A method for selecting an aluminum tube welding parameter boundary

Abstract

This invention discloses a method for selecting welding parameter boundaries for aluminum tubes, relating to the field of aluminum tube welding. The method includes the following steps: S1, setting a fixed linear velocity benchmark; S2, obtaining the critical power point through small-sample experiments; S3, establishing linear models for cold welding boundaries and overheating boundaries; S4, generating a dynamic operating window; and S5, dynamic correction and continuous learning. This invention scientifically selects a linear velocity benchmark, accurately obtains the critical power point through small-sample step-by-step experiments, and constructs a linear model using the least squares method to avoid discrete interpolation errors, achieving full linear velocity boundary prediction. It introduces a dynamic safety redundancy strategy to generate operating windows and recommended operating points, improving process fluctuation tolerance and ensuring weld stability. Through a dynamic correction mechanism and a composite sampling scheme, the model is continuously optimized to adapt to production changes, improving the accuracy, stability, and production adaptability of welding parameter selection, and reducing the risk of quality accidents.

Description

Technical Field

[0001] This invention relates to the field of aluminum tube welding technology, specifically a method for selecting the boundary of aluminum tube welding parameters. Background Technology

[0002] Aluminum tubes, with their advantages of light weight, good thermal conductivity, and strong corrosion resistance, are used in refrigeration and air conditioning, aerospace, automobile manufacturing, fluid transportation and other fields. Welding is the core process in aluminum tube processing and forming, and the rationality of the selection of process parameters determines the quality of the weld.

[0003] In existing technologies, the welding operation window in aluminum tube welding production is easily affected by multiple coupled factors such as fluctuations in ambient temperature and humidity, batch differences in aluminum tube material, and the operating status of welding equipment. This causes the safe range of power-linear speed to deviate. Traditional processes use an offline static calibration mode, which cannot detect the drift changes of the operation window in real time during production. This results in increased costs for scrapping and reworking aluminum tube materials. Furthermore, when production conditions change, it is impossible to quickly transmit standardized parameter boundary ranges, leading to poor welding quality consistency. Summary of the Invention

[0004] The purpose of this invention is to provide a method for selecting the welding parameter boundaries of aluminum tubes, in order to solve the problem mentioned in the background art that the welding operation window is easily affected by multiple coupling factors, causing the power-linear velocity safety range to deviate, making it impossible to quickly transmit the standardized parameter boundary range, resulting in poor welding quality consistency.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for selecting the boundary of welding parameters for aluminum tubes, comprising the following steps: S1. Set a fixed linear speed benchmark: Select the commonly used linear speed of the current production line as the benchmark, record the current environmental conditions, aluminum tube batch, and equipment status as the benchmark working condition for this calibration, conduct a stability pre-test, and confirm that the current working condition is stable. S2. Obtaining the critical power point through small sample experiments: Select experimental samples and obtain the critical power for cold welding and overheating through small sample experiments under a fixed linear velocity reference. S3. Establish linear models for cold welding boundary and overheating boundary: Preprocess the critical power data and construct linear equations for cold welding boundary and overheating boundary based on the least squares method. S4. Generate a dynamic operating window: Substitute the real-time linear velocity into the model to calculate the theoretical critical power, and introduce a dynamic safety redundancy strategy to shrink the theoretical boundary inward to form the actual operating window. The center of the calculated window is used as the recommended operating point. S5. Dynamic Correction and Continuous Learning: Incremental iteration of the model is achieved by introducing time weighting and sliding window mechanisms. A hierarchical early warning system is established based on dynamic confidence intervals to adjust for minor process drifts and update boundary data.

[0006] Preferably, in step S1, the selection of the linear speed benchmark is based on production coverage and model robustness. The linear speed with the highest daily usage ratio of the production line is selected. When the production line covers multiple linear speeds, the usage frequency and capacity ratio of each linear speed are calculated, and the speed is selected accordingly.

