Aluminum alloy electric welding quality evaluation method and system

Abstract

The invention provides an aluminum alloy electric welding quality evaluation method and system. The method comprises the steps that first welding data and second welding data of aluminum alloy in the two spot welding processes of dynamic electrode pressure and welding current are obtained respectively; the first welding data and the second welding data are preprocessed, dynamic fluctuation characteristics, stability characteristics, extreme event characteristics and energy accumulation characteristics are extracted from the preprocessed first welding data, a current steady state effective value is extracted from the preprocessed second welding data, and heat is separated out; performing correlation analysis on all the extracted features by using a Pearson's correlation coefficient, and obtaining multiple target features according to a correlation analysis result; and the multiple target features are identified based on the classification prediction model of the random forest, a final classification result is obtained, and the final classification result is one of pseudo soldering, normality and cracks. According to the invention, the aluminum alloy electric welding quality can be efficiently detected in real time.

Description

Technical Field

[0001] This invention relates to the field of electric welding quality assessment technology, and in particular to a method and system for assessing the quality of aluminum alloy electric welding. Background Technology

[0002] With the deepening of Industry 4.0, modern industry has increasingly higher requirements for welding quality. The requirements for welding quality in smart factories have been upgraded from the traditional strength standard to full-process traceability and digital control. The innovative research and development of resistance welding process monitoring and quality prediction technology has become a key breakthrough in improving manufacturing quality and has gradually become the focus of scholars' research.

[0003] Resistance spot welding is a typical multi-physics coupled manufacturing process involving complex interactions of electrical conduction, mechanical stress, Joule heating effect, metal phase transformation, and molten metal flow, resulting in a highly complex physical process. Within the millisecond-level welding cycle, key process parameters such as current intensity, preset electrode pressure, and energizing time exhibit strong nonlinear coupling characteristics. These are compounded by random disturbances such as electrode surface oxidation, plate gap fluctuations, and cooling water temperature changes, leading to significant uncertainties in the welding energy input and weld nugget growth process. The molten pool formation process is confined between the plates and difficult to observe directly. Furthermore, the duration of the liquid metal phase transformation is extremely short; even minute current fluctuations or pressure deviations can cause the weld nugget size to deviate from the critical value, resulting in hidden defects such as spatter, incomplete welds, or interface detachment. This is especially true for aluminum and aluminum alloys, which possess excellent electrical and thermal conductivity.

[0004] In summary, the quality of aluminum alloy resistance spot welding is strongly influenced by multiple parameters. Traditional destructive testing methods are inefficient and cannot achieve real-time monitoring, so there is an urgent need for an efficient non-destructive evaluation method. Summary of the Invention

[0005] The purpose of this invention is to provide a method and system for evaluating the quality of aluminum alloy electric welding, which aims to solve the problems of low efficiency and inability to achieve real-time monitoring in traditional detection technologies.

[0006] In a first aspect, the present invention provides a method for evaluating the quality of aluminum alloy electric welding, characterized in that the method includes: First and second welding data of aluminum alloy were obtained in two spot welding processes under dynamic electrode pressure and welding current, respectively. The first welding data and the second welding data are preprocessed, and dynamic fluctuation characteristics, stability characteristics, extreme event characteristics and energy accumulation characteristics are extracted from the preprocessed first welding data. The steady-state effective value of current and heat release are extracted from the preprocessed second welding data. The Pearson correlation coefficient was used to perform correlation analysis on all extracted features, and multiple target features were obtained based on the correlation analysis results. The classification prediction model based on random forest is used to identify the multiple target features to obtain the final classification result, which is one of the following: poor weld, normal, or crack.

[0007] In some embodiments, the preprocessing of the first welding data and the second welding data includes: Fourier transform is performed on the first and second welding data to convert the time-domain signal into a frequency-domain signal and analyze the frequency components of the signal. Set the corresponding frequency threshold according to the signal characteristics to identify and filter out high-frequency noise components; The filtered frequency domain signal is subjected to an inverse Fourier transform to convert it back to the time domain signal, resulting in a smoothed signal after noise reduction.

