Intelligent optimization method for fastener production process

Abstract

The application provides an intelligent optimization method for a fastener production process, comprising the following steps: step one, data acquisition and preprocessing; step two, screening out key process parameters; step three, establishing a quality prediction model by using a multiple linear regression method with the key process parameters as input and the quality index as output; step four, determining the optimal recommended value and the allowable fluctuation range of each key process parameter according to the quality prediction model and the target quality range; and step five, performing production optimization according to the allowable fluctuation range of the key process parameters and performing real-time monitoring. Through whole-process data acquisition and correlation, the application realizes the identification of key process parameters and the determination of the optimization window in the whole-process procedure of fastener production by using correlation analysis, multiple linear regression and grid search, and the method is easy to implement, reduces the optimization cost, and takes into account the quality and production efficiency.

Description

Technical Field

[0001] This invention relates to the field of fastener manufacturing technology, and in particular to an intelligent optimization method for the fastener manufacturing process. Background Technology

[0002] The production of fasteners (such as bolts, nuts, and screws) typically involves multiple processes, including cold heading, thread rolling, heat treatment, and surface treatment. The process parameters of each process (such as cold heading pressure, die temperature, thread rolling speed, heat treatment temperature, holding time, and cooling rate) directly determine the key performance indicators of the final product, such as hardness, dimensional accuracy, and thread quality. Currently, most fastener manufacturers have installed basic sensors and data acquisition devices at each process stage to obtain some process parameter data. They have also installed quality inspection equipment such as hardness testers and dimensional measuring instruments at the end of the production line. However, this data is scattered across different processes and equipment, and has not yet been effectively integrated and utilized.

[0003] In recent years, some patents have attempted to optimize specific processes in fastener production using data analysis methods. For example, Chinese invention patent CN120893331A discloses a method for optimizing fastener heat treatment process parameters based on big data analysis. This method collects process parameters, equipment operating parameters, environmental parameters, and performance testing parameters during the heat treatment process to construct sub-optimization models for hardness uniformity, residual stress balance, and deformation index. These sub-models are then integrated into a multi-objective comprehensive optimization model, and an improved multi-objective intelligent optimization algorithm is used to solve for the optimal combination of heat treatment process parameters. This method performs relatively in-depth big data modeling and optimization for a single heat treatment process, demonstrating a certain degree of advancement. However, this method still has the following shortcomings: First, the optimization scope is limited to the heat treatment process and does not cover other key processes such as cold heading, thread rolling, and surface treatment. The final quality of fasteners is often the result of the coupling effect of multiple processes, and it is difficult to achieve global optimization by optimizing a single process. Second, this method uses complex models such as principal component analysis dimensionality reduction, random forest feature importance ranking, and multi-objective intelligent optimization algorithms, which require a large amount of data, consume a lot of computing resources, and require professional personnel to maintain, making it difficult for small and medium-sized fastener enterprises to apply directly. Third, this method does not establish a closed-loop mechanism that feeds back the end quality inspection results to the parameters of the preceding processes, making it difficult to quickly locate the cause when quality problems occur. Summary of the Invention

[0004] This invention addresses the aforementioned problems and aims to provide an intelligent optimization method for fastener production. Through full-process data collection and correlation, it achieves quality traceability across multiple processes. It employs correlation analysis, multiple linear regression, and grid search to identify key process parameters and determine optimization windows throughout the fastener production process. The method is easy to implement, reduces optimization costs, and provides a quantified optimal parameter window and a hierarchical response mechanism, balancing quality and production efficiency.

[0005] Specifically, the first aspect of the present invention provides an intelligent optimization method for a fastener manufacturing process, comprising the following steps: Step 1: Collect data at each stage of fastener production and at the end of the production line, and preprocess the collected data; Step 2: Correlate the process parameters of each process with the quality inspection data, and select key process parameters based on the correlation coefficient between the process parameters and quality indicators; Step 3: Using key process parameters as input and quality indicators as output, establish a quality prediction model using the multiple linear regression method; Step 4: Based on the quality prediction model and the target quality range, determine the optimal recommended value and allowable fluctuation range for each key process parameter; Step 5: Optimize production based on the allowable fluctuation range of key process parameters and monitor it in real time.

