Annular light spot laser welding quality prediction method, system and equipment and storage medium

Abstract

The invention relates to the technical field of laser welding, and discloses an annular light spot laser welding quality prediction method, system and equipment and a storage medium. The method comprises the following steps: constructing a training set and a test set according to historical process parameters and historical welding seam quality results of annular light spot laser welding; constructing a regression model based on a multi-layer sensor; training a multi-layer perceptron-based regression model according to the training set to obtain a trained multi-layer perceptron-based regression model; verifying the trained regression model based on the multi-layer perceptron based on the test set; and the current technological parameters are input into the trained regression model based on the multi-layer sensor, and a current welding quality predicted value is obtained. According to the method, the multi-layer sensor with the nonlinear modeling capacity is adopted, the complex coupling relation between the laser welding process parameters and the welding quality indexes is modeled, the model is used for annular light spot laser welding quality prediction, the prediction precision is improved, and effective evaluation on the uncertainty of the prediction result is provided.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of laser welding, in particular to a ring-shaped light spot laser welding quality prediction method, system, device and storage medium. BACKGROUND

[0002] As an advanced welding technology with high efficiency and high energy density, laser welding has been widely used in manufacturing fields such as automobiles, aerospace, rail transportation, shipbuilding and precision structural parts processing. Compared with traditional welding methods, laser welding has the remarkable advantages of small heat input, narrow weld, fast welding speed, small deformation and high automation level. In recent years, with the continuous improvement of laser power level and the development of laser spot regulation technology, ring-shaped light spot laser welding has gradually become one of the key process means for connecting and processing of structural steel, high-strength steel and other medium-thick plate materials.

[0003] In ring-shaped light spot laser welding, welding quality is affected by the coupling of various process parameters, such as outer ring power, inner ring (center) power, welding speed, defocusing amount, etc. Reasonably setting the process parameter combination is of great significance to achieve deep penetration welding, high-strength joints and good forming. However, the selection of process parameters in current industrial production process mainly relies on empirical rules or a small amount of offline tests, which lacks systematicness and real-time performance, and it is difficult to quickly respond to welding quality control requirements under different working conditions. SUMMARY

[0004] The main purpose of the present application is to provide a ring-shaped light spot laser welding quality prediction method, system, device and storage medium, which aims to solve at least one of the above technical problems.

[0005] To achieve the above-mentioned purpose, the present application provides a ring-shaped light spot laser welding quality prediction method, comprising: constructing a training set and a test set according to historical process parameters and historical weld quality results of ring-shaped light spot laser welding; constructing a regression model based on a multilayer perceptron; training the regression model based on the multilayer perceptron according to the training set to obtain a trained regression model based on the multilayer perceptron; verifying the trained regression model based on the multilayer perceptron based on the test set; obtaining current process parameters, inputting the current process parameters into the trained regression model based on the multilayer perceptron, and obtaining a current welding quality prediction value.

[0006] In some embodiments, the training set and the test set are constructed according to the historical process parameters and the historical weld quality results of the ring-shaped light spot laser welding, comprising: obtaining historical process parameters and corresponding historical weld quality results of ring-shaped light spot laser welding; constructing a sample data set according to the historical process parameters and the historical weld quality results; performing structuralization and standardization processing on the sample data set to obtain a processed data set; dividing the processed data set into a training set and a test set.

[0007] In some embodiments, the structuralization and standardization processing on the sample data set to obtain a processed data set comprises: constructing five-tuples of three process parameters and two welding quality indicators in the sample data set; performing de-duplication, missing value filling and outlier removal processing on the five-tuples to obtain processed five-tuples; unifying the processed five-tuples into a two-dimensional structured table format based on the physical magnitude and dimension standard of the process parameters to obtain a penetration and width data set; performing standardization processing on the penetration and width data set to obtain a processed data set.

[0008] In some embodiments, the constructing a multi-layer perceptron-based regression model comprises: constructing a multi-layer perceptron-based penetration regression model; constructing a multi-layer perceptron-based width regression model; taking the multi-layer perceptron-based penetration regression model and the multi-layer perceptron-based width regression model as a multi-layer perceptron-based regression model; wherein the multi-layer perceptron-based penetration regression model and the multi-layer perceptron-based width regression model are independent of each other and have the same regression model structure.

[0009] In some embodiments, the multi-layer perceptron-based penetration regression model comprises an input layer, a plurality of hidden layers and an output layer; wherein the input layer comprises three nodes corresponding to three process parameters respectively, the hidden layers adopt an activation function, and the output layer is a single node; the output layer of the multi-layer perceptron-based penetration regression model outputs a standardized predicted value of penetration; the output layer of the multi-layer perceptron-based width regression model outputs a standardized predicted value of width.

[0010] In some embodiments, the training the multi-layer perceptron-based regression model according to the training set to obtain a trained multi-layer perceptron-based regression model comprises: performing round-by-round iterative training on the multi-layer perceptron-based penetration regression model based on the training set and a supervised target to obtain a trained penetration regression model; training the multi-layer perceptron-based regression model based on the training set and a supervised target, to obtain a trained fusion width regression model; generating a trained multi-layer perceptron-based regression model based on the trained fusion depth regression model and the trained fusion width regression model.

