A data-driven method and system for optimizing bolt machining process parameters

Abstract

This invention relates to the fields of data processing and machine learning technology, and in particular to a data-driven method and system for optimizing bolt machining process parameters. The method includes: acquiring machining data of the target equipment in real time and preprocessing it to form a process dataset; extracting multiple key features from the process dataset to form a multimodal feature vector; obtaining optimization tasks, executing different model building strategies based on the correlation between optimization tasks to obtain a quality process correlation model; and finding the optimal combination of process parameters that meets the machining quality indicators by using a multi-objective Bayesian optimization algorithm adapted to different model building strategies based on the constructed quality process correlation model.

Description

Technical Field

[0001] This invention relates to the fields of data processing and machine learning technology, and in particular to a data-driven method and system for optimizing bolt machining process parameters. Background Technology

[0002] Data-driven approaches are profoundly changing traditional manufacturing. Their application value is particularly significant in bolt machining, especially in the machining of grooves (such as internal hexagonal, Torx, and Phillips head slots) where high precision and consistency are required.

[0003] Traditional process parameter optimization methods lack comprehensive utilization of multimodal data and correlation modeling between multiple quality indicators, making it difficult to achieve intelligent parameter adaptation in dynamic processing environments, resulting in fluctuations in processing quality and low efficiency.

[0004] To address this issue, we propose a data-driven approach to optimize process parameters, aiming to solve the problem of difficulty in adapting process parameters in dynamic processing environments. Summary of the Invention

[0005] This invention achieves intelligent recommendation of process parameters through feature extraction, quantification of correlation of quality indicators, differential model construction strategy, and multi-objective Bayesian optimization.

[0006] The technical solution proposed in this invention is: a data-driven method for optimizing bolt machining process parameters, the method comprising:

[0007] Real-time acquisition and preprocessing of processing data from the target equipment to form a process dataset; extraction of multiple key features from the process dataset to form a multimodal feature vector;

[0008] Obtain optimization tasks, and execute different model building strategies based on the correlation between optimization tasks to obtain a quality process correlation model;

[0009] Based on the constructed quality-process correlation model, the optimal combination of process parameters that meets the processing quality index is found by using a multi-objective Bayesian optimization algorithm adapted to different model construction strategies.

[0010] Preferably, the historical processing data includes target equipment operating status data, process parameter data, raw material data, and quality index data;

[0011] The extraction of multiple key features from the process dataset to form a multimodal feature vector includes:

[0012] Extract key features from historical processing data; these key features include: the average current of the spindle. Peak current Average amplitude Peak amplitude Average speed Peak speed Average feed rate and material hardness Maximum axial cutting depth of bolt groove Maximum radial cutting width and symmetry Constructing multimodal feature vectors ;in, These represent the normalized spindle current average, peak current, average amplitude, peak amplitude, average rotational speed, peak speed, average feed rate, material hardness, maximum axial depth of cut, maximum radial width of cut, and symmetry, respectively.

[0013] Preferably, the step of obtaining optimization tasks involves executing different model building strategies based on the correlation between optimization tasks to obtain a quality process correlation model, including:

[0014] Obtain optimization tasks and calculate the correlation between each optimization task, i.e., the correlation between quality indicators;

[0015] Cluster analysis is performed on the correlations, and based on the analysis results, different model building strategies are implemented to construct corresponding process correlation models.

[0016] Preferably, the step of obtaining optimization tasks and calculating the correlation between each optimization task, i.e., the correlation between quality indicators, includes:

[0017] Measure the linear correlation between any two quality indicators, including:

[0018] For any two quality indicators, if their historical observations are obtained, the Pearson correlation coefficient between the two quality indicators is: ;

[0019] in, and Indicators of quality and exist Historical observations at that time Indicates the number of historical observations; and Indicators and The average of historical observations;

[0020] Measure the nonlinear correlation between any two quality metrics, including:

[0021] Calculate the maximum information data for any two quality indicators. ;

[0022] in, express A set of grid divisions; Represents a grid Mutual information estimation under the following conditions , Indicates the grid in Number of intervals in the direction, ; Indicates the number of horizontal or vertical grid lines;

[0023] Calculate the distance correlation between any two quality indicators ;in, This represents the distance covariance between two quality indicators, and the distance variance. Distance variance ;

[0024] Then, the comprehensive correlation score of any two quality indicators is: ; , , This represents the fusion weighting coefficient.