[0006] Furthermore, the data collection in step one includes: Cold heading process: Collect cold heading forming pressure and mold temperature; Cold heading pressure refers to the force applied by the cold heading machine punch when it forges metal wire into the shape of a bolt head or blank. It is usually measured in kilonewtons. Cold heading pressure directly affects the degree of metal flow and work hardening effect. Too low a pressure will result in incomplete filling of the head, missing material, or cracks; too high a pressure may cause die overload, accelerated wear, or even internal micro-cracks in the blank. During cold heading, intense friction between the mold and the metal generates heat. Increased mold temperature can affect the fluidity of the metal and the life of the mold. Excessively high mold temperature may cause lubricant failure and mold sticking, while excessively low temperature may increase the resistance to metal flow.

[0007] Thread rolling process: collect thread rolling pressure and thread rolling plate feed speed; Thread rolling pressure refers to the radial force exerted by the thread rolling machine when rolling threads onto the bolt blank through the relative movement of two thread rolling plates. The magnitude of thread rolling pressure is closely related to the fullness of the thread profile, surface roughness, and residual stress distribution. Insufficient pressure will lead to incomplete thread profile and a smaller pitch diameter; excessive pressure may cause burns on the thread surface or even chipping of the thread rolling plates.

[0008] Thread rolling plate feed speed: refers to the axial movement speed of the thread rolling plate during the thread rolling process, usually measured in meters per minute. The feed speed affects the thread forming time. Too fast a speed may result in a rough thread surface and incomplete tooth profile; too slow a speed will reduce production efficiency.

[0009] Heat treatment process: collecting heating temperature, holding time, and cooling rate; Heating temperature: refers to the highest temperature to which fasteners are heated in a heat treatment furnace, usually in degrees Celsius. For medium carbon steel or alloy steel fasteners, the quenching heating temperature directly affects the degree of austenitization, and thus determines the martensite content and hardness after quenching.

[0010] Holding time: This refers to the time a fastener is held at a heated temperature, usually measured in minutes. The holding time needs to ensure uniform temperature between the fastener's core and surface, and to allow for complete austenitization. Too short a time will result in incomplete core transformation; too long a time will increase energy consumption and may lead to grain growth.

[0011] Cooling rate: refers to the rate at which a fastener cools down from its heating temperature when quenched in a cooling medium, usually measured in degrees Celsius per second. The cooling rate determines the degree of martensitic transformation and the magnitude of residual stress.

[0012] Surface treatment process: sampling and processing time, drying temperature; The immersion or spraying time refers to the time that fasteners are immersed in or sprayed in the surface treatment bath, usually measured in seconds or minutes. The immersion time affects the coating thickness, uniformity, and adhesion. Too short a time results in a thin coating with insufficient corrosion resistance; too long a time may lead to an excessively thick coating, increased costs, or even sagging.

[0013] Drying temperature: refers to the temperature at which the surface-treated fasteners are heated in a drying oven, usually expressed in degrees Celsius. The drying temperature determines the degree of curing and cross-linking of the coating. If the temperature is too low, curing will be incomplete and the coating will easily peel off; if the temperature is too high, the coating may age, discolor, or the substrate may soften due to tempering.

[0014] At the end of the production line: collect data on product hardness, key dimensions, and appearance pass rate.

[0015] Hardness: refers to the ability of a fastener to resist localized indentation deformation, usually expressed as Rockwell hardness or Vickers hardness. Hardness is a core indicator of the mechanical properties of fasteners and directly reflects the quality of heat treatment. A hardness value that is too low indicates insufficient tensile strength; a value that is too high may lead to hydrogen embrittlement or decreased toughness.

[0016] Critical dimensions include the major diameter, pitch diameter, minor diameter, head height, and shank diameter of the thread. These dimensions determine the interchangeability of the fastener assembly and the reliability of the connection. Dimensional deviations can lead to assembly failure or uncontrolled preload.

[0017] Appearance pass rate: The percentage of items that pass the inspection, whether visual or by manual sampling, reflects surface quality issues such as cracks, burrs, rust, and coating defects.

[0018] Furthermore, the preprocessing in step one includes data cleaning, missing value handling, feature extraction, and data standardization.

[0019] The raw data collected often contains problems such as noise, missing values, and dimensional differences. Using it directly will affect the accuracy of subsequent analysis, so preprocessing is required.

[0020] Missing value handling includes: for individual missing data, filling in the missing data with the average of the parameter at the previous and next time points; for consecutive missing data exceeding a preset threshold, deleting the missing data segment.

[0021] Data standardization includes: using the min-max standardization method to uniformly convert all process parameters to the [0,1] range.

[0022] Furthermore, the data cleaning includes removing outliers that exceed the equipment's range or physical limits, and deleting samples with incomplete records.