[0008] In some embodiments, extracting dynamic fluctuation characteristics, stability characteristics, extreme event characteristics, and energy accumulation characteristics from the preprocessed first welding data includes: The first welding data is a dynamic electrode force curve, which includes multiple consecutive moments and the electrode force corresponding to each moment. Each feature includes at least one indicator, wherein the dynamic fluctuation feature includes peak-to-peak value, root mean square value, and slope; the stability feature includes variance, standard deviation, and interquartile range; the extreme event feature includes maximum value, kurtosis, and negative inflection point; and the energy accumulation feature includes absolute energy and area under the curve.

[0009] In some embodiments, extracting the steady-state effective value of the current and the heat generated from the preprocessed second welding data includes: The second welding data is a current curve, which includes multiple consecutive time points and the current corresponding to each time point; The steady-state RMS value of the current can be obtained using the following formula: ; in, Let k be the moving average at time k. The number of sampling points selected during the steady-state phase of the current. Let i be the current value at the i-th sampling point; In some embodiments, the heat of release is obtained according to the following formula: ; in, In order to release heat, For the power-on time, This is a dynamic resistance, which is constant. Let t be the current value at time t.

[0010] In some embodiments, the step of using the Pearson correlation coefficient to perform correlation analysis on all extracted features and obtaining multiple target features based on the correlation analysis results includes: Based on the results of the correlation analysis, features exhibiting multicollinearity were screened. Specifically, only one of the multiple strongly correlated indicators reflecting the same feature was retained, and only one of the multiple strongly correlated indicators exhibiting complementary characteristics was retained. A strong correlation index is defined as a Pearson correlation coefficient between two indicators that is greater than the first threshold. The Pearson correlation coefficient between the two indicators is calculated using the following formula: ; in, The Pearson correlation coefficient between indicators x and y. and The first and second indices are respectively the first and second indices x and y. i One observation value, and are the mean values ​​of indicators x and y, respectively, and n is the total number of indicators.

[0011] In some embodiments, the random forest-based classification prediction model identifies the multiple target features to obtain a final classification result; including: The multiple target features are connected to obtain the features to be identified, and the features to be identified are input into a classification prediction model based on random forest to obtain the final identification result. For a new input feature to be identified, each decision tree outputs a class prediction value based on its feature path. The final classification result is obtained by the mode of the votes from each base classifier: ; in, For the final classification result, For the set of all possible category labels, This is an indicator function that returns 1 if the condition is true, and 0 otherwise. c is the category label.

[0012] Secondly, the present invention provides an aluminum alloy electric welding quality assessment system, the system comprising: The data acquisition module is used to acquire the first welding data and the second welding data of aluminum alloy in two spot welding processes: dynamic electrode pressure and welding current. The feature extraction module is used to preprocess the first welding data and the second welding data, and extract dynamic fluctuation features, stability features, extreme event features and energy accumulation features from the preprocessed first welding data, and extract the steady-state effective value of current and heat release from the preprocessed second welding data. The feature analysis module is used to perform correlation analysis on all extracted features using the Pearson correlation coefficient, and to obtain multiple target features based on the correlation analysis results. The welding prediction module is used to identify the multiple target features based on a random forest classification prediction model to obtain a final classification result, which is one of the following: poor weld, normal weld, or crack.

[0013] Thirdly, the present invention provides a storage medium that stores one or more programs that, when executed by a processor, implement the above-described aluminum alloy electric welding quality assessment method.

[0014] Fourthly, the present invention provides an electronic device, the electronic device comprising a memory and a processor, wherein: The memory is used to store computer programs; When the processor executes the computer program stored in the memory, it implements the above-mentioned aluminum alloy electric welding quality assessment method.

