Production process quality system risk assessment method for digital workshop

Through the multivariate statistical analysis and the Hotling statistical control chart, the accuracy of the digital workshop risk assessment is solved, effective evaluation and fault diagnosis of production process quality is achieved, and the accuracy of risk management and production process are improved. stability.

CN118982240APending Publication Date: 2024-11-19AEROSUN CORP
View PDF 0 Cites 2 Cited by

Patent Information

Application Number
CN202411068226.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-06
Publication Date
2024-11-19

AI Technical Summary

Technical Problem

Risk assessment in digital workshops is difficult to accurately control project risks, and traditional methods cannot effectively extract information from production process data, resulting in unmet fault diagnosis requirements.

Method used

Using multivariate statistical analysis method, the Hotling Statistics (T2) control chart is constructed, the risk level is determined and improvement measures are proposed. By collecting production process quality indicator parameters, using multivariate statistical analysis to obtain correlation indicator parameters, constructing T2 control charts, and analyzing Risk level and make recommendations for improvement.

Benefits of technology

It realizes risk assessment and fault diagnosis of the quality of the production process in the digital workshop, and improves the accuracy of risk management and the stability of the production process.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118982240A_ABST
    Figure CN118982240A_ABST
Patent Text Reader

Abstract

The invention discloses a production process quality system risk assessment method for a digital workshop, and the method comprises the steps: collecting index parameters affecting the production process quality, obtaining the index parameters of the production process quality of an enterprise production department according to the digital workshop, and analyzing the types and specifications of the parameters. A multi-dimensional vector is established for related index parameter data related to a product production process by using a multivariate statistical analysis method. A control chart is adopted to display the influence of the production process quality parameters on the production process quality, and the control upper limit in the control chart is calculated and determined through the data of the quality parameters. According to the control chart, risk existing points exceeding the control upper limit are found so as to determine the risk level, product quality rectification suggestions are provided, and production process quality management is completed. And informing an enterprise production department of a risk assessment result and improvement measures, improving the production efficiency, reducing the planning risk, and completing the risk assessment of the production process quality system of the digital workshop.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to a risk assessment method, in particular to a production process quality system risk assessment method for a digital workshop, and belongs to the technical field of flexible manufacturing. Background Art

[0002] The system risk assessment of the digital workshop is also called safety evaluation. Its main purpose is to ensure the normal production and equipment safety of the digital production line. With the development of intelligent manufacturing technology, the digital production line is also moving towards high automation and high integration. The increased complexity of the digital production line also increases the possibility of related failures. In the actual production process, due to the lack of a large number of actual case references and similar digital production line project risk management experience, the establishment of a risk assessment system is difficult, which leads to a large discrepancy between the risk assessment results of the digital workshop and the actual results of the digital workshop, which will eventually make the project risk difficult to control.

[0003] In the actual production process of a digital workshop, as the variables, processes, product performance and other requirements of the production process increase, the production process will become more complicated, so a large amount of production process data will be collected. However, traditional methods based on mechanism models or mathematical models can no longer extract effective information from the numerous production process data, and cannot meet the needs of fault diagnosis of the production process quality of the digital workshop. Summary of the invention

[0004] Purpose of the invention: In view of the problems and shortcomings in the prior art, the present invention provides a method for risk assessment of the production process quality system of a digital workshop, which completes the risk assessment analysis of the production process and ensures the normal production of the digital production line of the workshop.

[0005] Technical solution: A method for risk assessment of production process quality system in a digital workshop, including the following steps:

[0006] Step 1, collecting index parameters of production process quality;

[0007] Step 2: Use multivariate statistical analysis methods to obtain some index parameters that are strongly correlated with the quality of the production process;

[0008] Step 3: Construct Hotelling statistic (T 2 ) control chart;

[0009] Step 4: Determine the risk level, obtain risk assessment results based on the control chart, and propose improvement measures and suggestions for non-conforming products.

[0010] Step 1: Collect the index parameter data of the production process quality. According to the digital workshop, the index parameters of the production process quality are obtained. The index parameters of the production process quality include: transportation logistics, product storage, raw material quality, product design, product processing technology requirements, processing equipment, processing personnel, and processing environment related index parameter data.