[0025] Preferably, the cluster analysis of the correlation includes:

[0026] Constructing a Relationship Graph: Defining a Weighted Undirected Graph Among them, vertex set Represents quality indicators; edge set , Indicates the overall correlation threshold;

[0027] weight matrix Edge weight ;in, The square of the distance metric representing the physical meaning between two quality indicators; The variance of the distance between the physical meanings of two quality indicators;

[0028] Perform spectral clustering analysis, including:

[0029] Constructing the Laplace matrix ,in, Represents a diagonal matrix, with diagonal elements. ;

[0030] Normalized Laplace matrix: Perform eigenvalue decomposition: ,in, Indicates the first 1 eigenvalue, ; Indicates the quantity of quality indicators; Represents the eigenvalue weights;

[0031] Before choosing Construct the feature vector from the smallest non-zero eigenvalues. ;right Perform K-means clustering to obtain index groups ;

[0032] Modularity as an evaluation of cluster quality Among them, the comprehensive weighting coefficient index Quality Indicators Overall weight Quality Indicators Overall weight ;

[0033] Represents the category coefficient, if and In the same category, ;otherwise, .

[0034] Preferably, based on the analysis results, different model building strategies are executed to construct corresponding process correlation models, including:

[0035] if If so, then execute model building strategy one, that is, build an independent quality process correlation model;

[0036] if Then, execute model building strategy two, that is, build a global multi-task quality process correlation model;

[0037] if and If so, execute model building strategy three, that is, build a grouped multi-task model; otherwise, execute model building task four, that is, build a hierarchical hybrid quality process correlation model.

[0038] The construction of the independent quality process correlation model includes:

[0039] Meet the conditions and At that time, the first The independent model for each quality indicator is ;in, ;

[0040] ;in, , It represents a sub-feature vector composed of the same number of arbitrary features from the multimodal feature vector; Represents the kernel function. Represent a Gaussian process; Indicates noise;

[0041] The construction of the global multi-task quality process correlation model includes:

[0042] Meet the conditions and or ;

[0043] Maintain consistency between the model form and the independent quality process association model; for the function Core the linear model, that is: Among them, the first A latent function ; Indicates the number of latent functions; Indicates the first The first quality indicator The weighting coefficients of each latent function;

[0044] ;in, Represents the comprehensive correlation score matrix The Middle A comprehensive correlation score; This represents a threshold for the number of latent functions;

[0045] The construction of the grouped multi-task quality process correlation model includes:

[0046] Meet the conditions and At that time, group modeling is performed:

[0047] For the first Cluster Establish sub-models ;

[0048] in, Indicates belonging to The first cluster The first quality indicator The weighting coefficients of each latent function; Indicates belonging to The first cluster The first quality indicator One potential function;

[0049] Therefore, the grouped multi-task quality process correlation model is constructed as follows: ;in, Indicates belonging to The first cluster Noise in each quality indicator; ;

[0050] The construction of the hierarchical hybrid quality process correlation model includes:

[0051] Assume there is a hierarchical structure among the quality indicators, meaning that secondary quality indicators are influenced by primary quality indicators.

[0052] Construct common layer functions at the first-level indicator layer. ;

[0053] Construct specific layer functions at the secondary indicator layer. ;

[0054] The stratified mixed quality process correlation model is as follows: ; ; This represents the model output of the first-level indicator layer; This represents the model output of the second-level indicator layer;

[0055] Dynamic monitoring of the overall correlation score includes:

[0056] Establish a sliding time window Real-time data is collected and processed through a sliding time window to form a real-time sample set.

[0057] Multiple key features are extracted from the real-time acquired processing data to form a real-time multimodal feature vector. Using this real-time multimodal feature vector, a real-time comprehensive correlation score is calculated. ;

[0058] if Then use renew ;in, Indicates the threshold to be updated; express The norm of .