[0011] In some embodiments, the verifying the trained multi-layer perceptron-based regression model based on the test set comprises: obtaining a predicted value output by the trained multi-layer perceptron-based regression model based on the test set; calculating a residual of the predicted value and an actual value; calculating a mean and a standard deviation of the residual according to a distribution of the residual; constructing a confidence interval according to the mean and the standard deviation; verifying the trained multi-layer perceptron-based regression model based on the confidence interval.

[0012] In addition, to achieve the above object, the application further provides a ring-shaped light spot laser welding quality prediction system, comprising: a data acquisition module configured to construct a training set and a test set based on historical process parameters and historical weld quality results of ring-shaped light spot laser welding; a model construction module configured to construct a multi-layer perceptron-based regression model; a model training module configured to train the multi-layer perceptron-based regression model based on the training set, to obtain a trained multi-layer perceptron-based regression model; a model verification module configured to verify the trained multi-layer perceptron-based regression model based on the test set; a model prediction module configured to obtain current process parameters, input the current process parameters into the trained multi-layer perceptron-based regression model, and obtain a current weld quality prediction value.

[0013] In addition, to achieve the above object, the application further provides an electronic device, which comprises a memory, a processor, and a ring-shaped light spot laser welding quality prediction program stored in the memory and executable on the processor, and the ring-shaped light spot laser welding quality prediction program is configured to implement the ring-shaped light spot laser welding quality prediction method as described above.

[0014] In addition, to achieve the above object, the application further provides a storage medium storing a ring-shaped light spot laser welding quality prediction program, and the ring-shaped light spot laser welding quality prediction program is used to enable a processor to implement the ring-shaped light spot laser welding quality prediction method as described above.

[0015] The application provides a ring-shaped light spot laser welding quality prediction method, comprising the following steps: constructing a training set and a test set according to historical process parameters and historical weld quality results of ring-shaped light spot laser welding; constructing a multilayer perceptron-based regression model; training the multilayer perceptron-based regression model according to the training set to obtain a trained multilayer perceptron-based regression model; verifying the trained multilayer perceptron-based regression model based on the test set; obtaining current process parameters, inputting the current process parameters into the trained multilayer perceptron-based regression model, and obtaining a current weld quality prediction value. In the application, the multilayer perceptron with nonlinear modeling capability is used to model the complex coupling relationship between various process parameters and weld quality indexes in laser welding, a trained multilayer perceptron-based regression model is obtained, the ring-shaped light spot laser welding quality prediction is performed, and the prediction accuracy is improved. The model is based on multi-process parameter modeling and uncertainty quantification to cope with the complexity of weld quality prediction caused by the nonlinear coupling of multiple process parameters and provide effective evaluation of the uncertainty of the prediction results. BRIEF DESCRIPTION OF DRAWINGS

[0016] Figure 1 A flowchart of a ring-shaped light spot laser welding quality prediction method provided by the embodiment of the application is shown in the figure. Figure 2 A flowchart of a multilayer perceptron-based regression model construction process involved in the embodiment of the application is shown in the figure. Figure 3 A structural block diagram of a ring-shaped light spot laser welding quality prediction system provided by the embodiment of the application is shown in the figure. Figure 4 A structural block diagram of an electronic device provided by the embodiment of the application is shown in the figure.

[0017] The implementation, functional features and advantages of the application will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION

[0018] In order for those skilled in the art to better understand the technical solutions of the application, the following describes exemplary embodiments of the application with reference to the accompanying drawings, which include various details of the embodiments of the application to help understanding, and should be considered only as exemplary. Therefore, those skilled in the art should realize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the application. Similarly, in order to be clear and concise, the description in the following description omits the description of well-known functions and structures.

[0019] In the case of no conflict, the embodiments of the application and the features in the embodiments can be combined with each other.

[0020] As used herein, the term “and / or” includes any and all combinations of one or more of the associated listed items.

[0021] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the present application. As used herein, the singular forms “a”, “an” and “the” are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms “comprises” and / or “comprising,” when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof. “Connected” or “coupled” or similar terms are not limited to a direct connection or coupling, but also include an indirect connection or coupling, such as through an intermediary.

[0022] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art. It will be further understood that terms, such as those defined in commonly used dictionaries, should be interpreted as having a meaning that is consistent with their meaning in the context of the relevant art and the present disclosure, and will not be interpreted in an overly literal or overly formal sense unless expressly so defined herein.

[0023] In the technical solutions of the present application, the collection, storage, use, processing, transmission, provision and disclosure of user personal information involved in the technical solutions comply with relevant laws and regulations and do not violate public order and good customs. The use of user data in the technical solutions complies with relevant national laws and regulations (for example, “Information Security Technology Personal Information Security Specification” and the like). For example, appropriate measures are taken for personal information access control; restrictions are given to the display of personal information; the use purpose of personal information does not exceed the direct or reasonably related range; and the use of personal information eliminates explicit identity pointing and avoids precise positioning to a specific individual.