[0011] Step 2: Analyze parameter types and specifications. First, use multivariate statistical analysis methods to analyze the relevant index parameter data involved in the industrial process. Use the index parameter data of the production process quality to establish a multidimensional vector, analyze the collected production process quality parameters, and obtain some index parameters that are strongly correlated with the production process quality among all the index parameters.

[0012] Step 2.1: Use multivariate statistical analysis to determine the characteristic vector y of the index parameter data of the production process quality under normal conditions i , where y i is a p-dimensional vector of index parameters of production process quality, and p relevant parameter data in M ​​samples are obtained in the present invention.

[0013] Step 2.2, obtain the index parameter data that is strongly correlated with the production process quality. Multivariate statistical analysis can map the index parameter data of the production process quality to a high-dimensional phase space, in which the strength of the correlation can be reflected. The collected index parameter data of the production process quality is recorded as sample data. The sample data is used to construct a vector y i , calculate the vector y i The sample mean of and the overall sample covariance S, i = 1, ..., M; based on the vector y i T 2 (i) The calculation formula is:

[0014]

[0015] Among the multiple production process quality indicator parameter data obtained, some parameter data may have no correlation with the production process quality, which makes T 2 The calculation of T is not in line with reality, so it is necessary to obtain the index parameter data that is strongly correlated with the quality of the production process so that T 2 The calculation is in line with reality. Therefore, the feature matrix Y M×p , the feature matrix Y M×p Transformed into a set of linearly uncorrelated principal components. First, the feature matrix Y M×p =[y1,y2,...,y M ] T Subtract the column mean to center it, that is Y * The singular value decomposition formula is:

[0016] Y * =UΨV T

[0017] where U and V are M×M and p×p orthogonal matrices, and Ψ is an M×p diagonal matrix with diagonal elements λ1≥λ2≥…≥λ p ≥0(i.e. Y * Therefore, the covariance matrix is ​​calculated as:

[0018] Y *T Y * =VΨU T UΨV T =VΨ 2 V T

[0019] Where VΨ2V T Yes *T Y * Therefore, the principal component calculation formula is:

[0020] Z=Y * V=UΨV T V=UΨ

[0021] The first principal component Z(:,1)=Y * v1 has the largest sample variance among all principal components. * The first principal component direction of . Therefore, the overall sample covariance S is calculated as:

[0022]

[0023] It should be noted that the principal components Z(:, i) and Z(:, j) are mutually orthogonal. Therefore, the covariance matrix S of the principal components is z is a diagonal term λ1 2 ≥λ2 2 ≥…≥λ p 2 The diagonal matrix of T. 2 (i) The calculation formula becomes:

[0024]

[0025] Where Z(i,:) is the projection of the sample in the direction of the principal component, that is, Furthermore, the score matrix Z can be obtained by considering only the first q eigenvalues ​​and eigenvectors (q < p) q The first q eigenvalues ​​and eigenvectors (q<p) are used to obtain the sub-matrix Z q Reflects the feature matrix Y M×pMost of the changes in , that is, the index parameter data represented by the first q eigenvalues ​​and eigenvectors are index parameter data that are strongly correlated with the quality of the production process. Therefore, the reduced dimension T can be obtained 2 for:

[0026] Step 3: Build T 2 Control charts;

[0027] Step 3.1, according to step 2, obtain the index parameter data that is strongly correlated with the quality of the production process, and then obtain the reduced dimension T from step 2 2 Calculate the new T 2 .

[0028] From step 2, we can get the dimension-reduced T 2 for: T for each sample 2 and its confidence interval can be plotted on a control chart, which characterizes the differences between the multivariate recursive characteristics of different samples. 2 The control chart shows the influence of the quality parameters of the production process on the quality of the production process. Through the data of the quality parameters, T is set. 2 The variables in the control chart are quality parameters.

[0029] Step 3.2, calculate and determine T 2 The confidence interval of the control chart is UCL, and the calculation formula of UCL is:

[0030]

[0031] Where M is the number of samples, α is the significance level of the F distribution, and F α,p,M-p is 100α% on an F-distribution with p and Mp degrees of freedom.

[0032] Step 4: By analyzing T 2 Control charts are used to determine the risk level and conduct risk assessment. 2 Find the risk point that exceeds UCL in the control chart and determine E as the risk assessment index of the production process quality system. The calculation formula of risk assessment index E is:

[0033]

[0034] Where I is T 2 The number of risk points in the control chart;

[0035] Further, determine the risk level and propose product quality improvement suggestions. The risk level is determined based on the production process quality system risk assessment indicators.

