A method for generating tool life DOE analysis response and a method for tool life regression analysis based on workpiece appearance grading.

Abstract

This invention discloses a method for generating a tool life (DOE) analysis response based on workpiece appearance grading and a tool life regression analysis method. At each rated cutting distance, workpiece appearance images are acquired and graded according to their levels, with one image corresponding to one level. Each level is then assigned a preset numerical value. For each rated cutting distance, the corresponding distance value is multiplied by the numerical value corresponding to the appearance image, resulting in a data product. The m data products are summed, and the sum is used as the DOE analysis response. This invention utilizes workpiece appearance grading and performs data processing, weighting, and summing to obtain the final result as the DOE analysis response. Furthermore, by acquiring images throughout the entire tool cutting process, it can comprehensively solve for the optimal X (tool design factor) combination over the entire tool life cycle.

Description

Technical Field

[0001] This invention relates to tool life (DOE) analysis, specifically a tool life DOE analysis response generation method and a tool life regression analysis method based on workpiece appearance grading. Background Technology

[0002] In tool life analysis (DOE) where the customer's workpiece is assessed based on the surface quality grading, the conventional DOE evaluation method involved evaluating and collecting the machined part (hereinafter referred to as workpiece appearance) as the DOE response for analysis: Figure 1 This is a schematic diagram of the existing technology of appearance classification through image. The appearance of the workpiece is compared by appearance inspection (magnified 50-200X in Keyence) and divided into: excellent, medium and poor.

[0003] The drawback of this analytical method is: The classification description of workpiece appearance belongs to logical variables. When more than three logical variables are used as responses, it is difficult to have a good regression model for analysis: generally, binary logistic regression is used.

[0004] It is difficult to incorporate the responses of the points collected over the entire cutting length into the analysis, making it difficult to analyze the responses of the points collected over the entire life cycle (cutting length).

[0005] If the wear value is obtained at a specific cutting distance (such as the rated life cutting distance of 100m), this logistic variable response can only be modeled using a logistic regression model, which is not a good model. Figure 2 This is a schematic diagram of the establishment of a three-category logistic regression model in the existing technology, indicating that all the parameters included in this analysis are not stable. Summary of the Invention

[0006] To address the shortcomings of the existing technologies, this invention provides a tool life DOE analysis response generation method and a tool life regression analysis method based on workpiece appearance grading. This invention utilizes workpiece appearance grades and performs data processing, then performs weighted summation, and the final result serves as the response for DOE analysis. Furthermore, images are acquired throughout the entire tool cutting process, enabling a relatively complete solution to obtain the optimal X (tool design factor) combination over the entire life cycle.

[0007] To achieve the above technical objectives, the present invention adopts the following technical solution: a method for generating tool life (DOE) analysis response based on workpiece appearance grading, comprising the following steps: Determine m rated cutting distances, where m is a positive integer greater than or equal to 1; determine n grades from best to worst, where n is a positive integer greater than or equal to 1; At each of the rated cutting distances, an appearance image of the workpiece is acquired, and the appearance images are classified according to the grade, with one appearance image corresponding to one grade; The levels are each assigned a preset numerical value; For each of the rated cutting distances, the corresponding distance value is multiplied by the numerical value corresponding to the appearance image to obtain a data product; The products of the m data points are multiplied and summed, and the sum is used as the response of the DOE analysis.

[0008] The preset values ​​corresponding to the levels decrease sequentially from excellent to poor.

[0009] n=3, and the three levels are excellent, medium and poor. The preset value for excellent is 100, the preset value for medium is 50, and the preset value for poor is 0.

[0010] m=5, and the rated cutting distances of the five are 20 meters, 40 meters, 60 meters, 80 meters, and 100 meters, respectively.

[0011] A tool life regression analysis method is provided, wherein the tool life regression analysis method uses the sum of the products of m data as the response variable.