[0024] In the related art, although certain progress has been made in modeling the laser welding process, some scholars attempt to use regression analysis, neural networks, support vector machines and other methods to establish a prediction model between process parameters and quality, but there are still the following deficiencies: first, the nonlinear coupling relationship between multiple parameters is complex, and traditional linear modeling is difficult to accurately describe the mapping relationship between the process and the weld quality; second, most existing models output single-value prediction results, lack of quantitative expression of prediction uncertainty, and cannot provide prediction confidence and risk assessment reference, which is not conducive to process optimization and quality control under complex working conditions; finally, in the actual welding process, due to factors such as sensor precision limitation, material property fluctuation and equipment stability, there are certain errors in the collected data itself, and the robustness of the model is insufficient in the presence of measurement errors and distribution deviation.

[0025] In the related art, there are the following problems: 1. In the laser welding process, there is a complex nonlinear coupling relationship between various process parameters (such as power, welding speed, and focal point position). The traditional linear model cannot accurately describe this high-order nonlinear relationship, resulting in limited prediction accuracy of the model. 2. Most existing methods only output a single prediction value, and cannot provide an evaluation of the reliability or error range of the prediction value, so the risk degree of the prediction error affecting the final welding quality cannot be quantified.

[0026] Therefore, the present application designs a ring-shaped light spot laser welding quality prediction method based on multi-process parameter modeling and uncertainty quantification. Considering the coupling effect of multiple process parameters, a welding quality modeling method with nonlinear modeling capability and output of prediction uncertainty is proposed to assist in realizing high-reliability and intelligent laser welding process parameter optimization and quality control.

[0027] The present application provides a ring-shaped light spot laser welding quality prediction method, system, device and storage medium.

[0028] The embodiment of the present application provides a ring-shaped light spot laser welding quality prediction method, referring to Figure 1 , Figure 1 The flowchart of an embodiment of the ring-shaped light spot laser welding quality prediction method of the present application is shown.

[0029] As shown in Figure 1 , the ring-shaped light spot laser welding quality prediction method comprises: Step S100: constructing a training set and a test set according to historical process parameters and historical weld quality results of ring-shaped light spot laser welding; Step S200: constructing a regression model based on a multilayer perceptron; Step S300: training the regression model based on the multilayer perceptron according to the training set to obtain a trained regression model based on the multilayer perceptron; Step S400: verifying the trained regression model based on the multilayer perceptron based on the test set; Step S500: obtaining current process parameters, inputting the current process parameters into the trained regression model based on the multilayer perceptron, and obtaining a current welding quality prediction value.

[0030] It should be noted that the execution subject in the present embodiment can be an electronic device, which can be a computer device with data processing function, or other devices that can realize the same or similar functions, and the present embodiment does not limit the execution subject, which is taken as a computer device for illustration in the present embodiment.

[0031] It can be understood that, in view of the problems of insufficient prediction accuracy of welding quality, lack of result confidence expression and difficulty in supporting laser welding risk prediction in the existing laser welding process, the embodiment discloses a ring light spot laser welding quality prediction method based on multi-process parameter modeling and uncertainty quantification, to cope with the complexity of weld quality prediction caused by the nonlinear coupling of multiple process parameters, and provide effective evaluation of the uncertainty of the prediction result.

[0032] Specifically, the method described in the embodiment uses a multilayer perceptron neural network (MLP) with nonlinear modeling capability to model the complex coupling relationship between multiple process parameters (such as outer ring power, inner ring power, welding speed) and welding quality indicators (penetration, width) in laser welding, extract key features that effectively reflect welding quality, and thus improve prediction accuracy; considering that actual welding data itself has errors, an uncertainty evaluation mechanism based on residual statistics is introduced based on model prediction output, to provide a confidence interval for welding quality prediction results, and enhance model interpretability and risk prediction ability.

[0033] In an embodiment, a training set and a test set are constructed according to historical process parameters and historical weld quality results of ring light spot laser welding, including: obtaining historical process parameters of ring light spot laser welding and corresponding historical weld quality results; constructing a sample data set according to the historical process parameters and the historical weld quality results; performing structured and standardized processing on the sample data set to obtain a processed data set; dividing the processed data set into a training set and a test set.

[0034] Specifically, the key quality indicators of the weld (such as penetration and width) obtained by ring light spot laser welding under different process parameters (including outer ring power, welding speed and center power) are collected. Specifically, it includes: through ring light spot laser welding experiments under different outer ring power (such as 0-4000w), center power (such as 500w-4000w) and welding speed (such as 10mm / s-100mm / s), by intercepting a typical cross section and observing the gold image using an optical microscope, the weld penetration and width indicators are obtained.

[0035] Exemplarily, according to the parameter combination scheme of orthogonal experiment, a plurality of structured sample data sets (including historical process parameters and historical weld quality results) with three process parameters (outer ring power, welding speed and center power) as input and weld quality indicators (penetration and width) as output are constructed, to provide basic data support for subsequent modeling.