[0036] Low risk: The production process quality system risk assessment index E is in the range of [0, 0.25), and its risk level is major risk.

[0037] Medium risk: The production process quality system risk assessment index is in the range of E[0.25, 0.5), and its risk level is high risk.

[0038] High risk: The production process quality system risk assessment index E is in the range of [0.5, 0.75), and its risk level is medium risk.

[0039] Major risk: The production process quality system risk assessment index E is in the range of [0.75, 1], and its risk level is low risk.

[0040] Further risk assessment results were obtained through T 2 The corresponding risk points in the control chart can be used to obtain the time and link of the production process abnormality. The digital workshop can make timely improvements based on this information to complete the management of the production process quality of the digital workshop. The risk assessment results will be further informed to the production department of the enterprise to complete the risk assessment of the planning and scheduling system of the digital workshop. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 is a flow chart of risk assessment of production process quality system according to an embodiment of the present invention;

[0042] Figure 2 is a flow chart of risk assessment of quality system of ultra-precision machining surface polishing process of an embodiment;

[0043] Figure 3 is the quality T of the ultra-precision surface polishing process of the embodiment 2 Control charts; DETAILED DESCRIPTION

[0044] The present invention is further explained below in conjunction with specific embodiments. It should be understood that these embodiments are only used to illustrate the present invention and are not used to limit the scope of the present invention. After reading the present invention, various equivalent forms of modifications to the present invention by those skilled in the art all fall within the scope defined by the claims attached to this application.

[0045] As above Figure 2 As shown, the subject of the present invention is the quality of the ultra-precision machining surface polishing process. Real-time control of the ultra-precision machining process is crucial to ensuring the quality of surface flatness, improving manufacturing performance and increasing competitive advantages in the global market.

[0046] The method steps for risk assessment of quality system of ultra-precision surface polishing process are as follows:

[0047] Step 1, collect the index parameters of the ultra-precision machining surface polishing quality, and obtain the index parameters of the ultra-precision machining surface polishing quality according to the digital workshop. The ultra-precision machining surface polishing workpieces collected in the embodiment of the present invention are the same products processed in the workshop, and the collected index parameters of the ultra-precision machining surface polishing quality are: polishing trajectory, polishing pressure, polishing speed, polishing liquid properties, polishing pad properties, raw material product batch, workpiece surface roughness before polishing, workpiece surface straightness before polishing, workpiece surface verticality before polishing, different machine tools for polishing front stage process, and polishing environment temperature. Determine the number of samples M collected to be 120 samples.

[0048] Step 2, analyzing parameter types and specifications, firstly, using multivariate statistical analysis methods to establish a multidimensional vector for the relevant index parameter data that affect the quality of ultra-precision surface polishing, and then performing multivariate statistical analysis on the collected production process quality parameters to obtain index parameter data that is strongly correlated with the production process quality;

[0049] Step 2.1: Use multivariate statistical analysis to determine the characteristic vector y of the index parameter data of the production process quality under normal conditions i , where y i is a 10-dimensional vector of index parameters of the ultra-precision machining surface polishing quality of this embodiment, i=1, ..., 120. The index parameters represented by the 10-dimensional vector are polishing trajectory, polishing pressure, polishing speed, polishing liquid properties, polishing pad properties, raw material product batch, workpiece surface roughness before polishing, workpiece surface straightness before polishing, workpiece surface verticality before polishing, and polishing environment temperature. In the present invention, 10 relevant parameter data in 120 samples are obtained.

[0050] Step 2.2, obtain indicator parameter data that is strongly correlated with the quality of the production process.

