JB/T 7510-1994 Orthogonal test method for process parameter optimization
Some standard content:
Machinery Industry Standard of the People's Republic of China
JB/T7510-94
Process Parameter Optimization Method
Orthogonal Experiment Method
1994-12-09 Issued
Ministry of Machinery Industry of the People's Republic of China
1995-10-01 Implementation
Subject Content and Scope of Application
3 Procedures and Requirements for Orthogonal Experiment
Application of Orthogonal Experiment Method to Optimize Process Parameters (Conditions) Example 4
Appendix A Common Use orthogonal table format (supplement) Appendix B Interaction table and table header design of commonly used orthogonal tables (supplement)
Appendix c
Example of optimizing process parameters (process conditions) by orthogonal test method (reference) (1)
Mechanical Industry Standard of the People's Republic of China
Process parameter optimization method
Orthogonal test method
1 Subject content and applicable scope
This standard specifies the procedures and requirements for optimizing process parameters (process conditions) by orthogonal test method. JB/T751094
This standard is applicable to the optimization test of multi-factor process parameters (process conditions) and also to the optimization test of multi-factor parameters in product design.
2 Terminology
2.1 Orthogonal test method (abbreviated as orthogonal test) Orthogonal test method is a scientific method for studying multi-factor optimization test by applying the orthogonal principle of orthogonal table and mathematical statistical analysis. It can select the combination of the best parameters or conditions of each factor with the least number of tests. 2.2 Orthogonal table
Orthogonal table is a standardized table designed according to the orthogonal principle. Its symbol is L. (m), where L represents the orthogonal table: l represents the number of rows of the orthogonal table (the number of tests that can be arranged); y represents the number of columns of the orthogonal table (the maximum number of test factors that can be accommodated); y represents the number of levels of each test factor. Orthogonal tables with different levels are called mixed level orthogonal tables. Its symbol is L, (m×m). 2.3 Assessment index
The characteristic quantity used to measure the test results in the orthogonal test. There are two types of assessment indexes: quantitative index and qualitative index. Quantitative index is an index directly expressed in quantity, such as output, efficiency, size, strength, etc.; qualitative index is an index that cannot be directly expressed in quantity, such as color, feel, appearance, etc.
2.4 Factor
Factor is an element that affects the test assessment index. There are two types of factors: controllable factors and uncontrollable factors. Controllable factors refer to factors that can be artificially controlled and adjusted in the test, such as temperature, pressure, speed, material, cutting amount, etc. Uncontrollable factors refer to factors that cannot be artificially controlled and adjusted for the time being, such as slight vibration of machine tools, slight wear of cutting tools, etc. The factors selected in the orthogonal test must be controllable factors. 2.5 Level
Level is the grade of the value or state taken by the factor in the test. 2.6 Interaction
In the orthogonal test, if the effect of one factor on the assessment index is affected by the change of another factor, then the two factors are said to have an interaction. If there is an interaction between factor A and factor B, it is recorded as A×B. 3 Procedures and requirements for orthogonal tests
3.1 Determine the assessment indicators
3.1.1 The assessment indicators should be quantitative, and the others should be qualitative indicators. Test scoring standards should be formulated. Through scoring, qualitative indicators can be quantified for easy analysis and comparison.
3.1.2 When there are two or more assessment indicators, a comprehensive scoring method should be adopted to convert multiple indicators into single indicators for easy comprehensive evaluation. 3.2 Select factors and determine the level
3.2.1 When selecting factors, various factors that may affect the assessment indicators should be classified first, and then, based on experience, controllable factors that may have a significant impact should be selected as test factors. JB/T7510-94
3.2.2 The number of selected test factors should not be too many, and generally 3 to 7 is appropriate, so as not to increase the workload of invalid tests. If the expected purpose is not achieved after the first round of tests, the test factors can be adjusted based on the first round of tests and then the test can be conducted again. 3.2.3 The number of levels of the factors should not be too many, generally 2 to 4 is appropriate. The number of levels of each factor can be the same or different. More levels can be taken for important factors.
3.2.4 The values of each level should be appropriately separated to facilitate the analysis of the test results. 3.3 Selection of orthogonal tables
3.3.1 Types and formats of orthogonal tables
There are many types of orthogonal tables. The commonly used types and formats of orthogonal tables are shown in Table 1 and Appendix A (supplement). Table 1
Number of bit buffers
Two-bit level
Three-bit level
Multiple bit buffers
Mixed bit levels
Orthogonal table type
L,(2) , L,(2\)、Lu(215)
L(3\)、Lr(3*)
Lu(4*), Las(5), Lo(7)
L,(4X2*),L(3×2),Ls(4×2*)
L(4*×2*),Lu(4X2).Lu(4*X2*)
L(8×2\),L(2×3),Lu(6×3*),L(3X4X2\)3.3.2 Principles for selecting orthogonal arrays
See Appendix A Table A1, Table A2 and Table A4
See Appendix A Table A3 and Table A7|| tt||See Appendix A Table A5, Table A6 and Table A8
See Appendix A Table A9~Table A18
3.3.2.1 The number of columns of the orthogonal table must not be less than the number of experimental factors, and the number of levels of the orthogonal table must be the same as the number of levels of each factor. 3.3.2.2 When the factors and levels are the same, if the test accuracy is high, an orthogonal table with a large number of tests should be selected; if the test cost is high or the test cycle is long, an orthogonal table with a small number of tests should be selected as much as possible. 3.3.2.3 If the number of levels of each experimental factor is not equal, the corresponding mixed level orthogonal table should generally be selected. 3.3.2.4 If the interaction between experimental factors is considered, the orthogonal table should be selected according to the number of interaction factors and the interaction arrangement principle.