[0012] It also includes the following steps: Select JMP for simulation fitting; Model building; Model fitting and evaluation; Find the optimal solution.

[0013] The construction of the fitting model includes the following steps: When selecting a role variable, select the sum of the products of m data points at point Y; Select the X factor at the effect in the model and construct it using a response surface; Click the Run button to start running the model.

[0014] The fitting model processing and judgment includes the following steps: Delete the insignificant items one by one; Model judgment and testing.

[0015] The optimal solution is obtained by the following steps: Set the sum of the products of m data to be as large as possible; Maximize the desired outcome to obtain the optimal solution.

[0016] In summary, the present invention has achieved the following technical effects: This invention classifies the workpiece appearance surface based on planned and defined acquisition points throughout the entire rated life (cutting distance), quantifies the grade data, and sums them after weighting to obtain the response Y of the tool life (DOE). Regression analysis is then performed to achieve the optimal solution for the X factor combination of workpiece appearance surface quality over the entire life cycle. This invention avoids the problem of difficulty in modeling logistic regression of workpiece appearance surface hierarchical logical variables in traditional methods; This invention avoids the problem of traditional logic modeling, which is difficult to implement the superposition of appearance-level logic variables throughout the entire life cycle (at the cutting meter collection point), and thus cannot obtain the optimal solution for the entire life cycle evaluation. This invention achieves a more ideal and accurate evaluation throughout the entire life cycle, resulting in the optimal solution (the combination of optimal tool design factors X). Attached Figure Description

[0017] Figure 1 This is a schematic diagram illustrating the appearance of existing technologies through image grading; Figure 2 This is a schematic diagram illustrating the establishment of a three-category logistic regression model using existing technologies; Figure 3 This is a schematic diagram of the process provided by the present invention; Figure 4 yes Figure 4 This is a schematic diagram illustrating the quantification of workpiece appearance grade data; Figure 5 This is a schematic diagram illustrating the removal of insignificant terms in DOE regression analysis; Figure 6 This is a graph showing the predicted values ​​versus actual values ​​from a DOE regression analysis. Figure 7 This is a schematic diagram of the fit summary and variance analysis of DOE regression analysis; Figure 8 These are the parameter estimates from the DOE regression analysis; Figure 9 This is the predictor profiler interface for DOE regression analysis. Detailed Implementation

[0018] The present invention will be further described in detail below with reference to the accompanying drawings.

[0019] This specific embodiment is merely an explanation of the present invention and is not intended to limit the invention. After reading this specification, those skilled in the art can make modifications to this embodiment without contributing any inventive step, but such modifications are protected by patent law as long as they are within the scope of the claims of the present invention.

[0020] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," "counterclockwise," "axial," "radial," and "circumferential" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0021] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0022] In this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0023] In this invention, unless otherwise explicitly specified and limited, "above" or "below" the second feature can mean that the first feature is in direct contact with the second feature, or that the first feature is in indirect contact with the second feature through an intermediate medium. Furthermore, "above," "over," and "on top" of the second feature can mean that the first feature is directly above or diagonally above the second feature, or simply that the first feature is at a higher horizontal level than the second feature. "Below," "below," and "under" the second feature can mean that the first feature is directly below or diagonally below the second feature, or simply that the first feature is at a lower horizontal level than the second feature.

[0024] Example: like Figure 3 As shown, a method for generating tool life (DOE) analysis response based on workpiece appearance grading includes the following steps: Determine m rated cutting distances, where m is a positive integer greater than or equal to 1; determine n grades from best to worst, where n is a positive integer greater than or equal to 1; At each rated cutting distance, acquire the appearance image of the workpiece, and classify the appearance images according to the grade, with one appearance image corresponding to one grade; The levels are mapped to preset values ​​to quantify the level data, transforming it from qualitative to quantitative analysis, which is beneficial for subsequent DOE analysis. For each rated cutting distance, the corresponding distance value is multiplied by the value corresponding to the appearance image to obtain a data product. In this step, a weighted processing method is used to multiply the rated cutting distance by the grade data, which enables the analysis of tool life throughout the entire cutting process. The products of m data points are multiplied and summed, and the sum is used as the response to the DOE analysis. In this step, a weighted summation method is used as the response Y of the DOE analysis, which transforms the original qualitative data into a quantitative one that can be applied to DOE analysis.