[0007] Preferably, in step S2, obtaining the critical power point through a small-sample experiment specifically includes the following steps: S21. Sample and Consumable Preparation: Cut aluminum tubes from the current batch, remove oil stains, discard the initial section during the experiment, test the effective middle section, calibrate the ultrasonic testing instrument, and preset the experimental parameters. S22. Cold Welding Critical Power Test: At a fixed linear speed, set the initial power to 80% of the normal power, and gradually increase the power in increments of 0.1kW. After welding the aluminum tube at each power level, use ultrasonic testing to determine if there are any incomplete fusion defects. Record the power value at which no cold welding defects are found for the first time as the cold welding critical power. ; S23. Overheating Critical Power Test: Set the initial power to 120% of the normal power, and increase the power in increments of 0.1kW. After each power level, determine whether overheating defects occur by metallographic observation or surface morphology inspection. Record the power level preceding the first occurrence of an overheating defect as the overheating critical power. ; S24. Multi-linear velocity verification: Select 2-3 different linear velocities and weld samples with different linear velocities onto the aluminum tube to obtain the corresponding linear velocity values. and .

[0008] Preferably, in step S3, establishing the linear model of the cold weld boundary and the overheated boundary specifically includes the following steps: S31. Data cleaning: Perform data integrity checks, physical rationality verification, and outlier analysis on the experimental data obtained in step S2. S32. Linear fitting of cold weld boundary and overheat boundary: Substitute the data into the linear equations of cold weld boundary and overheat boundary, solve the coefficients by the least squares method, and obtain the cold weld boundary model and overheat boundary model. S33. Select intermediate linear velocities that were not involved in the fitting and substitute them into the model to calculate theoretical values ​​for experimental verification. Select points near the boundary for extrapolation verification, evaluate the applicable boundary of the model, and quantify the prediction error of the model.

[0009] Preferably, in step S32, the linear equation of the cold welding boundary is set as follows: The linear equation for the overheated boundary is: ; Substitute the cleaned cold welding data into the normal equation system: ; Substitute the superheat data after cleaning into the normal equations: ; Solving in sequence, we get , , and By obtaining the value of , the cold welding boundary model and the overheating boundary model can be obtained; in, Indicates the critical power for cold welding. The intercept term represents the boundary of the cold weld. The term represents the slope at the cold weld boundary, and v represents the welding line velocity. Indicates the superheat critical power. The intercept term representing the overheated boundary, The slope term represents the overheated boundary.

[0010] Preferably, in step S4, the dynamic operation window first reads the current line speed setting value of the production line in real time. It automatically retrieves the parameters of the cold welding and overheating boundary models and calculates the theoretical critical power. and An inward contraction strategy was adopted to obtain the allowance on the cold welding side. and superheat side margin The actual operation window is , Calculate the golden midpoint within the operation window. .

[0011] The preferred formula for calculating the midpoint of the golden ratio is as follows: .

[0012] Preferably, in step S5, dynamic correction and continuous learning adopt a composite scheme of timed basic sampling and triggered enhanced sampling to maintain the frequency of sampling one set of welding samples at regular intervals for appearance inspection, ultrasonic non-destructive testing and mechanical property sampling, and increase the sampling frequency at key production nodes.

[0013] Compared with the prior art, the beneficial effects of the present invention are: This invention reduces extrapolation errors by scientifically selecting a linear velocity benchmark, accurately obtains the critical power point through small-sample step-by-step experiments, constructs a linear model using the least squares method to avoid discrete interpolation errors, achieves full linear velocity boundary prediction, introduces a dynamic safety redundancy strategy to generate operating windows and recommended working points, improves the tolerance to process fluctuations, ensures weld stability, continuously optimizes the model through a dynamic correction mechanism and a composite sampling scheme to adapt to production changes, and provides intuitive guidance through parameter visualization. Overall, it improves the accuracy, stability, and production adaptability of welding parameter selection, and reduces the risk of quality accidents. Attached Figure Description