[0051] Calculate the vector y i The sample mean of and the overall sample covariance S, based on y i Vector Calculation T 2 (i)

[0052]

[0053] Among the obtained multiple ultra-precision machining surface polishing quality index parameter data, some parameter data may have no correlation with the production process quality, resulting in the singularity of the covariance matrix, thus making T 2 The calculation of is not realistic. Therefore, we obtain 10 relevant parameter data from 120 samples to construct the feature matrix Y 120×10 . The feature matrix Y 120×10Transformed into a set of linearly uncorrelated principal components. First, the feature matrix Y 120×10 =[y1,y2,...,y 120 ] T Subtract the column mean to center it, that is Y * The singular value decomposition formula is:

[0054] Y * =UΨV T

[0055] where U and V are 120×120 and 10×10 orthogonal matrices, and Ψ is a 120×10 diagonal matrix with diagonal elements λ1≥λ2≥…≥λ 10 ≥0(i.e. Y * Therefore, the covariance matrix is ​​calculated as:

[0056] Y *T Y * =VΨU T UΨV T =VΨ 2 V T

[0057] Yes *T Y * Therefore, the principal component calculation formula is:

[0058] Z=Y * V=UΨV T V=UΨ

[0059] The first principal component Z(:,1)=Y * v1 has the largest sample variance among all principal components. * The first principal component direction of . Therefore, the overall sample covariance S is calculated as:

[0060]

[0061] It should be noted that the principal components Z(:, i) and Z(:, j) are mutually orthogonal. Therefore, the covariance matrix S of the principal components is z is a diagonal term λ1 2 ≥λ2 2 ≥...≥λ 10 2 The diagonal matrix of T. 2 The calculation formula becomes:

[0062]

[0063] Where Z(i,:) is the projection of the sample in the direction of the principal component, that is, In addition, in this embodiment, only the first four eigenvalues ​​and eigenvectors (4 < 10) are considered to obtain the sub-matrix Z4, where the index parameters represented by the first four eigenvalues ​​and eigenvectors are polishing trajectory, polishing pressure, polishing speed, and polishing liquid properties. This explains the characteristic matrix Y 120×10 Most of the changes in , that is, the index parameter data represented by the first four eigenvalues ​​and eigenvectors are index parameter data that are strongly correlated with the quality of the production process. Therefore, the reduced dimension T can be obtained 2 for:

[0064] Step 3: Build T 2 Control charts;

[0065] According to the calculated T 2 To construct Figure 3 T 2 Control chart; using T 2 The control chart shows the influence of relevant influencing factors on the polishing quality of ultra-precision machined surfaces.

[0066] Step 3.1, according to step 2, obtain the index parameter data that is strongly correlated with the quality of the production process, and then obtain the reduced dimension T from step 2 2 Calculate the new T 2 .

[0067] From step 2, we can get the dimension-reduced T 2 for:

[0068] Step 3.2, calculate and determine T 2 The confidence interval of the control chart is UCL, and the calculation formula of UCL is:

[0069]

[0070] The significance level α of the embodiment of the present invention is 0.05;

[0071] Step 4: Use T 2 Control charts determine risk levels and conduct risk assessments. 2 The number of risk points exceeding UCL in the control chart is 4.

[0072]

[0073] Furthermore, the risk level of the evaluation indicator is defined as low risk.

[0074] Further risk assessment results were obtained through T 2The corresponding risk points in the control chart can be used to obtain the time and link of the production process abnormality. The digital workshop can make timely improvements based on this information to complete the management of the production process quality of the digital workshop. The risk assessment results will be further informed to the production department of the enterprise to complete the risk assessment of the planning and scheduling system of the digital workshop.

Claims

1. A method for risk assessment of production process quality system in a digital workshop, characterized in that: The following steps are involved: Step 1, collecting index parameters of production process quality; Step 2, using multivariate statistical analysis methods to analyze the index parameters that have a strong correlation with the quality of the production process; Step 3: Construct Hotelling statistic (T 2 ) control chart; Step 4: Determine the risk level and obtain the risk assessment results based on the control chart.

2. The method for risk assessment of production process quality system of a digital workshop according to claim 1 is characterized in that: In the step 1, the index parameters of the production process quality are obtained according to the digital workshop, and the index parameters of the production process quality include: transportation logistics, product storage, raw material quality, product design, product processing technology requirements, processing equipment, processing personnel, and processing environment related index parameter data.

3. The method for risk assessment of production process quality system of a digital workshop according to claim 1 is characterized in that: In step 2, the parameter types and specifications are analyzed. First, the relevant indicator parameter data involved in the industrial process are analyzed using multivariate statistical analysis methods. A multidimensional vector is established using the indicator parameter data of the production process quality. The collected production process quality parameters are analyzed, and some indicator parameters that are strongly correlated with the production process quality are obtained from all indicator parameters.