[0023] Furthermore, the feature extraction includes extracting at least one of the following features from each production cycle: maximum value, minimum value, average value, standard deviation, and slope, for continuously sampled time series data.

[0024] Sensors for cold heading pressure and thread rolling pressure typically sample at high frequencies, with a single bolt production cycle potentially containing hundreds of data points. Inputting all raw data points into the model results in a massive dataset with high autocorrelation, unsuitable for direct correlation analysis. Feature extraction compresses high-dimensional time series into representative statistics: the maximum value reflects peak load; the average value reflects overall load level; the standard deviation reflects the degree of fluctuation, which can be used to determine die wear or wire unevenness; and the slope reflects the loading rate, which is related to impact energy. These features retain key information while significantly reducing data dimensionality.

[0025] Furthermore, in step two, the specific method for screening key process parameters is as follows: calculate the Pearson correlation coefficient between each process parameter and the quality indicator, sort them from largest to smallest absolute value, and select parameters whose sum of absolute values ​​of correlation coefficients exceeds 70% as key process parameters.

[0026] The Pearson correlation coefficient r measures the degree of linear correlation between two variables, with a value ranging from -1 to 1. The closer |r| is to 1, the stronger the correlation. A positive value indicates a positive correlation, meaning that as the parameter increases, the quality index also increases, while a negative value indicates a negative correlation.

[0027] A cumulative contribution rate of 70% is an engineering rule of thumb. If the threshold is too low, the omission of key parameters may have a significant impact on quality; if the threshold is too high, too many parameters will be selected, losing their "key" meaning. A 70% threshold strikes a balance between information completeness and simplification. In practice, the cumulative contribution of the first 3-5 parameters is usually sufficient to reach 70%.

[0028] Furthermore, step three specifically includes: using the key process parameters selected in step two as inputs and quality indicators as outputs, establishing a quality prediction model using the multiple linear regression method, and calculating the goodness of fit of the model.

[0029] Suppose there are m key process parameters selected, denoted as m. , , …, The quality indicators are The multiple linear regression equation is in the form of: ; in: For quality indicators; The intercept; This is the first regression coefficient; This is the second regression coefficient; Let m be the regression coefficient; This is the first critical process parameter; This is the second key process parameter; This refers to the m-th critical process parameter; This is random error; Using historical sample data, the optimal estimate of the regression coefficient is obtained by minimizing the sum of squared residuals. The residual of a sample is the error between the actual observed quality index and the quality index predicted by the model. The sum of squared residuals is the sum of the squared residuals of all samples. The solution process can be completed automatically using matrix operations or statistical software.

[0030] Furthermore, if the goodness of fit of the model is less than the set model validity threshold, then a quadratic term or interaction term is added to correct the model.

[0031] The formula for calculating goodness of fit is as follows: ; ; ; in: For goodness of fit; The sum of squared residuals reflects the degree of deviation between the model's predicted values ​​and the actual values; The total sum of squares reflects the overall degree of fluctuation of actual observed values ​​around their mean. Let be the actual value of the quality index for the i-th sample; This is the average of the actual values ​​of all quality indicators. Let be the predicted value of the quality index for the i-th sample; Furthermore, step four includes: taking the output of the quality prediction model falling within the target quality range as a constraint, using a grid search method to traverse the feasible domain of parameters to obtain a set of parameter combinations, taking the median of the set of solutions as the optimal recommended value, and taking the boundary of the set of solutions as the allowable fluctuation range.

[0032] Grid search method: Within the feasible region of parameters, a grid is divided with a certain step size. Each grid point is traversed, and the parameter values ​​are substituted into the quality prediction model to calculate the predicted quality index. All grid points whose predicted values ​​fall within the target range are then selected. The parameter combinations corresponding to these grid points constitute the solution set.

[0033] For each key process parameter, the minimum, maximum and distribution center of the parameter are statistically analyzed. The median is taken as the optimal recommended value, and the interval between the minimum and maximum values ​​is taken as the allowable fluctuation range.

[0034] Based on the allowable fluctuation range of key process parameters, the production line is adjusted accordingly to ensure that the collected process parameter data is within the allowable fluctuation range and is as close as possible to the optimal recommended value.

[0035] Furthermore, the real-time monitoring includes: real-time monitoring of key process parameters during production; issuing a yellow warning when the parameter exceeds the allowable fluctuation range but does not exceed 10% of the allowable fluctuation range; issuing a red warning and recommending shutdown for inspection when the parameter exceeds the allowable fluctuation range by more than 10% or deviates for more than 5 minutes. Attached Figure Description

[0036] To more clearly illustrate the technical solutions in the embodiments of this drawing or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this drawing. For those skilled in the art, other drawings can be obtained based on the structures shown in these drawings without creative effort.