[0014] Figure 1 This is a flowchart of a method for selecting the boundary of aluminum tube welding parameters according to the present invention. Detailed Implementation

[0015] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0016] Example: Refer to Figure 1 As shown: A method for selecting the boundary of welding parameters for aluminum tubes, specifically including the following steps: Step 1: Set a fixed linear speed benchmark: Based on actual production, select the most commonly used linear speed on the current production line as the benchmark, record the current environmental conditions, aluminum tube batch, and equipment status as the benchmark operating conditions for this calibration, conduct a stability pre-test, and confirm that the current operating conditions are stable. If defects or fluctuations occur, it is necessary to investigate equipment, material, or environmental problems, and start calibration again after the operating conditions are stable.

[0017] The selection of the linear speed baseline is based on production coverage and model robustness. The linear speed with the highest daily usage rate on the production line is selected. If the production line covers multiple linear speeds, the usage frequency and capacity share of each linear speed are calculated by statistically analyzing the production data of the past 30 days, and the speed with the highest frequency is selected. If the speeds are evenly distributed, the median value of the range is selected to reduce the extrapolation error of the model at high and low speeds. For trial production of new products or small-batch customized orders, the median value can be selected as a temporary baseline based on the linear speed range recommended in the process documents, and iterative optimization can be carried out after production data is accumulated.

[0018] The selection of a fixed linear velocity needs to take into account both representativeness and ease of operation. Choosing a commonly used linear velocity can reduce extrapolation errors in subsequent production. At the same time, a complete record of the baseline working condition provides a traceability basis for subsequent model correction. Using a fixed linear velocity as an anchor point, the two-dimensional power-linear velocity problem is simplified into a one-dimensional power problem, and the power boundary at that linear velocity is quickly located, providing an initial anchor point for subsequent linear modeling.

[0019] Step 2: Obtain the critical power point through small sample experiments: Select experimental samples and obtain the critical power for cold welding and overheating through small sample experiments under a fixed linear velocity reference.

[0020] The specific steps involved in obtaining the critical power point through a small-sample experiment are as follows: 21. Sample and Consumable Preparation: Cut aluminum tubes from the current batch, ensuring that there is no oxide layer damage, scratches or deformation on the surface, and wipe off the oil stains with anhydrous ethanol. To avoid the influence of end effect, discard the initial section during the experiment and only test the middle effective section. Calibrate the ultrasonic testing instrument and preset the experimental parameters. 22. Cold Welding Critical Power Test: At a fixed linear speed, set the initial power to 80% of the conventional power and start welding. After welding aluminum tubes at each power level, mark the power value and welding time on the sample for subsequent traceability. Observe whether there are grooves, pores, or poor formation caused by incomplete fusion on the weld surface. If the initial inspection fails, it is directly judged as cold welding. For samples that pass the appearance test, perform ultrasonic scanning using a transverse wave angle probe to perform 100% coverage inspection along both sides of the weld. If incomplete fusion defects are found, it is judged as cold welding. If a cold welding defect is detected at the current power, increase the power in steps of 0.1kW and repeat welding and inspection. When two consecutive samples at a certain power level have no cold welding defects, record that power as the cold welding critical power. ,exist Weld three more samples at the power level. If all of them pass the test, the value is confirmed to be stable. If fluctuations occur, the stability of the equipment or the consistency of the samples needs to be investigated. 23. Overheating Critical Power Test: Set the initial power to 120% of the normal power, start welding, and observe the weld surface for burn-through, collapse, or rough morphology caused by coarse grains. If the initial inspection fails, it is directly judged as overheating. For samples suspected of being overheated, take metallographic specimens, grind, polish, and etch them, and observe the grain size under a microscope. If the grain size in the weld area exceeds twice that of the base material, it is judged as overheating. If no overheating defect is detected at the current power, increase the power in increments of 0.1kW, and repeat welding and testing. When an overheating defect first appears at a certain power level, record the previous power level as the overheating critical power. ,exist Weld three more samples at the power level. If none of them show overheating defects, the value is confirmed to be stable. If fluctuations occur, the step size needs to be adjusted for more precise positioning. 24. Multi-linear velocity verification: Select two more different linear velocities and weld samples with different linear velocities onto the aluminum tube to obtain the corresponding linear velocity values. and .