[0025] In this invention, the preset values ​​corresponding to the grades decrease sequentially from excellent to poor.

[0026] In this embodiment, n=3, and the three levels are excellent, medium and poor. The preset value corresponding to excellent is 100, the preset value corresponding to medium is 50, and the preset value corresponding to poor is 0.

[0027] In other embodiments, n can be 6, and the levels from best to worst are 1, 2 to 6, with corresponding preset values ​​of 100, 80, 60, 40, 20, and 0.

[0028] Users use cutting tools to process workpieces. After processing, they take photos of the workpiece's appearance. Analysis of these photos determines the workpiece's appearance to be rated as excellent, average, or poor. For example, a user purchases cutting tools from our company and uses them to process the surface of a mobile phone casing. After processing, an image of the phone casing's appearance is captured using a machine vision system, such as Keyence's, and the image is categorized into excellent, average, and poor grades based on certain conditions. Our company uses the user's feedback grades, performs a series of processing steps, and uses this as the response for DOE analysis, subsequently analyzing tool life. Most commonly, the workpiece appearance grade can be obtained from the user. However, to facilitate our DOE regression analysis, we have developed a user system with equivalent requirements. This allows us to directly use our cutting tools to process workpieces. The purpose of this user system is not to process workpieces, but rather to obtain the appearance grade of the workpiece after processing with our cutting tools. This facilitates the acquisition of the workpiece appearance grade, which is then used to assist in analyzing tool life. It is important to understand that this system is not the inventive point of this invention; it is merely a means to more easily obtain the grade. Meanwhile, the grading conditions are set by the user, and our company uses this grading system to assist in analyzing the tool life. Similarly, the grading conditions are not the point of creation of this invention.

[0029] This invention quantifies the grade data of workpiece appearance, shifting from a qualitative approach to judging the quality of workpiece appearance to a quantitative approach to analyzing tool life. Furthermore, by setting multiple cutting distances, this invention can provide feedback on tool wear information throughout the entire cutting process, rather than relying solely on the final appearance image after cutting (machining completion). This effectively eliminates errors and improves the accuracy of tool life analysis.

[0030] In this embodiment, m=5, and the five rated cutting distances are 20 meters, 40 meters, 60 meters, 80 meters, and 100 meters, respectively. When the tool is machining the surface of the workpiece, an image of the workpiece appearance is acquired every time a rated cutting distance is reached, and the image is graded.

[0031] Figure 4 This diagram illustrates the digitization of workpiece appearance grade data. In the diagram, "Version Scheme" represents different cutting tools, "X-Factor" represents tool information, 20 meters, 40 meters, 60 meters, 80 meters, and 100 meters represent five rated cutting distances, and "Y Response - Workpiece Appearance Grade (Machining Meters)" indicates the grade corresponding to the appearance image at different rated cutting distances. For example, scheme version A01 has five appearance images at five rated cutting distances with grades of: Poor, Poor, Poor, Poor, Poor, Poor; scheme version A03 has five appearance images at five rated cutting distances with grades of: Medium, Medium, Medium, Poor, Poor, Poor. "Y Response - Workpiece Appearance Grade" "Datafication" means converting the grades into preset values. For example, "poor, poor, poor, poor, poor, poor" in scheme version A01 is converted to 0, 0, 0, 0, 0, and "medium, medium, medium, poor, poor" in scheme version A03 is converted to 50, 50, 50, 0, 0. "Y-Response - Workpiece Appearance Grading Datafication Weighted Summation" means multiplying the preset value by the corresponding rated cutting distance. For example, the "Y-Response - Workpiece Appearance Grading Datafication Weighted Summation" in scheme version A01 is 0, which is the result of multiplying "0, 0, 0, 0, 0" by "20 meters, 40 meters, 60 meters, 80 meters, 100 meters" respectively and then adding them together. Specifically, 0 × 20 + 0×40 + 0×60 + 0×80 + 0×100 = 0. The "Y Response - Workpiece Appearance Grading Data Weighted Sum" in scheme version A03 is 6000, which is the sum of the products of "50, 50, 50, 0, 0" and "20 meters, 40 meters, 60 meters, 80 meters, 100 meters" respectively. Specifically, it is 50×20 + 50×40 + 50×60 + 0×80 + 0×100 = 6000. The result of these multiplications and sums is used as the DOE response.