[0037] Figure 1 This is a flowchart of the steps of the present invention; The purpose, features, and advantages of this accompanying drawing will be further explained in conjunction with the embodiments and with reference to the accompanying drawing. Detailed Implementation

[0038] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be described and illustrated below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. All other embodiments obtained by those skilled in the art based on the embodiments provided by this invention without inventive effort are within the scope of protection of this invention.

[0039] Obviously, the accompanying drawings described below are merely some examples or embodiments of the present invention. Those skilled in the art can apply the present invention to other similar scenarios based on these drawings without any inventive effort. Furthermore, it is understood that although the efforts made in this development process may be complex and lengthy, for those skilled in the art related to the content disclosed in this invention, modifications to design, manufacturing, or production based on the technical content disclosed in this invention are merely conventional technical means and should not be construed as insufficient disclosure of the present invention.

[0040] Unless otherwise specified, all embodiments and optional embodiments of the present invention can be combined with each other to form new technical solutions.

[0041] Unless otherwise specified, all technical features and optional technical features of this invention can be combined to form new technical solutions.

[0042] Unless otherwise specified, all steps of the present invention may be performed sequentially or randomly, preferably sequentially. For example, the method includes steps (a) and (b), indicating that the method may include steps (a) and (b) performed sequentially, or it may include steps (b) and (a) performed sequentially. For example, the mention that the method may also include step (c) indicates that step (c) may be added to the method in any order. For example, the method may include steps (a), (b), and (c), or it may include steps (a), (c), and (b), or it may include steps (c), (a), and (b), etc.

[0043] Unless otherwise specified, the terms "comprising" and "including" as used in this invention can be open-ended or closed-ended. For example, "comprising" and "including" can mean that other components not listed may also be included, or that only the listed components may be included.

[0044] Unless otherwise specified, the term "or" is inclusive in this invention. For example, the phrase "A or B" means "A, B, or both A and B". More specifically, the condition "A or B" is satisfied by any of the following conditions: A is true (or exists) and B is false (or does not exist); A is false (or does not exist) and B is true (or exists); or both A and B are true (or exist).

[0045] To better understand the solutions of the embodiments of the present invention, some related terms and concepts that may be involved in the embodiments of the present invention will be introduced below.

[0046] (1) Pearson correlation coefficient, a statistic that measures the degree of linear correlation between two variables, with a value range of [-1, 1]. The closer to 1, the stronger the correlation. A positive value indicates a positive correlation, and the quality index increases as the parameter increases. A negative value indicates a negative correlation.

[0047] (2) Multiple linear regression is a statistical modeling method that studies the linear relationship between a dependent variable and multiple independent variables, establishes a quantitative relationship model between process parameters and quality indicators, and is used to predict the quality results under different parameter combinations, providing a basis for parameter optimization.

[0048] In this embodiment, as Figure 1 As shown, an intelligent optimization method for fastener manufacturing process includes the following steps: Step 1: Collect data at each stage of fastener production and at the end of the production line, and preprocess the collected data; Step 2: Correlate the process parameters of each process with the quality inspection data, and select key process parameters based on the correlation coefficient between the process parameters and quality indicators; Step 3: Using key process parameters as input and quality indicators as output, establish a quality prediction model using the multiple linear regression method; Step 4: Based on the quality prediction model and the target quality range, determine the optimal recommended value and allowable fluctuation range for each key process parameter; Step 5: Optimize production based on the allowable fluctuation range of key process parameters and monitor it in real time.

[0049] In this embodiment, taking the production line for producing M8×1.25 hexagonal head bolts as an example, the preparation work for the data acquisition steps is as follows: A Kistler 9136B piezoelectric pressure sensor with a range of 0-30kN and an accuracy of ±0.5% is installed at each station of the cold heading machine. PT100 thermocouples are embedded near the mold. The sampling frequency is 1kHz, and 200 pressure data points are recorded for each bolt production cycle. The mold temperature is recorded once per cycle. An FOCD type pressure sensor with a range of 0-15kN is installed on the feed mechanism of the thread rolling machine. The feed speed of the thread rolling plate is collected by an encoder, and an average pressure value and a feed speed value are recorded for each bolt. K-type thermocouples with an accuracy of ±2℃ are installed in the heating zone of the continuous mesh belt furnace. The holding time is recorded by PLC. An infrared cooling rate monitor is installed at the outlet of the quenching tank. A set of data is recorded for each batch. Processing time and drying oven temperature were recorded on the Dacromet coating line, with one set of data recorded for each batch; At the end of the production line, a Mitutoyo LSM-506 laser diameter gauge is installed to measure the major diameter, pitch diameter, and shank diameter of the thread; a Wilson Hardness BP3001 hardness tester is used for a 5% sampling rate to test the Rockwell hardness (HRC); and a Cognex VisionPro vision system is used to detect appearance defects such as cracks and burrs. Average hardness, dimensional compliance rate, and appearance compliance rate are calculated for each batch.