[0021] During the experiment, other variables need to be kept constant, and a step-by-step approximation method is used to locate the critical value to avoid the boundary becoming blurred due to excessive step size. Discrete data of cold welding and overheating boundaries are obtained through at least 3 sets of experimental points to provide a basis for subsequent linear fitting.

[0022] Step 3: Establish linear models for cold welding boundary and overheating boundary: Preprocess the critical power data and construct linear equations for cold welding boundary and overheating boundary based on the least squares method.

[0023] Establishing a linear model for the cold weld boundary and the overheated boundary specifically includes the following steps: 31. Data cleaning: Perform data integrity checks, physical rationality verification, and outlier analysis on the acquired experimental data to ensure the purity of the modeling foundation; 32. Linear fitting of cold weld boundary and overheat boundary: Substitute the data into the linear equations of cold weld boundary and overheat boundary, solve the coefficients by least squares method, and obtain the cold weld boundary model and overheat boundary model. 33. Select intermediate linear velocities that were not involved in the fitting and substitute them into the model to calculate theoretical values ​​for experimental verification. Select points near the boundary for extrapolation verification, evaluate the applicable boundary of the model, and quantify the prediction error of the model.

[0024] The linear equation for the cold welding boundary is set as follows: The linear equation for the overheated boundary is: ; Substitute the cleaned cold welding data into the normal equation system: ; Substitute the superheat data after cleaning into the normal equations: ; Solving in sequence, we get , , and By obtaining the value of , the cold welding boundary model and the overheating boundary model can be obtained; in, Indicates the critical power for cold welding. The intercept term represents the boundary of the cold weld. The term represents the slope at the cold weld boundary, and v represents the welding line velocity. Indicates the superheat critical power. The intercept term representing the overheated boundary, The slope term represents the overheated boundary.

[0025] By transforming discrete experimental points into a continuous mathematical model, boundary prediction of the entire velocity range can be achieved, avoiding the errors of traditional methods that rely on discrete point interpolation.

[0026] Step 4: Generate a dynamic operating window: Substitute the real-time linear velocity into the model to calculate the theoretical critical power, and introduce a dynamic safety redundancy strategy to shrink the theoretical boundary inward to form the actual operating window. The center of the calculated window is used as the recommended operating point.

[0027] The dynamic operation window is generated by first reading the current line speed setting of the production line in real time. It automatically retrieves the parameters of the cold welding and overheating boundary models and calculates the theoretical critical power. and To prevent quality accidents caused by power grid fluctuations, instantaneous changes in gas source pressure, or material tolerances, an inward contraction strategy is adopted to obtain a margin on the cold welding side. and superheat side margin The actual operation window is , Calculate the golden midpoint within the operation window. .

[0028] The formula for calculating the midpoint of the Fibonacci retracement is as follows: ; The power is set at this default point to maximize the tolerance of the welding process to process fluctuations and ensure the most stable weld formation.

[0029] Step 5, Dynamic Correction and Continuous Learning: Incremental iteration of the model is achieved by introducing time weighting and sliding window mechanisms. Dynamic thresholds are set based on the goodness of fit of recent data. A hierarchical early warning system is established based on dynamic confidence intervals to adjust for minor process drifts and actively update boundary data.