[0015] Compared with the prior art, the present invention has the following advantages: First, this invention acquires first and second welding data for aluminum alloys during two spot welding processes involving dynamic electrode pressure and welding current, providing diverse data sources for subsequent analysis and enabling a more comprehensive reflection of various states during the welding process. Since high-frequency noise and other interferences exist during the welding process, affecting data accuracy, this invention performs preprocessing operations on the acquired data, including Fourier transform, filtering out high-frequency noise components based on frequency thresholds, and inverse Fourier transform. This effectively removes noise interference, obtains smooth signals, and improves data quality. Simultaneously, it extracts various features from the preprocessed data, such as dynamic fluctuation characteristics, stability characteristics, extreme event characteristics, energy accumulation characteristics, steady-state effective value of current, and heat release. These features describe key information about the welding process from different perspectives. Furthermore, considering that the extracted features may suffer from multicollinearity, which increases model complexity and computational load and reduces evaluation accuracy, this invention addresses these issues. This invention utilizes Pearson correlation coefficients to perform correlation analysis on all features, filtering out multiple target features and removing redundant information, enabling the model to more efficiently utilize key features for evaluation. Finally, addressing the uncertainty of aluminum alloy resistance spot welding quality and the difficulty in detecting hidden defects, this invention uses a random forest classification and prediction model to identify multiple target features. The random forest algorithm can handle high-dimensional data and, through a voting mechanism of multiple decision trees, improves the accuracy and stability of classification results, accurately identifying different welding quality states such as incomplete welds, normal welds, and cracks. Attached Figure Description

[0016] Figure 1This is a flowchart of an aluminum alloy electric welding quality assessment method proposed in an embodiment of the present invention; Figure 2 The dynamic electrode force curve before noise reduction and smoothing; Figure 3 The dynamic electrode force curve before noise reduction and smoothing; Figure 4 The current curve before noise reduction and smoothing; Figure 5 The current curve after noise reduction and smoothing; Figure 6 A graph showing the correlation coefficient between the feature and the diameter of the melt core; Figure 7 This is a correlation coefficient graph after removing some features; Figure 8 This is a schematic diagram of data sampling for the Random Forest algorithm; Figure 9 Here is a flowchart of the Random Forest algorithm training process; Figure 10 The following is an example of the final classification results, where (a) is a cold solder joint, (b) is a normal solder joint, and (c) is a cracked solder joint. Figure 11 This is a schematic diagram of the structure of an aluminum alloy electric welding quality assessment system proposed in an embodiment of the present invention.

[0017] The following detailed description, in conjunction with the accompanying drawings, will further illustrate the present invention. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. Unless otherwise defined, the technical or scientific terms used herein should have the ordinary meaning understood by those skilled in the art. The terms "comprising" and similar expressions used herein mean that the element or object preceding the word covers the element or object listed after the word and its equivalents, but does not exclude other elements or objects.

[0019] First, the applicant discovered that existing research largely focuses on signals such as dynamic resistance and electrode displacement, but these are easily affected by oxide films and thermal conductivity in aluminum alloy spot welding, resulting in insufficient feature discrimination. Dynamic electrode force signals can directly reflect the thermal expansion effect during weld nugget growth and are less affected by equipment type, exhibiting higher robustness. Therefore, this invention uses dynamic electrode force as the core monitoring signal, combines it with current characteristics to construct a multi-dimensional feature vector, and builds an offline resistance spot welding quality prediction model.

[0020] Specifically, such as Figure 1 As shown, this is a method for evaluating the quality of aluminum alloy electric welding according to an embodiment of the present invention. The method includes steps S101 to S104, wherein: Step S101: Obtain the first welding data and the second welding data of aluminum alloy in the two spot welding processes of dynamic electrode pressure and welding current respectively; It should be noted that, in this embodiment, the first welding data is a dynamic electrode force curve, which includes multiple consecutive moments and the electrode force corresponding to each moment; the second welding data is a current curve, which includes multiple consecutive moments and the current corresponding to each moment.