[0036] In an embodiment, the sample data set is structured and standardized to obtain a processed data set, including: constructing a five tuple of three process parameters and two welding quality indicators in the sample data set; performing de-duplication, missing value filling and outlier removal processing on the five tuple to obtain a processed five tuple; uniformly storing the processed five tuple as a two-dimensional structured table format based on the physical magnitude and dimension standard of the process parameters to obtain a penetration and width data set; and performing standardization processing on the penetration and width data set to obtain a processed data set.

[0037] Specifically, the collected process parameters and corresponding welding result data (weld quality indicators) are structured and divided into a training set and a test set according to a preset ratio, for example, an 8:2 ratio.

[0038] Exemplarily, each set of experimental data is uniformly recorded as a five tuple containing three process input parameters and two welding quality indicators, the three process input parameters including outer ring laser power, center laser power and welding speed, and the two welding quality indicators including penetration and width; all sample data is de-duplicated, missing value filled and outlier removed, and stored uniformly as a two-dimensional structured table format according to the physical magnitude and dimension standard of the process parameters; Exemplarily, the three process input parameters are samples in the data set, and the two quality indicators are labels corresponding to the task, and the generated penetration and width data set can be represented as , ; wherein, is the th input sample, represents the size format of the input sample, and are the penetration and width values corresponding to the th input sample, is the total number of samples. Based on the principle of random uniform sampling, all structured sample data is divided into a training set and a test set according to a ratio of, for example, 8:2, and the test set is reserved for evaluating the generalization performance of the prediction model.

[0039] It can be understood that, considering the dimensional difference between different process parameters and weld quality indicators, the embodiment can adopt Z-score standardization for unified processing.

[0040] Specifically, the standardization processing: considering the physical unit and order of magnitude difference between different process parameters and weld quality indicators, Z-score standardization method can be used for unified processing of all input process parameters and output welding quality indicators. For each input variable (including outer ring power , center power , welding speed ) and output variable (penetration and width , respectively, and their mean and standard deviation in the dataset are calculated, which are expressed as:

[0041] wherein, is the th value in the input or output variable, is the corresponding normalized result. The corresponding normalized input variable and output variable can be expressed as , , , and , respectively.

[0042] In this embodiment, the modeling of multiple process parameters of ring-shaped spot laser welding is adopted to solve the problem of complex coupling relationship between the energy distribution and welding effect of ring-shaped spot. The multilayer perceptron neural network (MLP) with nonlinear modeling capability is used to model the complex coupling relationship between multiple process parameters (such as outer ring power, inner ring power, and welding speed) and welding quality indicators (such as penetration and width) in laser welding, extract key features that effectively reflect welding quality, and thus improve the prediction accuracy.

[0043] In an embodiment, the regression model based on the multilayer perceptron is constructed, including: constructing a penetration regression model based on the multilayer perceptron; constructing a width regression model based on the multilayer perceptron; and taking the penetration regression model based on the multilayer perceptron and the width regression model based on the multilayer perceptron as the regression model based on the multilayer perceptron; wherein the penetration regression model based on the multilayer perceptron and the width regression model based on the multilayer perceptron are independent of each other and have the same regression model structure.

[0044] In an embodiment, the penetration regression model based on the multilayer perceptron includes an input layer, a plurality of hidden layers, and an output layer; wherein the input layer includes three nodes corresponding to three process parameters, the hidden layers use activation functions, and the output layer is a single node. The output layer of the penetration regression model based on the multilayer perceptron outputs the normalized predicted value of the penetration. The output layer of the width regression model based on the multilayer perceptron outputs the normalized predicted value of the width.

[0045] Specifically, a regression model based on the multilayer perceptron (MLP) is constructed to fit the nonlinear relationship between the process parameters and the penetration and width, respectively. For example, Figure 2As shown, the model structure design: for the nonlinear mapping relationship between the input variables and the penetration, the input variables and the width, respectively, a regression model based on MLP is constructed. The two mapping tasks are independent of each other and the corresponding regression model structures are consistent. The MLP model is a feedforward neural network, which consists of an input layer, several hidden layers and an output layer. The input layer includes three nodes, respectively corresponding to three standardized process parameters: outer ring power , center power , welding speed . The number of hidden layers and nodes can be adjusted according to the experimental results, and the activation function is used to increase the nonlinear expression effect. The output layer is a single node, which outputs the standardized prediction value of the penetration and the width respectively.

[0046] Exemplarily, the mathematical expression of the model: this embodiment takes two hidden layers as an example, and the number of nodes in each hidden layer is 64. The penetration prediction model structure is input layer-hidden layer 1-hidden layer 2-output layer. Among them, the input layer can be expressed as:

[0047] Wherein, is the input process parameter vector.

[0048] The mathematical formula of hidden layer 1 can be expressed as:

[0049] Wherein, is the weight matrix of the network layer, is the bias vector of the network layer, is the activation function. Exemplarily, in this embodiment, the rectified linear unit function (ReLU) can be used as the activation function, and its expression is as follows:

[0050] Wherein, when the input value is less than 0, the output value is equal to 0; when the input is greater than or equal to 0, the output is equal to the input value.