[0050] Data cleaning: The cold heading pressure of batch 8 was -3.2kN, indicating a sensor malfunction; all data for this batch was deleted. The mold temperature record for batch 12 was 498℃, far exceeding the equipment's upper limit, indicating a poor thermocouple contact; this data point was deleted. A total of 2 abnormal batches were removed, leaving 28 valid batches.

[0051] Missing value handling: The heat preservation time data for batch 15 was missing, so it was filled by the average of 31.5 min from batch 14 (32 min) and batch 16 (31 min). Data for other batches were complete.

[0052] Feature extraction: For each bolt, 200 cold heading pressure points are used to calculate the peak pressure, average value, and standard deviation. For a specific bolt, the calculated pressure data is: peak pressure 23.2 kN, average value 18.5 kN, and standard deviation 2.3 kN. Individual parameters such as die temperature and thread rolling pressure are used directly from their original values.

[0053] Data standardization: Taking heating temperature as an example, the minimum value among the 28 batches is 835℃, and the maximum value is 860℃. Therefore, the standardization of a batch at 845℃ is (845-835) / (860-835) = 0.4. All parameters are converted to the [0,1] interval using this formula.

[0054] Using hardness HRC as the quality index, the Pearson correlation coefficients between various process parameters and hardness were calculated, and the results are as follows: The correlation coefficients for heating temperature, cold heading pressure, holding time, and cooling rate are 0.85, 0.62, 0.58, 0.45, 0.21, 0.18, 0.15, 0.12, and 0.08, respectively. The sum of the first four correlation coefficients is 2.50, accounting for 77% of the sum of the absolute values ​​of all correlation coefficients (3.24), exceeding the 70% threshold. Therefore, the key process parameters selected are: heating temperature, cold heading pressure, holding time, and cooling rate.

[0055] Using heating temperature, cold heading pressure, holding time, and cooling rate as inputs, and hardness HRC as output, a multiple linear regression model was established. Standardized data from 28 batches were used, and the regression coefficients were solved using the least squares method. The results were obtained using an Excel data analysis toolkit. ; The fitted equation is as follows: ; The residual sum of squares was calculated to be 38.6, the total sum of squares was 227.1, and the goodness of fit was 0.83, which is greater than the preset threshold of 0.7. Therefore, the model is effective and there is no need to introduce quadratic or interaction terms. The target hardness range for this embodiment is 34 HRC-36 HRC.

[0056] Based on equipment capabilities and process experience, the feasible range of parameters in this embodiment is: Heating temperature: 830℃-860℃, in increments of 2℃; Cold heading pressure: 21kN-25kN, in increments of 0.5kN; Incubation time: 28-35 minutes, in 1-minute increments; Cooling rate: 40℃-55℃ / s, step size 2℃ / s; The grid search generated approximately 15×9×8×8=8640 parameter combinations. Each combination was substituted into the model to calculate the predicted hardness. Combinations with predicted values ​​in the range of 34HRC-36 HRC were selected, resulting in approximately 1200 effective combinations.

[0057] The effective range of each parameter in statistical efficient combinations: Heating temperature: 838℃-852℃, center value 845℃, allowable fluctuation ±7℃; Cold heading pressure: 22.2 kN-23.8 kN, center value 23.0 kN, allowable fluctuation ±0.8 kN; Insulation time: 30.5 min - 33.5 min, with a center value of 32.0 min, and an allowable fluctuation of ±1.5 min; Cooling rate: 44℃ / s-50℃ / s, center value 47℃ / s, allowable fluctuation ±3℃ / s; Traditional parameter optimization methods typically provide an "optimal value," but in actual production, due to factors such as sensor accuracy, equipment inertia, and environmental fluctuations, it is difficult to precisely control parameters at a single value point. Overly stringent requirements can lead to frequent alarms and downtime, impacting production efficiency. This invention uses a grid search to obtain an "allowable fluctuation window." As long as the parameters fall within this window, the predicted quality will meet requirements. This approach ensures product quality while allowing for reasonable tolerances in the production process, achieving a balance between quality and efficiency.