[0021] Step S102: Preprocess the first welding data and the second welding data, and extract dynamic fluctuation characteristics, stability characteristics, extreme event characteristics and energy accumulation characteristics from the preprocessed first welding data, and extract the steady-state effective value of current and heat release from the preprocessed second welding data; After synchronously acquiring the parameter curves of dynamic electrode force and welding current during the welding experiment, it was found that the actual acquired welding current and inter-electrode voltage signals inevitably contain various noise and interference components. Directly applying these signals to subsequent feature extraction would lead to significant errors, thus affecting the accuracy of the quality assessment model. Therefore, it is necessary to preprocess the original acquired signals to eliminate noise interference and other adverse factors. This embodiment uses Fourier transform to remove high-frequency noise for signal denoising. The basic principle and steps are as follows: First, Fourier transform is performed on the acquired original welding current and inter-electrode voltage signals to convert the time-domain signal into a frequency-domain signal, and the frequency components of the signal are analyzed. Second, an appropriate frequency threshold is set according to the signal characteristics to identify and filter out high-frequency noise components. Next, inverse Fourier transform is performed on the filtered frequency-domain signal to convert it back to a time-domain signal. Finally, a smoothed signal after denoising is obtained for subsequent analysis and calculation. Through the frequency-domain filtering operation of Fourier transform, high-frequency noise in the signal can be effectively separated and removed while retaining the main characteristics of the signal. The dynamic electrode force curve and current curve before and after denoising are shown below. Figures 2 to 5 As shown.

[0022] In addition, in some embodiments, the dynamic electrode force is related to the melting of the core metal. In order to capture key information reflecting the welding quality from the dynamic electrode force signal, four types of core features were extracted from the dynamic electrode force curve: dynamic fluctuation features, stability features, extreme event features and energy accumulation features. A total of 11 representative features (indicators) were selected, as shown in Table 1 below, which can effectively characterize the time-varying characteristics and abnormal modes of the electrode force.

[0023] Table 1 Feature Extraction and Significance Analysis of Dynamic Electrode Force Signals

[0024] Furthermore, in some embodiments, current is the source of welding energy, and therefore the most significant factor affecting welding quality. The medium-frequency resistance welding control system features a high-precision current detection module and a closed-loop feedback adjustment mechanism, enabling effective dynamic compensation during the welding process. Repeatability test data shows that when the real-time current value reaches the set threshold, the current fluctuation amplitude in the steady-state phase is very small. Therefore, when extracting features from the current signal, stability-related features are not given excessive consideration.

[0025] The steady-state effective value of the current is calculated using the sliding window averaging method. N sampling points are selected during the steady-state phase of the current, which represents the width of the sliding window. This is the moving average value at time k, which is the steady-state effective value of the current. for: ; in, Let k be the moving average at time k. The number of sampling points selected during the steady-state phase of the current. Let be the current value at the i-th sampling point.

[0026] When welding current passes through the resistance of the welding zone, heat is generated, causing the temperature field to rise and forming a weld joint. According to Joule's law, the heat generated during the entire energizing time tw... for: ; During the resistance spot welding process of aluminum alloys, dynamic resistance It remains almost unchanged. Under the same welding conditions, the dynamic resistance can be approximated as a constant r, therefore the heat expression can be transformed into: ; in, In order to release heat, For the power-on time, This is a dynamic resistance, which is constant. Let t be the current value at time t.

[0027] Step S103: Use Pearson correlation coefficient to perform correlation analysis on all extracted features, and obtain multiple target features based on the correlation analysis results; It should be noted that in some embodiments, features exhibiting multicollinearity are first screened based on the results of correlation analysis. Specifically, only one of the multiple strongly correlated indicators that reflect the same feature is retained, and only one of the multiple strongly correlated indicators that have complementary representations is retained. A strong correlation index is defined as a Pearson correlation coefficient between two indicators that is greater than the first threshold. The Pearson correlation coefficient between the two indicators is calculated using the following formula: ; in, The Pearson correlation coefficient between indicators x and y. and The first and second indices are respectively the first and second indices x and y. i One observation value, and are the mean values ​​of indicators x and y, respectively, and n is the total number of indicators.