[0051] The mathematical formula of hidden layer 2 can be expressed as:

[0052] Wherein, is the weight matrix of the network layer, is the bias vector of the network layer, and the other variables in the formula can refer to the parameter explanation of the above hidden layer 1 mathematical formula.

[0053] The mathematical formula of the output layer can be expressed as:

[0054] wherein, is the normalized penetration prediction value; is the network layer weight matrix, is the network layer bias vector, is the output of the hidden layer 2.

[0055] It should be noted that, by analogy, another set of MLP models with the same structure can be constructed for the prediction of the fusion width, and the predicted fusion width is , which will not be described here.

[0056] In an embodiment, the multilayer perceptron-based regression model is trained according to the training set to obtain a trained multilayer perceptron-based regression model, including: performing round-by-round iterative training on the multilayer perceptron-based penetration regression model based on the training set and a supervised target to obtain a trained penetration regression model; performing round-by-round iterative training on the multilayer perceptron-based fusion width regression model based on the training set and a supervised target to obtain a trained fusion width regression model; and generating a trained multilayer perceptron-based regression model according to the trained penetration regression model and the trained fusion width regression model.

[0057] Specifically, as shown in Figure 2 , the penetration and fusion width regression models are trained on the training set respectively, and the performance of the trained models is verified on the test set, while the uncertainty quantification index is generated based on the prediction residual to form the quality evaluation output containing the prediction value and its confidence interval.

[0058] Exemplarily, using the constructed multilayer perceptron-based penetration regression model and fusion width regression model, the process parameters in the training set are respectively taken as the model input, and the corresponding normalized values of the penetration or fusion width and are taken as the supervised target to perform round-by-round iterative training. The number of training cycles can be flexibly set according to the experimental effect, and the present embodiment does not limit it.

[0059] Exemplarily, in each round of training, the mean square error (MSE) between the model prediction output and the true value is calculated, and the model weight parameters are adjusted through the back propagation algorithm to minimize the loss function. The mathematical expression of the mean square error function is as follows:

[0060] wherein, is the total number of samples, and respectively. In backpropagation, the network weights and biases can be iteratively updated by different optimization algorithms such as stochastic gradient descent (SGD) or Adam optimizer, and the selection of the optimization algorithm can be determined according to the actual effect, and the embodiment is not limited thereto. Taking SGD as an example, the update process of the weights and biases of each layer can be expressed as follows:

[0061] wherein, is a learning rate, and are the gradients of the loss function with respect to the weights and biases of the i-th layer, respectively. The process is iterated until the loss function converges to a set threshold or reaches a maximum training round, thereby obtaining the optimal model parameters for welding quality prediction, and a trained multilayer perceptron-based regression model is generated according to the optimal model parameters.

[0062] In an embodiment, the trained multilayer perceptron-based regression model is verified based on the test set, including: obtaining the predicted value output by the trained multilayer perceptron-based regression model based on the test set; calculating the residual error between the predicted value and the actual value; calculating the mean and standard deviation of the residual error according to the distribution of the residual error; constructing a confidence interval according to the mean and standard deviation; and verifying the trained multilayer perceptron-based regression model according to the confidence interval.

[0063] Specifically, as shown in FIG. 6, according to each sample in the test set, the residual error Figure 2 between the model predicted value and the actual value is calculated. According to the distribution of the residual error, the mean and standard deviation are calculated, and a confidence interval is constructed with or interval, thereby reflecting the uncertainty of the predicted value. The calculation formula of the residual error is as follows:

[0064] Through quantitative analysis and visual display of the prediction residual error, the embodiment not only provides the predicted value of the welding quality, but also synchronously provides the uncertainty interval of the predicted value, thereby enhancing the explainability and safety guarantee capability of the model prediction, and being suitable for laser welding manufacturing scenarios requiring high reliability control.

[0065] ​​​​It can be understood that in machine learning and statistical modeling, a trained model is usually used to make predictions on new data (test set). However, the prediction results of the model often deviate from the true values, and this deviation is the residual. By analyzing the distribution of the residual, the uncertainty of the model prediction can be understood. The mean and standard deviation are important statistics to describe the distribution characteristics of the data, and the confidence interval can more intuitively reflect the possible fluctuation range of the predicted value, helping to evaluate the reliability of the model prediction.

[0066] In this embodiment, error distribution modeling is introduced into the regression model output, and ± nσ is used to construct a confidence band, thereby providing prediction reliability for welding quality evaluation and supporting manufacturing risk prediction. By calculating the residual, analyzing the residual distribution, calculating the mean and standard deviation of the residual, and constructing the confidence interval, the uncertainty of the model prediction can be more comprehensively understood, and the performance of the model can be evaluated. For example, if the confidence interval is narrow, the model prediction is more accurate; if the confidence interval is wide, the model prediction has greater uncertainty, and the model needs to be further improved.