[0059] Preferably, the step of finding the optimal combination of process parameters that satisfies the processing quality index based on the constructed quality process correlation model, through a multi-objective Bayesian optimization algorithm adapted to different model construction strategies, includes:

[0060] Obtain the corresponding quality process correlation model;

[0061] Based on different quality correlation models, a multi-objective Bayesian optimization algorithm is used to optimize them. The optimization stops after reaching the maximum number of iterations, yielding the optimal combination of process parameters, including:

[0062] Constructing a multi-objective optimization objective function ;

[0063] The constraints are: ; ;

[0064] in, This represents the difference between the process parameter and its corresponding threshold value. This indicates that the predicted value of the quality indicator is less than the quality indicator threshold. The probability of; This represents the acceptable probability value of risk. Indicates the first The output of the quality process model corresponding to each quality indicator;

[0065] Based on different model building strategies, variations are made to the acquisition function in the multi-objective Bayesian optimization algorithm;

[0066] The objective function is solved using a multi-objective Bayesian optimization algorithm after sampling a variant of the objective function. After reaching the preset maximum number of iterations, the optimal parameter feature vector is obtained. .

[0067] Preferably, the variation of the acquisition function in the multi-objective Bayesian optimization algorithm based on different model building strategies includes:

[0068] For model building strategy one, the acquisition function ; Indicates the independent overvolume improvement in quality indicators; Indicates the first Independent overvolume improvement for each quality indicator;

[0069] For model building strategy two, a quasi-Monte Carlo integral approximation is used to obtain the acquisition function. ;

[0070] in, Indicates the origin from a multivariate Gaussian distribution The target value vector sampled in the middle; This represents the mean vector of quality indicators predicted by the quality process correlation model. This represents the variance vector of the quality indicators predicted by the quality process correlation model. Indicates the first One standard Gaussian sampling point, Indicates the number of sampling points. express The volumetric improvement; Indicates Gaussian transform;

[0071] For model building strategy three, the acquisition function ; Indicates the number of groups. Indicates the first The data acquisition function for each group;

[0072] For model building strategy four, the collection function of the first-level indicator layer ,in, This represents the mean vector of quality indicators predicted by the quality process correlation model at the first-level indicator layer.

[0073] Secondary indicator layer acquisition function ; Indicates the first The expected improvement in each quality indicator.

[0074] A data-driven bolt machining process parameter optimization system, comprising:

[0075] The data acquisition and preprocessing module is used to acquire and preprocess the processing data;

[0076] The correlation analysis module is used to calculate the comprehensive correlation score between quality indicators and perform cluster analysis;

[0077] The model building module is used to execute different modeling strategies based on the correlation.

[0078] The storage module is used to store the process dataset and the optimal parameter feature vector.

[0079] A computer-readable storage medium storing a computer program that is executed by a processor to implement the aforementioned data-driven method for optimizing bolt machining process parameters.

[0080] The beneficial effects of this invention are:

[0081] 1. This invention achieves differentiated model building strategies by quantifying the linear and nonlinear correlations between quality indicators, thereby improving the prediction accuracy and efficiency of quality process correlation models.

[0082] 2. This invention adjusts the multi-objective optimization algorithm for different model building strategies. Specifically, it designs a data acquisition function for the multi-objective Bayesian optimization algorithm that adapts to different model building strategies, making the multi-objective optimization algorithm more efficient and accurate in finding the optimal process parameters. Attached Figure Description

[0083] Figure 1 This is a flowchart of a data-driven method for optimizing bolt processing parameters according to the present invention. Detailed Implementation

[0084] The following description is intended to disclose the present invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious modifications will occur to those skilled in the art. The basic principles of the invention defined in the following description can be applied to other embodiments, modifications, improvements, equivalents, and other technical solutions that do not depart from the spirit and scope of the invention.

[0085] It is understood that the term "a" should be understood as "at least one" or "one or more", that is, in one embodiment, the number of an element can be one, while in another embodiment, the number of the element can be multiple, and the term "a" should not be understood as a limitation on the number.