[0028] Furthermore, the interrelationships between features can be assessed using feature correlation analysis. Correlation describes the association between input features and between features and outputs. In statistics, positive correlation indicates that two measured indicators show a synchronous growth trend, while negative correlation indicates that when one indicator increases, the other decreases. This association characteristic includes both the direction of covariation (positive or negative) and the strength of the linkage (the magnitude of the correlation coefficient). The correlation coefficient ranges from -1 to 1. The closer the absolute value of the correlation coefficient is to 1, the stronger the correlation between the variables; while an absolute value close to 0 indicates a weak or non-existent correlation between the variables.

[0029] Furthermore, to visually demonstrate the correlations between various features, this study employed a correlation heatmap for visualization analysis. The heatmap uses color intensity to represent the magnitude of the correlation coefficient, thus clearly reflecting the association between features. In the heatmap, a redder color indicates a stronger positive correlation, while a bluer color indicates a stronger negative correlation. Features with no significant correlation are displayed in a neutral color.

[0030] like Figure 6The figure shows a thermogram illustrating the correlation between features extracted from dynamic electrode force and current curves and the melt core diameter. Multicollinearity analysis of the features reveals that: the interquartile range (IQR) is highly correlated with the standard deviation (Std), variance, and peak-to-peak value, indicating that all four reflect the fluctuation characteristics of the electrode force and exhibit multicollinearity; therefore, only the standard deviation is retained for subsequent modeling. The absolute energy is strongly correlated with the area under the curve (AUC), which can be considered a complementary representation of energy accumulation; therefore, only the absolute energy is retained for subsequent modeling.

[0031] After removing the four features—interquartile range (IQR), variance, peak-to-peak value, and area under the curve (AUC)—the correlation coefficient is calculated again, as follows: Figure 7 As shown, some features are strongly positively correlated with the weld nugget diameter, such as the steady-state effective value of the current and the absolute energy; others are moderately positively correlated, such as Joule heat (Q), root mean square (RMS) value, and negative TP. A low correlation coefficient does not mean that the feature has no impact on the welding result. In actual welding, nonlinear effects may be hidden within features with low linear correlation: for example, certain features need to reach a specific range to significantly affect the weld nugget, and combinations of features may produce synergistic effects. Traditional linear models struggle to analyze these nonlinear relationships. Therefore, in the subsequent modeling stage, all features are retained as input data, and nonlinear algorithms are used to explore the mapping relationship between the extracted features and the welding results.

[0032] Step S104: The multi-target features are identified using a random forest-based classification prediction model to obtain a final classification result, which is one of the following: cold weld, normal, or crack.

[0033] It should be noted that Random Forest is an ensemble classifier composed of a series of decision tree classifiers. Its core idea is to construct multiple independent and differentiated decision trees within the Bagging framework, utilizing a collective voting mechanism to improve the model's generalization ability. This algorithm overcomes the overfitting defect of a single decision tree model through two key mechanisms: (1) Bootstrap sampling randomness: A sampling method with replacement is used to generate k distinct subsets from the original training set, such as... Figure 8 Each subset is approximately 63.2% the size of the original data. The unsampled data (i.e., out-of-bag data) can be used to estimate model error.

[0034] (2) Feature space randomness: During the node splitting process of each decision tree, only k candidate features are allowed to be randomly selected from all m features (usually k= This encourages the diversification of base learners by limiting the feature search space.

[0035] ntree differential decision trees, i.e., ntree base classifiers, are generated using the aforementioned dual randomness. Each tree is not pruned until a preset termination condition is met. For a new input sample x, each decision tree outputs a class prediction value Ci(x) based on its feature path. The final classification result is determined by the mode of the votes from each base classifier. ; in, For the final classification result, For the set of all possible category labels, This is an indicator function that returns 1 if the condition is true, and 0 otherwise. c is the category label.