3.4 Table header design
3.4.1 Arrangement of experimental factors
3.4.1.1 When the number of experimental factors is equal to the number of columns in the orthogonal table, the factors with more difficult level changes should be placed in the first column, and the factors with easier level changes should be placed in the last column. The remaining factors can be arranged arbitrarily. 3.4.1.2 When the number of experimental factors is less than the number of columns in the orthogonal table, if there are empty columns in the table, if the interaction is not considered, the empty columns can be used as error columns. Their position is generally placed in the middle or at the back.
3.4.1.3 When the interaction between factors is considered, the arrangement of the interaction should be in accordance with the provisions of Tables B1 to B6 in Appendix B (Supplement). Tables B1 and B3 are the arrangement rules for the interaction columns of two-level orthogonal tables. The numbers in the top row and the rightmost column in the table represent the column numbers of the orthogonal table. The numbers at the intersection of the vertical and horizontal columns indicate the column where the interaction of the two columns of factors is located. Table B2 and Table B4 are the application of the rules of Table B1 and Table B3.
Table B5 is the arrangement rule of the interaction of the three-level orthogonal table. It is different from the arrangement of the interaction of the two-level orthogonal table in that the interaction of any two columns of factors must occupy two columns. Table B6 is the application of the rule of Table B5. Generally, for an m-level orthogonal table, the interaction between two columns of factors should occupy one column. Interactions should generally not be arranged in the same column as factors to avoid confusion. 3.4.2 Allocating test conditions
Allocating test conditions is to arrange the value or state of each level of each test factor according to the level number of the orthogonal table. In order to eliminate artificial systematic errors, when arranging the value or state of each level of the test factor, a random method can be used to determine it. 2
3.4.3 Design the test result data record column JB/T7510-94
According to the number of assessment index items and the analysis method used for the test results, draw the assessment index and calculation data record columns on the right and bottom of the orthogonal table respectively.
3.5 Conduct the test
3.5.1 When conducting the test, the test conditions of each number must be strictly followed. The order of the tests of each number cannot be changed at will. They can be arranged at will, and the order in the table is not required. 3.5.2 In the test, the test sequence of the orthogonal table must be strictly followed and each test result value must be correctly recorded in the corresponding assessment index column. 3.6 Calculate and analyze the test results
3.6.1 Calculate the sum K+ of the assessment index values of the same bit level in each column and record it in the corresponding column below the orthogonal table. K, the subscript i represents the bit level, and j represents the number of columns.
3.6.2 When using range analysis, the range R of the sum of the same level assessment indicators in each column should be calculated.. R, can be calculated as follows: For two-level factors
For three-level and multi-level factors
R,-Ku-K
R, -Ku.. Ku
The calculation results should be filled in the corresponding columns below the orthogonal table, (1)
··(2)
3.6.3When using variance analysis, the total value of the assessment index T, the sum of squares of the height differences of each level in each column S, and its degree of freedom f, the total sum of squares of deviations S and its degree of freedom f. and their calculation results should be filled in the corresponding columns below the orthogonal table. T, S, and S are calculated using the following formulas:
Where: yi—the assessment index value of the th test; yi—the number of tests.
In the formula: k, the sum of the values of the same level assessment indicators in each column; T—the sum of the values of the assessment indicators:
m-level;
\-number of tests;
j1.2.3,....
3.6.4 Use range analysis or variance analysis to find out the order of influence of each factor on the assessment indicators. 4
（6）
3.6.4.1 For general orthogonal tests, the order of influence of each factor on the assessment indicators can be sorted out by comparing the range R of the sum of the same level assessment indicators of each factor. The larger R, the greater the influence. 3.6.4.2 For orthogonal tests with strict test error requirements, in order to eliminate the influence of errors, variance analysis should be used to find out the significance of the influence of each factor and sort out the order of influence. 3.6.5 Determine the optimal level combination of each factor. In the absence of interaction, if the larger the assessment index, the better, the level with the largest K of each factor (for a mixed level orthogonal table, K should be divided by the number of levels) should be taken as the optimal level. Otherwise, the level with the smallest K (or K divided by the number of levels) should be taken as the optimal level. For factors with interaction, the optimal level can only be determined after interaction analysis [see Chapter C4 of Appendix C (reference)]
3.6.6 Draw a trend chart of the relationship between the level of each factor and the test result index as needed to find out the trend of index changes.
JB/T7510 94
3.6.7 To find a better combination of levels, you can select new levels of relevant factors according to the trend chart and conduct another test. 3. 7 Verification
The optimization scheme obtained through calculation and analysis is verified. Only those that meet the expected goals can be promoted and implemented in production. Examples of using orthogonal test method to optimize process parameters (conditions) Examples of using orthogonal test method to optimize process parameters (process conditions) are shown in Appendix C. JB/T7510-94
Appendix A
Common orthogonal table format
(Supplement)Www.bzxZ.net
Table A1L(2\)
Test number
Note: The interaction between any two columns is at most one. La(2')
Test number
Table A3L,(3*)
Test number
Note: The interaction between any two columns is the other two columns, JB/T7510-94
Table A4·L(2\)
Test number
Table A5L(45)
Test number
Note: The interaction between any two columns is the other three columns. 6
JB/T751094
Test number
Note: The interaction between any two columns is the other four columns,
Tip: This standard content only shows part of the intercepted content of the complete standard. If you need the complete standard, please go to the top to download the complete standard document for free.