[0086] refer to Figure 1 The technical solution provided by this invention is: a data-driven method for optimizing bolt processing parameters, the method comprising:

[0087] Step 1: Acquire and preprocess the processing data of the target equipment in real time to form a process dataset; extract multiple key features from the process dataset to form a multimodal feature vector; the historical processing data includes target equipment operating status data, process parameter data, raw material data, and quality index data. Specifically, extracting multiple key features from the process dataset to form a multimodal feature vector includes the following steps:

[0088] Key features are extracted from historical processing data, including the average spindle current. Peak current Average amplitude Peak amplitude Average speed Peak speed Average feed rate and material hardness Maximum axial cutting depth of bolt groove Maximum radial cutting width and symmetry Constructing multimodal feature vectors ;in, These represent the normalized spindle current average, peak current, average amplitude, peak amplitude, average rotational speed, peak speed, average feed rate, material hardness, maximum axial depth of cut, maximum radial width of cut, and symmetry, respectively.

[0089] Step 2: Obtain optimization tasks, and execute different model building strategies based on the correlation between optimization tasks to obtain a quality process correlation model. This includes the following steps:

[0090] Step 2.1: Obtain optimization tasks and calculate the correlation between each optimization task, i.e., the correlation between quality indicators. Specifically:

[0091] To measure the linear correlation between any two quality indicators, that is, for any two quality indicators, obtain their historical observations, and then the Pearson correlation coefficient between the two quality indicators is: ;

[0092] in, and Indicators of quality and exist Historical observations at that time Indicates the number of historical observations; and Indicators and The average of historical observations;

[0093] if If the correlation between the two quality indicators is weak, then it is determined that the correlation between the two quality indicators is weak.

[0094] if If the correlation between the two quality indicators is moderate, then the correlation between them is determined to be moderate.

[0095] if If the correlation between the two quality indicators is strong, then it can be determined that there is a strong correlation between them.

[0096] Measure the nonlinear correlation between any two quality metrics, including:

[0097] Calculate the maximum information data for any two quality indicators (capturing arbitrary functional relationships). ;

[0098] in, express A set of grid divisions; Represents a grid Mutual information estimation under the following conditions , Indicates the grid in Number of intervals in the direction, ; Indicates the number of horizontal or vertical grid lines;

[0099] if If the nonlinear relationship between the two quality indicators is weak, then it is determined that the relationship between the two quality indicators is weak.

[0100] if If the nonlinearity between the two quality indicators is deemed to be moderate, then the nonlinearity between them is considered to be moderate.

[0101] if If the nonlinear relationship between the two quality indicators is strong, then it can be determined that the relationship between them is strong; among them, , This represents the threshold for non-linear relationships.

[0102] Calculate the distance correlation between any two quality metrics (capture arbitrary dependencies). ;in, This represents the distance covariance between two quality indicators, and the distance variance. Distance variance .

[0103] The composite correlation score of any two quality indicators is ; , , , This represents the fusion weighting coefficient.

[0104] Step 2.2: Perform cluster analysis on the correlations. Based on the analysis results, implement different model building strategies to construct corresponding process correlation models, specifically including the following steps:

[0105] Step 2.21: Construct a correlation graph, including:

[0106] Define a weighted undirected graph Among them, vertex set Represents quality indicators; edge set , Indicates the overall correlation threshold;

[0107] weight matrix Edge weight ;in, The square of the distance metric representing the physical meaning between two quality indicators; The variance of the distance between the physical meanings of two quality indicators.

[0108] Step 2.22: Perform spectral cluster analysis, including:

[0109] Constructing the Laplace matrix ,in, Represents a diagonal matrix, with diagonal elements. ;

[0110] Normalized Laplace matrix: ;

[0111] Perform eigenvalue decomposition: ,in, Indicates the first 1 eigenvalue, ; Indicates the quantity of quality indicators; Represents the eigenvalue weights;

[0112] Before choosing Construct the feature vector from the smallest non-zero eigenvalues. ;right Perform K-means clustering to obtain index groups .

[0113] Modularity as an evaluation of cluster quality Among them, the comprehensive weighting coefficient index Quality Indicators Overall weight Quality Indicators Overall weight . Represents the category coefficient, if and In the same category, ;otherwise, .

[0114] Step 2.23, if and Then, execute model building strategy one, which is to build an independent quality process correlation model, specifically:

[0115] No. The independent model for each quality indicator is ;in, .

[0116] ;in, , It represents a sub-feature vector composed of the same number of arbitrary features from the multimodal feature vector; Represents kernel functions (e.g., quadratic exponential kernel, Matrn kernel). Represent a Gaussian process; Indicates noise.