4. The method for risk assessment of production process quality system of a digital workshop according to claim 1 is characterized in that: The implementation process of step 2 is: Step 2.1: Use multivariate statistical analysis to determine the characteristic vector y of the index parameter data of the production process quality under normal conditions i , where y i is the p-dimensional vector of index parameters of production process quality; Step 2.2, obtain the index parameter data that is strongly correlated with the production process quality; record the collected index parameter data of the production process quality as sample data; use the sample data to construct the vector y i , calculate the vector y i The sample mean of and the overall sample covariance S, i = 1, ..., M; based on the vector y i T 2 (i) The calculation formula is: Construct feature matrix Y M×p , the feature matrix Y M×p Transformed into a set of linearly uncorrelated principal components; feature matrix Y M×p =[y1,y2,…,y M ] T Subtract the column mean to center it, that is Y * The singular value decomposition formula is: Y * =UΨV T where U and V are M×M and p×p orthogonal matrices, and Ψ is an M×p diagonal matrix with diagonal elements λ1≥λ2≥…≥λ p ≥0, the covariance matrix calculation formula is: Y *T Y * =UΨU T UV T =VΨ 2 V T Where VΨ 2 V T Yes *T Y * The eigenvalue decomposition and principal component calculation formula are: Z=Y * V=UΨV T V=UΨ The first principal component Z(:,1)=Y * v1 has the largest sample variance among all principal components, where the eigenvector v1 is Y * The first principal component direction of the overall sample covariance S is calculated as: The principal components Z(:,i) and Z(:,j) are mutually orthogonal, so the covariance matrix S of the principal components z is a diagonal term λ1 2 ≥λ2 2 ≥…≥λ p 2 The diagonal matrix, T 2 (i) The calculation formula becomes: where Z(i, :) is the projection of the sample in the direction of the principal component, that is In addition, the score matrix Z can be obtained by only considering the first q eigenvalues and eigenvectors q , q < p, the first q eigenvalues and eigenvectors are used to obtain the score matrix Z q which reflects the feature matrix Y M×p Most of the variations, that is, the index parameter data represented by the first q eigenvalues and eigenvectors are the index parameter data strongly correlated with the quality of the production process; Therefore, the reduced-dimensional T 2 is as follows 5. The method for risk assessment of production process quality system of a digital workshop according to claim 1 is characterized in that: Step 3, construct T 2 Control chart, the specific implementation process is: Step 3.1, obtain the reduced dimension T 2 , T of each sample 2 and its confidence interval are plotted on a control chart using T 2 The control chart shows the influence of production process quality parameters on production process quality. Set T 2 The variables in the control chart are quality parameters; Step 3.2, calculate and determine T 2 The confidence interval of the control chart is UCL, and the calculation formula of UCL is: Where M is the number of samples, α is the significance level of the F distribution, and F α,p,M-p is 100α% on an F-distribution with p and Mp degrees of freedom.

6. The method for risk assessment of production process quality system of a digital workshop according to claim 1 is characterized in that: In step 4, by analyzing T 2 Control chart to determine the risk level and conduct risk assessment; first in T 2 Find the risk point exceeding UCL in the control chart and determine E as the risk assessment index of the production process quality system; the calculation formula of risk assessment index E is: Where I is T 2 The number of risk points in the control chart.

7. The method for risk assessment of production process quality system of a digital workshop according to claim 1 is characterized in that: Determine the risk level based on the production process quality system risk assessment indicators; Low risk: The production process quality system risk assessment index E is in the range of [0,0.25); Medium risk: The production process quality system risk assessment index E is in the range of [0.25, 0.5); High risk: The production process quality system risk assessment index E is in the range of [0.5, 0.75); Major risk: The production process quality system risk assessment index E is in the range of [0.75,1].

8. The method for risk assessment of production process quality system of a digital workshop according to claim 1 is characterized in that: By T 2 The corresponding risk points in the control chart can be used to obtain the time and link of the production process abnormality. The digital workshop can make timely improvements based on this information to complete the management of the production process quality of the digital workshop.

Citation Information

Cited By

  • Vehicle production quality risk multi-source data collaborative evaluation full-chain optimization method

    CN120509746A

  • Vehicle production quality risk multi-source data collaborative evaluation full-chain optimization method

    CN120509746B