Orthogonal test method is also called orthogonal test design method, which uses“Orthogonal table”To arrange and analyze multifactor experimentsstatistical method。The advantages of this method are that the number of tests is small, the effect is good, the method is simple, the use is convenient, and the efficiency is high.In the study of more complex problems, there are often many factors.We will examine the relevant impactsTest indexThe conditions of are called factors, such as the components in the formula.The different states of various factors to be investigated in the test are called levels, such as the different contents (or proportions) of a certain component in the formula test.In order to seek the optimalProduction conditionsIt is necessary to test various factors and different levels of various factors, which is the problem of multi factor optimization.[1]
Chinese name
Orthogonal test method
Meaning
A Design Method for Studying Multiple Factors and Levels
Orthogonal test method refers to a method of arranging and organizing testsscientific method。It uses a set of standardized tables, namelyOrthogonal tableTo designTest planAnd analyze the test results, and can select a few from many test conditionsRepresentativenessStrong test conditions, and through the data of these tests, find a betterProduction conditionsThat is, the optimal or better scheme.
The orthogonal test method is actuallyoptimizationOne of.Due to the rich content of orthogonal test method, it can not only solve multiple factorsOptimization problemIt can also be used to analyze the influence of various factors on the test results, so as to grasp the main factors.Therefore, it has been separated from the optimization method and formed its own system.
Orthogonal tableIs a set of regular design tables, L is the code of the orthogonal table, n is the number of tests, and t isHorizontal number, c is the number of columns, that is, the number of factors that may be arranged most.
nature
(1) In each column, different numbers appear the same number of times.
(2) The numbers in any two columns are arranged in a complete and balanced way.
The above two points fully reflectOrthogonal tableThe two advantages ofDispersity, neat and comparable ".Generally speaking, each level of each factor meets each level of another factor once, which isOrthogonality。
step
Announce
edit
1. On the basis of investigation and research, the test subject is determined according to the key problems to be solved in scientific research and production practice.
2. Based on practical experience andtheoretical analysis And relatedInformation, analyze various factors that may affect the test results, find out the main factors from them, and determine the variation range of the main factors.
3. According to the specific characteristics of the test subject, the appropriate optimization method is selected.
4. Arrange according to the selected preferred methodTest planAnd operate strictly according to the test conditions to accurately determine the test results.
5. The test results are compared and analyzed to determineOptimal scheme。
After completing the test and collecting data, range analysis (also known asvariance analysis)。Range analysis is to consider that the influence of other factors on the results is balanced when factor A is considered, so that the difference of various levels of factor A is caused by factor A.
Analysis by range methodorthogonal test The results should lead to the following conclusions:
① Within the test scope, each pair ofTest indexThe impact of queuing from large to small.The maximum range of a column means that when the value of the column changes within the test range, the test index value changes the most.Therefore, the queue with the largest to smallest impact of each column on the test indicators is the queue with the largest to smallest value of the range D of each column.
② The change trend of test indexes with various factors.
③ Make the test index best and suitableOperating conditions(Appropriate factor level collocation).
④ The conclusions and further research directions are discussed.
Selection of better conditions
Announce
edit
Theoretically, if each factor is not affected by the level change of other factors, then simply combining the superior levels of each factor is a better test condition.However, in fact, when selecting better production conditions, we should also consider the primary and secondary factors, so that under the same conditions that meet the indicator requirements, some minor factors can be considered as high quality, high yieldLow consumptionThe principle of selecting the level is to obtain better production conditions that are more consistent with the actual requirements of the test.
Analytical method
Announce
edit
Direct comparison method
Direct comparison methodIt is a simple and direct comparison of test results.Although the direct comparison method gives a certain explanation of the test results, this explanation is qualitative and cannot tell us the best composition combination.Obviously, although this analysis method is simple, it is not satisfactory.
Visual analysis
visualanalytical methodIt is to analyze the problem through the average range of each factor.The so-called extreme difference is the average effectMaximumandminimum valueThe difference of.With extreme poor, we can find the main factors that affect the indicators and help us find the best combination of factor levels