[0067] In an example, based on the method described in this embodiment, a 4000w ring laser can be used to test Q235 structural steel (thickness not less than 12mm) by laser self-fusion welding, and the weld indicators generated under different welding process parameters are highly consistent with the model prediction results, effectively verifying the effectiveness and reliability of the method described in this embodiment in welding quality prediction.

[0068] The embodiment provides a ring-shaped light spot laser welding quality prediction method, comprising: constructing a training set and a test set according to historical process parameters and historical weld quality results of ring-shaped light spot laser welding; constructing a regression model based on a multilayer perceptron; training the regression model based on the multilayer perceptron according to the training set to obtain a trained regression model based on the multilayer perceptron; verifying the trained regression model based on the multilayer perceptron based on the test set; obtaining current process parameters, inputting the current process parameters into the trained regression model based on the multilayer perceptron, and obtaining a current welding quality prediction value. In this embodiment, the multilayer perceptron with nonlinear modeling capability is used to model the complex coupling relationship between the multiple process parameters and the welding quality indicators in laser welding, to obtain the trained regression model based on the multilayer perceptron, to predict the ring-shaped light spot laser welding quality, and to improve the prediction accuracy. The model is based on multi-process parameter modeling and uncertainty quantification to cope with the complexity of weld quality prediction caused by nonlinear coupling of multiple process parameters, and to provide effective evaluation of the uncertainty of the prediction results.

[0069] Reference Figure 3 , Figure 3 is a structural block diagram of an embodiment of the ring-shaped light spot laser welding quality prediction system.

[0070] As shown in Figure 3 The annular light spot laser welding quality prediction system comprises: A data acquisition module 10 is configured to construct a training set and a test set according to historical process parameters and historical weld quality results of annular light spot laser welding. A model construction module 20 is configured to construct a multilayer perceptron-based regression model. A model training module 30 is configured to train the multilayer perceptron-based regression model according to the training set to obtain a trained multilayer perceptron-based regression model. A model verification module 40 is configured to verify the trained multilayer perceptron-based regression model based on the test set. A model prediction module 50 is configured to obtain current process parameters, input the current process parameters into the trained multilayer perceptron-based regression model, and obtain a current weld quality prediction value.

[0071] The embodiment provides an annular light spot laser welding quality prediction system, adopts a multilayer perceptron neural network (MLP) having a nonlinear modeling capability, models a complex coupling relationship between various process parameters (for example, outer ring power, inner ring power, and welding speed) and welding quality indexes (for example, penetration and width) in laser welding, extracts key features effectively reflecting welding quality, and then improves prediction accuracy; considering that actual welding data itself has errors, an uncertainty evaluation mechanism based on residual statistics is introduced based on model prediction output, a confidence interval is provided for welding quality prediction results, and model interpretability and risk prediction capability are enhanced.

[0072] It should be noted that technical details not described in detail in the annular light spot laser welding quality prediction system embodiment can refer to the application of the annular light spot laser welding quality prediction method provided by any embodiment of the application as described above, and will not be described here.

[0073] Based on the same inventive concept, the embodiment of the application also provides an electronic device. Figure 4 A structural block diagram of an electronic device provided by the embodiment of the application is shown in Figure 4 The embodiment of the application provides an electronic device comprising one or more processors 101, a memory 102, and one or more I / O interfaces 103. The memory 102 stores one or more programs, and when the one or more programs are executed by the one or more processors, the one or more processors implement the annular light spot laser welding quality prediction method in any of the above embodiments; the one or more I / O interfaces 103 are connected between the processor and the memory and are configured to realize information interaction between the processor and the memory.

[0074] The processor 101 is a device with data processing capability, including but not limited to a central processing unit (CPU) and the like; the memory 102 is a device with data storage capability, including but not limited to a random access memory (RAM, more specifically SDRAM, DDR, etc.), a read-only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), and a flash memory (FLASH); the I / O interface (read-write interface) 103 is connected between the processor 101 and the memory 102, and can realize information interaction between the processor 101 and the memory 102, including but not limited to a data bus (Bus) and the like.

[0075] In some embodiments, the processor 101, the memory 102, and the I / O interface 103 are connected to each other through the bus 104, and further connected to other components of the computing device.

[0076] In some embodiments, the one or more processors 101 include a field programmable gate array.

[0077] The embodiments of the present application also provide a computer readable medium. The computer readable medium stores a computer program, wherein the program is executed by a processor to implement the steps in the annular light spot laser welding quality prediction method in any of the above embodiments. The computer readable storage medium can be a volatile or non-volatile computer readable storage medium.

[0078] The embodiments of the present application also provide a computer program product, including computer readable code or a non-volatile computer readable storage medium carrying computer readable code, when the computer readable code is run in a processor of an electronic device, the processor in the electronic device executes the annular light spot laser welding quality prediction method.