[0036] In summary, this invention first acquires first and second welding data for aluminum alloys during two spot welding processes involving dynamic electrode pressure and welding current, providing diverse data sources for subsequent analysis and enabling a more comprehensive reflection of various states during the welding process. Since high-frequency noise and other interferences exist during the welding process, affecting data accuracy, this invention performs preprocessing operations on the acquired data, including Fourier transform, filtering out high-frequency noise components based on frequency thresholds, and inverse Fourier transform. This effectively removes noise interference, obtains smooth signals, and improves data quality. Simultaneously, it extracts various features from the preprocessed data, such as dynamic fluctuation characteristics, stability characteristics, extreme event characteristics, energy accumulation characteristics, steady-state effective value of current, and heat release. These features describe key information about the welding process from different perspectives. Furthermore, considering that the extracted features may suffer from multicollinearity, which increases model complexity and computational load and reduces evaluation accuracy, this invention addresses these issues. This invention utilizes Pearson correlation coefficients to perform correlation analysis on all features, filtering out multiple target features and removing redundant information, enabling the model to more efficiently utilize key features for evaluation. Finally, addressing the uncertainty of aluminum alloy resistance spot welding quality and the difficulty in detecting hidden defects, this invention uses a random forest classification and prediction model to identify multiple target features. The random forest algorithm can handle high-dimensional data and, through a voting mechanism of multiple decision trees, improves the accuracy and stability of classification results, accurately identifying different welding quality states such as incomplete welds, normal welds, and cracks.

[0037] like Figure 11 As shown, an embodiment of the present invention also proposes an aluminum alloy electric welding quality assessment system, the system comprising: Data acquisition module 10 is used to acquire the first welding data and the second welding data of aluminum alloy in two spot welding processes of dynamic electrode pressure and welding current, respectively; Feature extraction module 20 is used to preprocess the first welding data and the second welding data, and extract dynamic fluctuation features, stability features, extreme event features and energy accumulation features from the preprocessed first welding data, and extract the steady-state effective value of current and heat release from the preprocessed second welding data. The feature analysis module 30 is used to perform correlation analysis on all extracted features using the Pearson correlation coefficient, and to obtain multiple target features based on the correlation analysis results. The welding prediction module 40 is used to identify the multiple target features based on a random forest classification prediction model to obtain a final classification result, which is one of the following: cold weld, normal, or crack.

[0038] In another aspect, the present invention also proposes a storage medium having stored one or more programs thereon, which, when executed by a processor, implement the above-described method for evaluating the quality of aluminum alloy welding.

[0039] In another aspect, the present invention also proposes an electronic device, including a memory and a processor, wherein the memory is used to store a computer program, and the processor is used to execute the computer program stored in the memory to implement the above-mentioned aluminum alloy electric welding quality assessment method.

[0040] Those skilled in the art will understand that the logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can mean any means that can contain stored, communicated, propagated, or transmitted programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0041] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0042] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0043] While embodiments of the present invention have been described in detail above, it will be apparent to those skilled in the art that various modifications and variations can be made to these embodiments. However, it should be understood that such modifications and variations fall within the scope and spirit of the invention as set forth in the claims. Furthermore, the invention described herein may have other embodiments and can be implemented or carried out in various ways.

Claims

1. A method for evaluating the quality of aluminum alloy electric welding, characterized in that, The method includes: First and second welding data of aluminum alloy were obtained in two spot welding processes under dynamic electrode pressure and welding current, respectively. The first welding data and the second welding data are preprocessed, and dynamic fluctuation characteristics, stability characteristics, extreme event characteristics and energy accumulation characteristics are extracted from the preprocessed first welding data. The steady-state effective value of current and heat release are extracted from the preprocessed second welding data. The Pearson correlation coefficient was used to perform correlation analysis on all extracted features, and multiple target features were obtained based on the correlation analysis results. The classification prediction model based on random forest is used to identify the multiple target features to obtain the final classification result, which is one of the following: poor weld, normal, or crack.

2. The method for evaluating the quality of aluminum alloy electric welding according to claim 1, characterized in that, The first welding data and the second welding data are preprocessed; include: Fourier transform is performed on the first and second welding data to convert the time-domain signal into a frequency-domain signal and analyze the frequency components of the signal. Set the corresponding frequency threshold according to the signal characteristics to identify and filter out high-frequency noise components; The filtered frequency domain signal is subjected to an inverse Fourier transform to convert it back to the time domain signal, resulting in a smoothed signal after noise reduction.