[0117] Step 2.24, if , or Then, the second model building strategy will be executed, which is to build a global multi-task quality process correlation model, specifically as follows:

[0118] Maintain consistency between the model form and the independent quality process association model. For the function... Core the linear model, that is: Among them, the first A latent function ; Indicates the number of latent functions; Indicates the first The first quality indicator The weight coefficients of each latent function.

[0119] ;in, Represents the comprehensive correlation score matrix The Middle A comprehensive correlation score; This represents the threshold for the number of latent functions.

[0120] Step 2.25, if and If the condition is met, then execute model building strategy three, which involves building a grouped multi-task model; otherwise, execute model building task four, which involves building a hierarchical hybrid quality process correlation model. Specifically:

[0121] For the first Cluster Establish sub-models ;

[0122] in, Indicates belonging to The first cluster The first quality indicator The weighting coefficients of each latent function; Indicates belonging to The first cluster The first quality indicator A potential function.

[0123] Therefore, the grouped multi-task quality process correlation model is constructed as follows: ;in, Indicates belonging to The first cluster Noise in each quality indicator; .

[0124] Step 2.26: Construct a hierarchical hybrid quality process correlation model. The specific steps are as follows:

[0125] Assume there is a hierarchical structure among the quality indicators, where secondary quality indicators are influenced by primary quality indicators. Construct a common layer function at the primary indicator layer. Construct specific layer functions at the secondary indicator layer. .

[0126] Therefore, the stratified mixed quality process correlation model is as follows: ; ; This represents the model output of the first-level indicator layer; This represents the model output of the second-level indicator layer.

[0127] For example, when a milling machine performs grooving on bolts, the sub-feature vector we are interested in includes features such as the average spindle feed rate. Peak speed Average speed Maximum axial cutting depth of bolt groove Maximum radial cutting width The outputs (quality indicators) of concern include groove depth deviation, surface roughness, and groove symmetry error.

[0128] Step 2.27: Dynamically monitor the comprehensive correlation score, including:

[0129] Establish a sliding time window Real-time data is collected and processed through a sliding time window to form a real-time sample set.

[0130] Multiple key features are extracted from the real-time acquired processing data to form a real-time multimodal feature vector. Using this real-time multimodal feature vector, a real-time comprehensive correlation score is calculated. ;if Then use renew ;in, Indicates the threshold to be updated; express The norm (Frobenius norm).

[0131] For example, in bolt machining, surface roughness is related to cutting force fluctuation, which in turn is related to vibration amplitude. These three factors are correlated, and the correlation is non-linear. Groove depth deviation is related to tool runout, which in turn is related to spindle vibration amplitude. These three factors are also correlated, and the correlation is moderately linear.

[0132] By analyzing correlations and performing differentiated modeling, the accuracy of model predictions is improved while the computational cost of the model is reduced, thus increasing efficiency.

[0133] Step 3: Based on the constructed quality-process correlation model, use a multi-objective Bayesian optimization algorithm adapted to different model construction strategies to find the optimal combination of process parameters that satisfies the processing quality indicators. This specifically includes the following steps:

[0134] Obtain the corresponding quality-process correlation model; based on different quality-process correlation models, optimize them using a multi-objective Bayesian optimization algorithm. Stop optimization after reaching the maximum number of iterations to obtain the optimal combination of process parameters, including:

[0135] Constructing a multi-objective optimization objective function ;

[0136] The constraints are: ; ;

[0137] in, This represents the difference between a process parameter and its corresponding threshold value (deterministic inequality constraints, such as a spindle feed rate less than the maximum feed rate, or a spindle speed less than the maximum spindle speed). This indicates that the predicted value of the quality indicator is less than the quality indicator threshold. The probability of; This represents the acceptable probability value of risk. Indicates the first The output of the quality process model corresponding to each quality indicator.

[0138] Before solving the objective function using the multi-objective Bayesian optimization algorithm, variations of the acquisition function in the multi-objective Bayesian optimization algorithm are made based on different model building strategies. These variations include the following steps:

[0139] For model building strategy one, since the quality metrics are independent, the data collection function... ; Indicates the independent overvolume improvement in quality indicators; Indicates the first Independent overvolume improvement for each quality indicator.