[0032] This invention digitizes user-defined workpiece appearance grading data. The three-category grading of workpiece appearance is a logistic variable. In traditional logistic regression, using three-category grading to build a regression model fails; moreover, three-category grading cannot achieve optimal solution over the entire lifespan (analysis of all data at the cutting distance acquisition points). This invention, by digitizing the data, solves the problems of logistic variables and logistic regression modeling.

[0033] This invention acquires surface appearance images at different cutting distances and performs grading and data processing. Since the grading data of the workpiece surface is a large item in the DOE (Tool Execution Environment) output response, the longer the cutting distance, the larger the grading data of the workpiece surface, and the better the tool life performance. This invention can quantitatively analyze tool life by using workpiece appearance grading, and acquires data at different cutting distances, which is consistent with the principles of tool life measurement.

[0034] This invention digitizes the workpiece surface appearance classification data at various sampling points at different cutting distances (in meters) throughout the life cycle, multiplies it by the cutting distance in meters, and sums all the data to achieve scheme evaluation over the entire life cycle. The number of sampling points at different cutting distances depends on the cutting characteristics and wear curve of this type of tool. To avoid the impact of differences in the number of sampling points on scheme evaluation, the following approach is typically used in actual DOE processes: Before the formal DOE scheme is implemented, the wear curve of the existing scheme (scheme requiring improvement) is investigated (while considering testing costs, as many data points as possible are collected to obtain an accurate wear curve, and the number of sampling points is defined according to the characteristics of the wear curve); after the definition is completed, the number of sampling points should not be changed during the entire DOE testing phase unless necessary, to ensure the effectiveness of scheme evaluation and DOE analysis.

[0035] This invention replaces the traditional method of using a single rated cutting distance (e.g., 100 meters) as the response Y of the workpiece surface grading data as a weighted sum for regression fitting model analysis. This method effectively improves the quality of the model, solves the modeling problem, and can achieve the optimal combination of X (tool design factors) over the entire life cycle.

[0036] A tool life regression analysis method is proposed, which uses the sum of the products of m data points as the response variable.

[0037] It also includes the following steps: Choose JMP for simulation fitting; specifically, choose the least squares method: Model building includes: when selecting role variables, choosing the sum of the products of m data points at Y; selecting the X factor and constructing it using a response surface at the model effect construction point; and clicking the run button to start running the model. Model fitting and evaluation; this includes: sequentially deleting insignificant terms, Figure 5 This is a diagram illustrating the removal of insignificant terms; model judgment and testing. Figure 6 This is a predicted-actual value plot, representing the dispersion of predicted and actual values ​​in the workpiece appearance grading data during the model judgment and verification stages. The p-value = 0.0399 < 0.05, indicating a good correlation between the predicted and actual values. Figure 7 This is a diagram illustrating the fit summary and analysis of variance. Adjusted R-squared represents "adjusted R-squared". 2 "R said "R 2 Adjust R 2 With R 2 The difference is a factor in measuring whether overfitting exists. When the difference is large, overfitting exists. In this embodiment, R is adjusted. 2 =0.651797 and R 2 =0.878129. Compared with the original scheme that could not be modeled, the model of this application is valid and significantly optimized. At the same time, the sum of squares and mean square of the errors are 241908314 and 34558331, respectively, indicating that the error term accounts for a low proportion and the model is well constructed. Figure 8 These are parameter estimates. VIF stands for variance inflation factor, which measures the degree of multicollinearity. Two VIF values ​​are slightly greater than 5, while the rest are less than 5, indicating that there is non-severe multicollinearity and that the model in this application is well constructed.