[0079] Those skilled in the art can understand that all or some of the steps in the above disclosed method, the functions of the modules / units in the system and the device can be implemented as software, firmware, hardware, and appropriate combinations thereof. In the hardware implementation, the division between the functional modules / units mentioned in the above description does not necessarily correspond to the division of physical components; for example, one physical component can have multiple functions, or one function or step can be performed by several physical components in cooperation. Some or all of the physical components can be implemented as software executed by a processor, such as a central processing unit, a digital signal processor, or a microprocessor, or as hardware, or as an integrated circuit, such as an application specific integrated circuit. Such software can be distributed on a computer readable storage medium, which can include computer storage media (or non-transitory media) and communication media (or transitory media).

[0080] As those skilled in the art will appreciate, the term computer storage media includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable program instructions, data structures, program modules or other data. Computer storage media include, but are not limited to, random access memory (RAM), read only memory (ROM), erasable programmable read only memory (EPROM), electrically erasable programmable read only memory (EEPROM), static random access memory (SRAM), flash memory or other memory technology, portable compact disc read only memory (CD-ROM), digital versatile disk (DVD) or other optical disk storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium which can be used to store the desired information and which can be accessed by a computer. Further, as those skilled in the art will appreciate, communication media typically embodies computer readable program instructions, data structures, program modules or other data in a modulated data signal such as a carrier wave or other transport mechanism and includes any information delivery media. The term "modulated data signal" means a signal that has one or more of its characteristics changed or set in a manner so as to encode information in the signal. By way of example, and not limitation, communication media includes wired media such as a wired network or direct-wired connection, and wireless media such as wireless networks, cellular telephone networks, code division multiple access (CDMA) networks, and other terrestrial and satellite radio frequency communication networks. Thus the computer readable program instructions and / or other program modules can be embodied in a computer readable storage medium, which can be any device or article that is enab!ed to store and / or carry computer readable program instructions and / or data structures. The computer readable storage medium can also be distributed over networked computer systems so that the computer readable program instructions and / or other program modules are stored and executed in a distributed fashion.

[0081] Computer readable program instructions described herein can be downloaded to respective computing / processing devices from a computer readable storage medium or to an external computer or external storage device via a network, for example, the Internet, a local area network, a wide area network and / or a wireless network. The network can comprise copper transmission cables, optical transmission fibers, wireless transmission, routers, firewalls, switches, gateway computers and / or edge servers. A network adapter card or network interface in each computing / processing device receives computer readable program instructions from the network and forwards the computer readable program instructions for storage in a computer readable storage medium within the respective computing / processing device.

[0082] Computer readable program instructions for carrying out operations of the present application can be assembler instructions, instruction-set-architecture (ISA) instructions, machine instructions, machine dependent instructions, microcode, firmware instructions, state-setting data, or either source code or object code written in any combination of one or more programming languages, including an object oriented programming language such as Smalltalk, C++ or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The computer readable program instructions can execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection can be made to an external computer (for example, through the Internet using an Internet Service Provider). In some embodiments, electronic circuitry including, for example, programmable logic circuitry, field-programmable gate array (FPGA), or programmable logic array (PLA) can execute the computer readable program instructions by utilizing state information of the computer readable program instructions to personalize the electronic circuitry, in order to perform aspects of the present application.

[0083] The computer program product described herein can be embodied in a specific manner by hardware, software, or a combination thereof. In an optional embodiment, the computer program product is embodied as a computer storage medium. In another optional embodiment, the computer program product is embodied as a software product, such as a software development kit (SDK), and the like.

[0084] The computer program product described herein can be embodied in a specific manner by hardware, software, or a combination thereof. In an optional embodiment, the computer program product is embodied as a computer storage medium. In another optional embodiment, the computer program product is embodied as a software product, such as a software development kit (SDK), and the like.

[0085] These computer readable program instructions can be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions / acts specified in the flowchart and / or block diagram block or blocks. These computer readable program instructions can also be stored in a computer readable storage medium that can include a non-transitory computer readable storage medium that can be a computer- readable storage medium having no data storage cycles that change state. The instructions can be executed by one or more processors of a computer, other programmable data processing apparatus, or other devices to produce a computer-implemented process such that the instructions which execute via the one or more processors of the computer or other programmable data processing devices create means for implementing the functions / acts specified in the flowchart and / or block diagram block or blocks. The computer program instructions can also be loaded onto a computer, other programmable data processing apparatus, or other devices to cause a series of operational steps to be performed on the computer, other programmable apparatus or other devices to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide processes for implementing the functions / acts specified in the flowchart and / or block diagram block or blocks.

[0086] The computer readable program instructions can also be loaded onto a computer, other programmable data processing apparatus, or other device to cause a series of operational steps to be performed on the computer, other programmable apparatus or other device to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide processes for implementing the functions / acts specified in the flowchart and / or block diagram block or blocks.

[0087] The flow and block diagrams in the drawings show the architectural, functional, and operational views of possible implementations of systems, methods, and computer program products according to the present application. In this regard, each block in the flow and block diagrams can represent a module, a segment, or a portion of instructions which contain one or more executable instructions for implementing the specified logical function(s). In some alternative implementations, the functions noted in the blocks can occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently or the blocks may

[0088] Example embodiments have been disclosed and, although specific terms are employed, they are used in a generic and descriptive sense only and not for purposes of limitation. In some instances, it will be apparent to those skilled in the art that features, characteristics or elements described with reference to one embodiment can be used in combination with features, characteristics or elements described with reference to other embodiments unless otherwise explicitly stated. Accordingly, it will be understood that various changes in form and details can be made without departing from the scope of the present application as set forth in the appended claims.

