DP(t) from the bottom of the keyhole to the laser head; The keyhole curve data acquisition module collects and processes data through the optical coherence imaging system to obtain the keyhole curve data DP1(t) in the spot ring laser welding experiment; Keyhole depth curve fitting module, which removes outliers and multiple reflection points and refits the keyhole depth curve; The keyhole depth point data preprocessing module preprocesses the keyhole depth point data DP0(t) collected along the length of the weld using the optical coherent imaging system; the preprocessing includes removing outliers, multiple reflection points, and spike interference noise; The keyhole depth signal curve fitting module performs curve fitting on the keyhole depth signal using a sliding window minimum algorithm after removing outliers, multiple reflection points, and spike interference noise; The prediction model input parameter determination module constructs a keyhole depth data set based on the obtained keyhole curve data to predict the busbar penetration depth. The input parameters of the busbar penetration prediction model are selected based on the actual working conditions. The busbar melting depth prediction module predicts the busbar melting depth for the selected test data set using the constructed BP neural network model, GA-BP neural network model and neural network model integrating process parameters, and compares and analyzes the predicted results with the actual melting depth results.

7. A computer device comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the method for predicting the penetration depth of a power battery busbar spot ring laser welding as described in any one of claims 1 to 5.

8. A computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the steps of the method for predicting penetration depth of spot ring laser welding of a power battery busbar according to any one of claims 1 to 5.

9. An information data processing terminal, comprising the power battery busbar spot ring laser welding penetration prediction system according to claim 6.

Images