[0030] Dynamic correction and continuous learning employ a composite scheme of timed basic sampling and triggered enhanced sampling. It maintains a frequency of sampling one set of welding samples at regular intervals for appearance inspection, ultrasonic non-destructive testing, and mechanical property sampling, ensuring the continuity of basic data. At the same time, it automatically increases the sampling frequency at key production nodes to cover parameter fluctuation risk points in the entire production scenario.

[0031] The time-weighted incremental iteration introduces a time decay weight coefficient to ensure that the model adapts to the current production conditions first. During weekly iterations, newly collected valid data are added to the original dataset according to weights, and the weighted least squares method is used to refit the cold welding boundary coefficient and the overheating boundary coefficient to generate the updated model equation. The full reconstruction trigger conditions for different scenarios are automatically triggered when key working conditions change, such as changing the aluminum tube material grade or the wall thickness deviation exceeding the threshold, major overhaul of welding equipment, or adjustment of process scheme. After two consecutive incremental iterations, the model prediction error still shows an upward trend, indicating that the original dataset can no longer cover the current working conditions.

[0032] Establish a three-tiered response mechanism: early warning, intervention, and verification. Level 1 warning: When the deviation of 3 consecutive sets of data is within 10%, it is judged as a slight drift, and the system will automatically fine-tune the model coefficients; Level 2 warning: When the deviation is 15%, it is judged as moderate drift. The system will pause automatic adjustment, lock the current operation window and complete a small sample validation experiment. The model will be manually updated based on the experimental results. Level 3 warning: When the deviation exceeds 15% or 5 sets of abnormal data appear consecutively, it is judged as a serious drift. The system will forcibly limit the power adjustment range and trigger the shutdown review process.

[0033] This invention first selects a commonly used linear velocity as a benchmark based on actual production conditions. After recording the benchmark operating conditions and confirming them through stability pre-experiments, the two-dimensional power-linear velocity problem is simplified into a one-dimensional power problem. Then, the critical power point is obtained through small-sample experiments. Qualified aluminum tube samples and calibration equipment are prepared first, and the critical power for cold welding and overheating is tested respectively using a step-approximation method. Two different linear velocities are then selected for supplementary verification. Subsequently, after cleaning the experimental data, a linear model of the cold welding and overheating boundaries is constructed based on the least squares method. The applicability of the model is evaluated through intermediate linear velocity verification and extrapolation verification. Then, the production line linear velocity is read in real time and substituted into the model to calculate the theoretical critical power. A dynamic safety redundancy strategy is introduced to shrink the theoretical boundary inward to form an actual operating window, and the golden midpoint of the window is calculated as the recommended operating point to ensure welding stability. At the same time, a time-weighted and sliding window mechanism is used to realize incremental iteration of the model. Data is collected through timed sampling and triggered enhanced sampling, and the model is dynamically corrected in combination with a graded early warning system.

[0034] Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for selecting a boundary of welding parameters for an aluminum tube, characterized by, Includes the following steps: S1. Set a fixed linear speed benchmark: Select the commonly used linear speed of the current production line as the benchmark, record the current environmental conditions, aluminum tube batch, and equipment status as the benchmark working condition for this calibration, conduct a stability pre-test, and confirm that the current working condition is stable. S2. Obtaining the critical power point through small sample experiments: Select experimental samples and obtain the critical power for cold welding and overheating through small sample experiments under a fixed linear velocity reference. S3. Establish linear models for cold welding boundary and overheating boundary: Preprocess the critical power data and construct linear equations for cold welding boundary and overheating boundary based on the least squares method. S4. Generate a dynamic operating window: Substitute the real-time linear velocity into the model to calculate the theoretical critical power, and introduce a dynamic safety redundancy strategy to shrink the theoretical boundary inward to form the actual operating window. The center of the calculated window is used as the recommended operating point. S5. Dynamic Correction and Continuous Learning: Incremental iteration of the model is achieved by introducing time weighting and sliding window mechanisms. A hierarchical early warning system is established based on dynamic confidence intervals to adjust for minor process drifts and update boundary data.