3. The method for evaluating the quality of aluminum alloy electric welding according to claim 1, characterized in that, The extraction of dynamic fluctuation characteristics, stability characteristics, extreme event characteristics, and energy accumulation characteristics from the preprocessed first welding data includes: The first welding data is a dynamic electrode force curve, which includes multiple consecutive moments and the electrode force corresponding to each moment. Each feature includes at least one indicator, wherein the dynamic fluctuation feature includes peak-to-peak value, root mean square value, and slope; the stability feature includes variance, standard deviation, and interquartile range; the extreme event feature includes maximum value, kurtosis, and negative inflection point; and the energy accumulation feature includes absolute energy and area under the curve.

4. The method for evaluating the quality of aluminum alloy electric welding according to claim 3, characterized in that, The step of extracting the steady-state effective value of the current and the heat generated from the pre-processed second welding data includes: The second welding data is a current curve, which includes multiple consecutive time points and the current corresponding to each time point; The steady-state RMS value of the current can be obtained using the following formula: ； in, Let k be the moving average at time k. The number of sampling points selected during the steady-state phase of the current. Let be the current value at the i-th sampling point.

5. The method for evaluating the quality of aluminum alloy electric welding according to claim 4, characterized in that, The heat of release can be obtained using the following formula: ； in, In order to release heat, For the power-on time, This is a dynamic resistance, which is constant. Let t be the current value at time t.

6. The method for evaluating the quality of aluminum alloy electric welding according to claim 5, characterized in that, The process involves using Pearson correlation coefficient to perform correlation analysis on all extracted features, and obtaining multiple target features based on the correlation analysis results; including: Based on the results of the correlation analysis, features exhibiting multicollinearity were screened. Specifically, only one of the multiple strongly correlated indicators reflecting the same feature was retained, and only one of the multiple strongly correlated indicators exhibiting complementary characteristics was retained. A strong correlation index is defined as a Pearson correlation coefficient between two indicators that is greater than the first threshold. The Pearson correlation coefficient between the two indicators is calculated using the following formula: ； in, The Pearson correlation coefficient between indicators x and y. and The first and second indices are respectively the first and second indices x and y. i One observation value, and are the mean values ​​of indicators x and y, respectively, and n is the total number of indicators.

7. The method for evaluating the quality of aluminum alloy electric welding according to claim 6, characterized in that, The random forest-based classification prediction model identifies the multiple target features to obtain the final classification result; including: The multiple target features are connected to obtain the features to be identified, and the features to be identified are input into a classification prediction model based on random forest to obtain the final identification result. For a new input feature to be identified, each decision tree outputs a class prediction value based on its feature path. The final classification result is obtained by the mode of the votes from each base classifier: ； in, For the final classification result, For the set of all possible category labels, This is an indicator function that returns 1 if the condition is true, and 0 otherwise. c is the category label.

8. A quality assessment system for aluminum alloy electric welding, characterized in that, The system includes: The data acquisition module is used to acquire the first welding data and the second welding data of aluminum alloy in two spot welding processes: dynamic electrode pressure and welding current. The feature extraction module is used to preprocess the first welding data and the second welding data, and extract dynamic fluctuation features, stability features, extreme event features and energy accumulation features from the preprocessed first welding data, and extract the steady-state effective value of current and heat release from the preprocessed second welding data. The feature analysis module is used to perform correlation analysis on all extracted features using the Pearson correlation coefficient, and to obtain multiple target features based on the correlation analysis results. The welding prediction module is used to identify the multiple target features based on a random forest classification prediction model to obtain a final classification result, which is one of the following: poor weld, normal weld, or crack.

9. A storage medium storing one or more programs that, when executed by a processor, implement the aluminum alloy electric welding quality assessment method as claimed in any one of claims 1-7.

10. An electronic device comprising a memory and a processor, wherein: The memory is used to store computer programs; When the processor is used to execute the computer program stored in the memory, it implements the aluminum alloy electric welding quality evaluation method as described in any one of claims 1-7.

Images