[0038] Finding the optimal solution includes: setting the sum of the products of m data points to be as large as possible; Figure 9 This is the predictive profiler interface for DOE regression analysis. "Workpiece appearance grading data weighted sum" represents the sum of the products of m data points, with its value falling between the lower and upper limits of the confidence interval. The confidence interval is the estimated range of the population parameter based on sample data at a certain confidence level, indicating "we have a certain degree of confidence that the population parameter falls within this interval." The VB weighted sum in this application falls within this interval, proving the feasibility of the proposed solution. Maximizing the desired outcome, the optimal solution is obtained, where the workpiece appearance grading data weighted sum is the largest (very large, without physical meaning, but close to the best value). The optimal solution for the system output factor combination is: rake angle 1: 26.7°; radial-clearance angle 1.26°; axial-clearance angle 6°; passivation value 8.45µm; flank face Ra 0.163µm; core thickness 3.84mm; helix angle 45.04°.

[0039] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Any simple modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention shall fall within the scope of the technical solution of the present invention.

Claims

1. A method for generating tool life (DOE) analysis response based on workpiece appearance grading, characterized in that, Includes the following steps: Determine m rated cutting distances, where m is a positive integer greater than or equal to 1; determine n grades from best to worst, where n is a positive integer greater than or equal to 1; At each of the rated cutting distances, an appearance image of the workpiece is acquired, and the appearance images are classified according to the grade, with one appearance image corresponding to one grade; The levels are each assigned a preset numerical value; For each of the rated cutting distances, the corresponding distance value is multiplied by the numerical value corresponding to the appearance image to obtain a data product; The products of the m data points are multiplied and summed, and the sum is used as the response of the DOE analysis.

2. The method for generating tool life (DOE) analysis response based on workpiece appearance grading according to claim 1, characterized in that, The preset values ​​corresponding to the levels decrease sequentially from excellent to poor.

3. The method for generating a tool life (DOE) analysis response based on workpiece appearance grading according to claim 1, characterized in that, n=3, and the three levels are excellent, medium and poor. The preset value for excellent is 100, the preset value for medium is 50, and the preset value for poor is 0.

4. The method for generating a tool life (DOE) analysis response based on workpiece appearance grading according to claim 1, characterized in that, m=5, and the rated cutting distances of the five are 20 meters, 40 meters, 60 meters, 80 meters, and 100 meters, respectively.

5. A method for regression analysis of tool life, characterized in that, The tool life regression analysis method uses the sum of the products of m data as described in any one of claims 1-4 as the response variable.

6. The tool life regression analysis method according to claim 5, characterized in that, It also includes the following steps: Select JMP for simulation fitting; Model building; Model fitting and evaluation; Find the optimal solution.

7. The tool life regression analysis method according to claim 6, characterized in that, The construction of the fitting model includes the following steps: When selecting a role variable, select the sum of the products of m data points at point Y; Select the X factor at the model effect and construct it using a response surface; Click the Run button to start running the model.

8. The tool life regression analysis method according to claim 7, characterized in that, The fitting model processing and judgment includes the following steps: Delete the insignificant items one by one; Model judgment and testing.

9. The tool life regression analysis method according to claim 8, characterized in that, The optimal solution is obtained by the following steps: Set the sum of the products of m data to be as large as possible; Maximize the desired outcome to obtain the optimal solution.

Images