Claims

1. A method for predicting the quality of annular spot laser welding, characterized in that, The method comprises the following steps: constructing a training set and a test set according to historical process parameters and historical weld quality results of ring spot laser welding; constructing a multilayer perceptron-based regression model; training the multilayer perceptron-based regression model according to the training set to obtain a trained multilayer perceptron-based regression model; verifying the trained multilayer perceptron-based regression model based on the test set; obtaining current process parameters, inputting the current process parameters into the trained multilayer perceptron-based regression model, and obtaining a current weld quality prediction value.

2. The method of claim 1, wherein, The method of constructing a training set and a test set according to historical process parameters and historical weld quality results of ring spot laser welding comprises the following steps: obtaining historical process parameters and corresponding historical weld quality results of ring spot laser welding; constructing a sample data set according to the historical process parameters and the historical weld quality results; performing structured and standardized processing on the sample data set to obtain a processed data set; dividing the processed data set into a training set and a test set.

3. The method of claim 2, wherein, The method of performing structured and standardized processing on the sample data set to obtain a processed data set comprises the following steps: constructing a five-tuple in the sample data set with three process parameters and two weld quality indicators; performing de-duplication, missing value filling and outlier removal processing on the five-tuple to obtain a processed five-tuple; unifying the processed five-tuple into a two-dimensional structured table format based on the physical magnitude and dimension standard of the process parameters to obtain a penetration depth and width data set; performing standardized processing on the penetration depth and width data set to obtain a processed data set.

4. The method of claim 1, wherein, The method of constructing a multilayer perceptron-based regression model comprises the following steps: constructing a multilayer perceptron-based penetration depth regression model; constructing a multilayer perceptron-based penetration width regression model; taking the multilayer perceptron-based penetration depth regression model and the multilayer perceptron-based penetration width regression model as a multilayer perceptron-based regression model; wherein the multilayer perceptron-based penetration depth regression model and the multilayer perceptron-based penetration width regression model are independent of each other and have the same regression model structure.

5. The method of claim 4, wherein, The multilayer perceptron-based penetration depth regression model comprises an input layer, a plurality of hidden layers and an output layer; wherein the input layer comprises three nodes corresponding to three process parameters, the hidden layers adopt an activation function, and the output layer is a single node; the output layer of the multilayer perceptron-based penetration depth regression model outputs a standardized penetration depth prediction value; the output layer of the multilayer perceptron-based penetration width regression model outputs a standardized penetration width prediction value.

6. The method of claim 4, wherein, The method of training the multilayer perceptron-based regression model according to the training set to obtain a trained multilayer perceptron-based regression model comprises the following steps: performing round-by-round iterative training on the multilayer perceptron-based penetration depth regression model based on the training set and a supervision target to obtain a trained penetration depth regression model; performing round-by-round iterative training on the multilayer perceptron-based penetration width regression model based on the training set and a supervision target to obtain a trained penetration width regression model; generating a trained multilayer perceptron-based regression model according to the trained penetration depth regression model and the trained penetration width regression model.

7. The method of any one of claims 1 to 6, wherein, The verifying the trained multi-layer perceptron-based regression model based on the test set comprises: obtaining a predicted value output by the trained multi-layer perceptron-based regression model based on the test set; calculating a residual between the predicted value and an actual value; calculating a mean value and a standard deviation of the residual according to a distribution of the residual; constructing a confidence interval according to the mean value and the standard deviation; verifying the trained multi-layer perceptron-based regression model according to the confidence interval.

8. A ring spot laser welding quality prediction system, characterized by, The method comprises: a data acquisition module configured to construct a training set and a test set according to historical process parameters and historical weld quality results of ring-shaped light spot laser welding; a model construction module configured to construct a multi-layer perceptron-based regression model; a model training module configured to train the multi-layer perceptron-based regression model according to the training set to obtain a trained multi-layer perceptron-based regression model; a model verification module configured to verify the trained multi-layer perceptron-based regression model based on the test set; a model prediction module configured to obtain current process parameters, input the current process parameters into the trained multi-layer perceptron-based regression model, and obtain a current weld quality prediction value.

9. An electronic device, comprising: The electronic device comprises a memory, a processor, and a ring-shaped light spot laser welding quality prediction program stored in the memory and executable on the processor, and the ring-shaped light spot laser welding quality prediction program is configured to implement the ring-shaped light spot laser welding quality prediction method according to any one of claims 1 to 7.

10. A storage medium, characterized by The storage medium stores a ring-shaped light spot laser welding quality prediction program, and the ring-shaped light spot laser welding quality prediction program is used to enable the processor to implement the ring-shaped light spot laser welding quality prediction method according to any one of claims 1 to 7 when executed.

Images