[0140] For model building strategy two, a quasi-Monte Carlo integral approximation is used to obtain the acquisition function. .in, Indicates the origin from a multivariate Gaussian distribution The target value vector sampled in the middle; This represents the mean vector of quality indicators predicted by the quality process correlation model. Indicates the first One standard Gaussian sampling point, Indicates the number of sampling points. express The volumetric improvement; This represents the Gaussian transform (Gaussian Copula transform). This represents the variance vector of the quality indicators predicted by the quality process correlation model.

[0141] For model building strategy three, the acquisition function ; Indicates the number of groups. Indicates the first Acquisition functions for each group.

[0142] For model building strategy four, the collection function of the first-level indicator layer ,in, This represents the mean vector of quality indicators predicted by the quality process correlation model at the first-level indicator layer.

[0143] Secondary indicator layer acquisition function ; Indicates the first The expected improvement in each quality indicator.

[0144] The objective function is solved using a multi-objective Bayesian optimization algorithm after sampling a variant of the objective function. After reaching the preset maximum number of iterations, the optimal parameter feature vector is obtained. The optimal parameter eigenvectors include, but are not limited to, the optimal average spindle speed, peak speed, and average feed rate.

[0145] The core of adjusting the multi-objective Bayesian optimization algorithm lies in designing different data acquisition functions. By using different acquisition functions, it is ensured that regardless of the modeling strategy employed, the optimal combination of process parameters can be found while considering the correlation of quality indicators, thereby improving the accuracy and stability of bolt machining. This method is applicable to industrial big data analysis platforms and can be used for data modeling and parameter optimization of the machining process, without relying on specific control system hardware.

[0146] This invention also provides a data-driven bolt machining process parameter optimization system, comprising:

[0147] The data acquisition and preprocessing module is used to acquire and preprocess the processing data;

[0148] The correlation analysis module is used to calculate the comprehensive correlation score between quality indicators and perform cluster analysis;

[0149] The model building module is used to execute different modeling strategies based on the correlation.

[0150] The storage module is used to store the process dataset and the optimal parameter feature vector.

[0151] The present invention also provides a computer-readable storage medium storing a computer program that is executed by a processor to implement the aforementioned data-driven method for optimizing bolt processing parameters.

[0152] The processes described above with reference to the flowcharts in the embodiments disclosed in this invention can be implemented as computer software programs. The embodiments disclosed in this invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication component, and / or installed from a removable medium. When the computer program is executed by a central processing unit (CPU), it performs the functions defined in the methods of this application. It should be noted that the computer-readable medium described above in this application can be a computer-readable signal medium or a computer-readable storage medium, or any combination of the two. Computer-readable storage media can be, for example, but not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatuses, or devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wire segments, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this application, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in connection with an instruction execution system, apparatus, or device. In this application, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on a computer-readable medium may be transmitted using any suitable medium, including but not limited to: wireless segments, wire segments, optical fibers, RF, etc., or any suitable combination thereof.

[0153] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation that may be implemented in systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0154] Those skilled in the art should understand that the embodiments of the present invention described above and shown in the accompanying drawings are merely examples and do not limit the present invention. The purpose of the present invention has been fully and effectively achieved. The functions and structural principles of the present invention have been shown and explained in the embodiments. Without departing from the principles described, the implementation of the present invention may have any changes or modifications.

Claims

1. A data-driven method for optimizing bolt machining process parameters, characterized in that, The method includes: Real-time acquisition and preprocessing of processing data from the target equipment to form a process dataset; extraction of multiple key features from the process dataset to form a multimodal feature vector; Obtain optimization tasks, and execute different model building strategies based on the correlation between optimization tasks to obtain a quality process correlation model; Based on the constructed quality-process correlation model, the optimal combination of process parameters that meets the processing quality index is found by using a multi-objective Bayesian optimization algorithm adapted to different model construction strategies.