2. The method of claim 1, wherein: In step S1, the selection of the linear speed benchmark is based on production coverage and model robustness. The linear speed with the highest daily usage rate of the production line is selected. When the production line covers multiple linear speeds, the usage frequency and capacity ratio of each linear speed are calculated, and the speed is selected.

3. The method of claim 1, wherein: In step S2, obtaining the critical power point through a small-sample experiment specifically includes the following steps: S21. Sample and Consumable Preparation: Cut aluminum tubes from the current batch, remove oil stains, discard the initial section during the experiment, test the effective middle section, calibrate the ultrasonic testing instrument, and preset the experimental parameters. S22, cold welding critical power test: under the fixed linear speed, set the starting power to 80% of the conventional power, increase the power by 0.1 kW step by step, after welding the aluminum pipe at each power, judge whether there is an unfused defect through ultrasonic detection, and record the power value without cold welding defect for the first time as the cold welding critical power ; S23. Overheating Critical Power Test: Set the initial power to 120% of the normal power, and increase the power in increments of 0.1kW. After each power level, determine whether overheating defects occur by metallographic observation or surface morphology inspection. Record the power level preceding the first occurrence of an overheating defect as the overheating critical power. ; S24. Multi-linear velocity verification: Select 2-3 different linear velocities and weld samples with different linear velocities onto the aluminum tube to obtain the corresponding linear velocity values. and .

4. The method for selecting the boundary of aluminum tube welding parameters according to claim 1, characterized in that: In step S3, establishing the linear model of the cold weld boundary and the overheated boundary specifically includes the following steps: S31. Data cleaning: Perform data integrity checks, physical rationality verification, and outlier analysis on the experimental data obtained in step S2. S32. Linear fitting of cold weld boundary and overheat boundary: Substitute the data into the linear equations of cold weld boundary and overheat boundary, solve the coefficients by the least squares method, and obtain the cold weld boundary model and overheat boundary model. S33. Select intermediate linear velocities that were not involved in the fitting and substitute them into the model to calculate theoretical values ​​for experimental verification. Select points near the boundary for extrapolation verification, evaluate the applicable boundary of the model, and quantify the prediction error of the model.

5. The method for selecting the boundary of aluminum tube welding parameters according to claim 4, characterized in that: In step S32, the linear equation of the cold welding boundary is set as follows: The linear equation for the overheated boundary is: ; Substitute the cleaned cold welding data into the normal equation system: ； Substitute the superheat data after cleaning into the normal equations: ； Solving in sequence, we get , , and By obtaining the value of , the cold welding boundary model and the overheating boundary model can be obtained; in, Indicates the critical power for cold welding. The intercept term representing the cold weld boundary, The term represents the slope at the cold weld boundary, and v represents the welding line velocity. Indicates the superheat critical power. The intercept term represents the overheated boundary. The slope term represents the overheated boundary.

6. The method for selecting the boundary of aluminum tube welding parameters according to claim 1, characterized in that: In step S4, the dynamic operation window first reads the current line speed setpoint of the production line in real time. It automatically retrieves the parameters of the cold welding and overheating boundary models and calculates the theoretical critical power. and An inward contraction strategy was adopted to obtain the allowance on the cold welding side. and superheat side margin The actual operation window is , Calculate the golden midpoint within the operation window. .

7. The method for selecting the boundary of aluminum tube welding parameters according to claim 1, characterized in that: The formula for calculating the midpoint of the Fibonacci retracement is as follows: 。 8. The method for selecting the boundary of aluminum tube welding parameters according to claim 1, characterized in that: In step S5, dynamic correction and continuous learning adopt a composite scheme of timed basic sampling and triggered enhanced sampling to maintain the frequency of sampling one set of welding samples at regular intervals for appearance inspection, ultrasonic non-destructive testing and mechanical property sampling, and increase the sampling frequency at key production nodes.

Images