[0058] The optimized process window was applied to the actual production line and run continuously for 10 batches. The results compared with the 10 batches before optimization are as follows: the average hardness before optimization was 33.2 HRC, and after optimization it was 35.1 HRC, an improvement of 1.9 HRC; the standard deviation of hardness before optimization was 2.8 HRC, and after optimization it was 1.3 HRC, a reduction of 54%, resulting in more stable production and helping to improve the yield rate and reduce the scrap rate; the hardness qualification rate before optimization was 89.5%, and after optimization it was 96.8%, an improvement of 7.3%, which greatly improved the production quality.

[0059] It should be noted that the present invention is not limited to the above-described embodiments. The above embodiments are merely examples, and any embodiments that have the same structure and perform the same effects as the technical concept within the scope of the present invention are included within the scope of the present invention. Furthermore, various modifications that can be conceived by those skilled in the art to the embodiments, and other ways of constructing by combining some of the constituent elements of the embodiments, without departing from the spirit of the present invention, are also included within the scope of the present invention.

Claims

1. A smart optimization method for fastener manufacturing process, characterized in that, Includes the following steps: Step 1: Collect data at each stage of fastener production and at the end of the production line, and preprocess the collected data; Step 2: Correlate the process parameters of each process with the quality inspection data, and select key process parameters based on the correlation coefficient between the process parameters and quality indicators; Step 3: Using key process parameters as input and quality indicators as output, establish a quality prediction model using the multiple linear regression method; Step 4: Based on the quality prediction model and the target quality range, determine the optimal recommended value and allowable fluctuation range for each key process parameter; Step 5: Optimize production based on the allowable fluctuation range of key process parameters and monitor it in real time.

2. The intelligent optimization method for fastener production process according to claim 1, characterized in that, The data collection in step one includes: Cold heading process: Collect cold heading forming pressure and mold temperature; Thread rolling process: collect thread rolling pressure and thread rolling plate feed speed; Heat treatment process: collecting heating temperature, holding time, and cooling rate; Surface treatment process: sampling and processing time, drying temperature; At the end of the production line: collect data on product hardness, key dimensions, and appearance pass rate.

3. The intelligent optimization method for fastener production process according to claim 1, characterized in that, The preprocessing in step one includes data cleaning, missing value handling, feature extraction, and data standardization.

4. The intelligent optimization method for fastener production process according to claim 3, characterized in that, The data cleaning includes removing outliers that exceed the equipment's range or physical limits, and deleting samples with incomplete records.

5. The intelligent optimization method for fastener production process according to claim 3, characterized in that, The feature extraction includes extracting at least one of the following features from each production cycle: maximum value, minimum value, average value, standard deviation, and slope, for continuously sampled time series data.

6. The intelligent optimization method for fastener production process according to claim 1, characterized in that, In step two, the specific method for screening key process parameters is as follows: calculate the Pearson correlation coefficient between each process parameter and the quality indicator, sort them from largest to smallest absolute value, and select parameters whose sum of absolute values ​​of correlation coefficients exceeds 70% as key process parameters.

7. The intelligent optimization method for fastener production process according to claim 1, characterized in that, Step three specifically includes: using the key process parameters selected in step two as inputs and quality indicators as outputs, establishing a quality prediction model using the multiple linear regression method, and calculating the goodness of fit of the model.

8. The intelligent optimization method for fastener production process according to claim 7, characterized in that, If the goodness of fit of the model is less than the set model validity threshold, then a quadratic term or interaction term is added to correct the model.

9. The intelligent optimization method for fastener production process according to claim 1, characterized in that, Step four includes: taking the output of the quality prediction model falling within the target quality range as a constraint, using a grid search method to traverse the feasible domain of parameters to obtain a set of parameter combinations, taking the median of the set as the optimal recommended value, and taking the boundary of the set as the allowable fluctuation range.

10. The intelligent optimization method for fastener production process according to claim 1, characterized in that, The real-time monitoring includes: real-time monitoring of key process parameters during production; issuing a yellow warning when a parameter exceeds the allowable fluctuation range but does not exceed 10% of the allowable fluctuation range; issuing a red warning and recommending shutdown for inspection when a parameter exceeds the allowable fluctuation range by more than 10% or deviates for more than 5 minutes.

Images