2. The data-driven method for optimizing bolt machining process parameters according to claim 1, characterized in that, The historical processing data includes target equipment operating status data, process parameter data, raw material data, and quality index data; The extraction of multiple key features from the process dataset to form a multimodal feature vector includes: Key features are extracted from historical processing data, including the average spindle current. Peak current Average amplitude Peak amplitude Average speed Peak speed Average feed rate and material hardness Maximum axial cutting depth of bolt groove Maximum radial cutting width and symmetry Constructing multimodal feature vectors ;in, These represent the normalized spindle current average, peak current, average amplitude, peak amplitude, average rotational speed, peak speed, average feed rate, material hardness, maximum axial depth of cut, maximum radial width of cut, and symmetry, respectively.

3. The data-driven method for optimizing bolt machining process parameters according to claim 2, characterized in that, The process of obtaining optimization tasks involves executing different model building strategies based on the correlation between optimization tasks to obtain a quality process correlation model, including: Obtain optimization tasks and calculate the correlation between each optimization task, i.e., the correlation between quality indicators; Cluster analysis is performed on the correlations, and based on the analysis results, different model building strategies are implemented to construct corresponding process correlation models.

4. The data-driven method for optimizing bolt machining process parameters according to claim 3, characterized in that, The process of acquiring optimization tasks and calculating the correlation between each optimization task, i.e., the correlation between quality indicators, includes: Measure the linear correlation between any two quality indicators, including: For any two quality indicators, if their historical observations are obtained, the Pearson correlation coefficient between the two quality indicators is: ;in, and Indicators of quality and exist Historical observations at that time Indicates the number of historical observations; and Indicators and The average of historical observations; Measure the nonlinear correlation between any two quality metrics, including: Calculate the maximum information data for any two quality indicators. ; in, express A set of grid divisions; Represents a grid Mutual information estimation under the following conditions , Indicates the grid in Number of intervals in the direction, ; Indicates the number of horizontal or vertical grid lines; Calculate the distance correlation between any two quality indicators ;in, This represents the distance covariance between two quality indicators, and the distance variance. Distance variance ; Then, the comprehensive correlation score of any two quality indicators is: ; , , This represents the fusion weighting coefficient.

5. The data-driven method for optimizing bolt machining process parameters according to claim 4, characterized in that, The clustering analysis of the correlations includes: Constructing a Relationship Graph: Defining a Weighted Undirected Graph Among them, vertex set Represents quality indicators; edge set , Indicates the overall correlation threshold; weight matrix Edge weight ;in, The square of the distance metric representing the physical meaning between two quality indicators; The variance of the distance between the physical meanings of two quality indicators; Perform spectral clustering analysis, including: Constructing the Laplace matrix ,in, Represents a diagonal matrix, with diagonal elements. ; Normalized Laplace matrix: ; Perform eigenvalue decomposition: ,in, Indicates the first 1 eigenvalue, ; Indicates the quantity of quality indicators; Represents the eigenvalue weights; Before choosing Construct the feature vector from the smallest non-zero eigenvalues. ;right Perform K-means clustering to obtain index groups ; Modularity as an evaluation of cluster quality Among them, the comprehensive weighting coefficient index Quality Indicators Overall weight Quality Indicators Overall weight ; Represents the category coefficient, if and In the same category, ;otherwise, .

6. The data-driven method for optimizing bolt machining process parameters according to claim 5, characterized in that, Based on the analysis results, different model building strategies are implemented to construct corresponding process correlation models, including: if If so, then execute model building strategy one, that is, build an independent quality process correlation model; if Then, execute model building strategy two, that is, build a global multi-task quality process correlation model; if and If so, execute model building strategy three, that is, build a grouped multi-task model; otherwise, execute model building task four, that is, build a hierarchical hybrid quality process correlation model. The construction of the independent quality process correlation model includes: Meet the conditions and At that time, the first The independent model for each quality indicator is ;in, ; ;in, , It represents a sub-feature vector composed of the same number of arbitrary features from the multimodal feature vector; Represents the kernel function. Represent a Gaussian process; Indicates noise; The construction of the global multi-task quality process correlation model includes: Meet the conditions and or ; Maintain consistency between the model form and the independent quality process association model; for the function Core the linear model, that is: Among them, the first A latent function ; Indicates the number of latent functions; Indicates the first The first quality indicator The weighting coefficients of each latent function; ;in, Represents the comprehensive correlation score matrix The Middle A comprehensive correlation score; This represents a threshold for the number of latent functions; The construction of the grouped multi-task quality process correlation model includes: Meet the conditions and At that time, group modeling is performed: For the first Cluster Establish sub-models ; in, Indicates belonging to The first cluster Quality Indicators The The weighting coefficients of each latent function; Indicates belonging to The first cluster The first quality indicator One potential function; Therefore, the grouped multi-task quality process correlation model is constructed as follows: ;in, Indicates belonging to The first cluster Noise in each quality indicator; ; The construction of the hierarchical hybrid quality process correlation model includes: Assume there is a hierarchical structure among the quality indicators, meaning that secondary quality indicators are influenced by primary quality indicators. Construct common layer functions at the first-level indicator layer. ; The kernel function represents the common layer functions; Construct specific layer functions at the secondary indicator layer. ; A kernel function representing a specific layer of functions; The stratified mixed quality process correlation model is as follows: ; ; This represents the model output of the first-level indicator layer; This represents the model output of the second-level indicator layer; Dynamic monitoring of the overall correlation score includes: Establish a sliding time window Real-time data is collected and processed through a sliding time window to form a real-time sample set. Multiple key features are extracted from the real-time acquired processing data to form a real-time multimodal feature vector. Using this real-time multimodal feature vector, a real-time comprehensive correlation score is calculated. ; if Then use renew ;in, Indicates the threshold to be updated; express The norm of .

7. The data-driven method for optimizing bolt machining process parameters according to claim 6, characterized in that, The constructed quality-process correlation model, through a multi-objective Bayesian optimization algorithm adapted to different model construction strategies, seeks the optimal combination of process parameters that satisfies the processing quality indicators, including: Obtain the corresponding quality process correlation model; Based on different quality correlation models, a multi-objective Bayesian optimization algorithm is used to optimize them. The optimization stops after reaching the maximum number of iterations, yielding the optimal combination of process parameters, including: Constructing a multi-objective optimization objective function ; The constraints are: ; ; in, This represents the difference between the process parameter and its corresponding threshold value. This indicates that the predicted value of the quality indicator is less than the quality indicator threshold. The probability of; This represents the acceptable probability value of risk. Indicates the first The output of the quality process model corresponding to each quality indicator; Based on different model building strategies, variations are made to the acquisition function in the multi-objective Bayesian optimization algorithm; The objective function is solved using a multi-objective Bayesian optimization algorithm after sampling a variant of the objective function. After reaching the preset maximum number of iterations, the optimal parameter feature vector is obtained. .

8. The data-driven method for optimizing bolt machining process parameters according to claim 7, characterized in that, The variations of the acquisition function in the multi-objective Bayesian optimization algorithm based on different model construction strategies include: For model building strategy one, the acquisition function ; Indicates the independent overvolume improvement in quality indicators; Indicates the first Independent overvolume improvement for each quality indicator; For model building strategy two, a quasi-Monte Carlo integral approximation is used to obtain the acquisition function. ;in, Indicates the origin from a multivariate Gaussian distribution The target value vector sampled in the middle; This represents the mean vector of quality indicators predicted by the quality process correlation model. This represents the variance vector of the quality indicators predicted by the quality process correlation model. Indicates the first One standard Gaussian sampling point, Indicates the number of sampling points. express The volumetric improvement; Represents the Gaussian Copula transform; For model building strategy three, the acquisition function ; Indicates the number of groups. Indicates the first The data acquisition function for each group; For model building strategy four, the collection function of the first-level indicator layer ,in, This represents the mean vector of quality indicators predicted by the quality-process correlation model at the first-level indicator layer; the data collection function at the second-level indicator layer. ; Indicates the first The expected improvement in each quality indicator.

9. A data-driven bolt machining process parameter optimization system, characterized in that, include: The data acquisition and preprocessing module is used to acquire and preprocess the processing data; The correlation analysis module is used to calculate the comprehensive correlation score between quality indicators and perform cluster analysis; The model building module is used to execute different modeling strategies based on the correlation. The storage module is used to store the process dataset and the optimal parameter feature vector.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that is executed by a processor to implement a data-driven method for optimizing bolt processing parameters as described in any one of claims 1-8.

Images