The hottest variable retrieval method is a method

  • Detail

Variable retrieval method -- a method to improve product quality

introduction in industrial enterprise management, product quality management is a very important problem. In order to improve product quality, people have put forward many methods, such as value engineering, failure mode effect analysis, fail safe, etc. Among various quality improvement techniques, experimental design method (D.O.E) is an important one, which was first proposed by Fisher. Through the joint efforts of many people, this method has been continuously improved and perfected. Dorian shanin gradually formed his own complete system, such as polytropic graph method, component retrieval method, variable retrieval technology and so on. After the product is produced, its performance needs to be tested. In this process, we will encounter unstable indicators. We can solve this problem by using the above D.O.E method. Variable retrieval technology is an element of D.O.E of Dorian * shaning style. Its main purpose is to find out the variables that affect the main causes of product quality by designing a series of experiments. When there are 5 ~ 15 variables controlling a process, this technology is often used. Next, variable retrieval technology is used to solve a problem in quality control. 1. Basic principle of variable retrieval technology. Excessive fluctuation of product quality will undoubtedly seriously affect production. To find out the real reason, we can directly look for it from the variables that affect the product, that is, the so-called "variable retrieval technology". The variables that affect the main causes of product quality are called "strong factors" and "weak factors" respectively

The prerequisite for the application of variable retrieval method is that there are 5 ~ 15 variables controlling a process, and the characteristic parameters of products must be continuous index variables. The details are as follows:

(1) if the variables in order of importance are a, B, C, D, e, F,..., and the characteristic parameter of the product is y, that is, y=y (a, B, C, D, e, f,...), it determines the range of variables a, B, C, D, e, F,..., a changes from XA to X 'a, B changes from XB to X' B,..., etc

(2) each variable corresponds to two cases, X good GA, GB, GC,... And X bad Ba, BB, BC

(3) do two experiments, a good experiment and a bad experiment, so as to measure their results yg=y (GA, GB, causing small piston damage and even oil leakage GC...), yb=y (BA, BB, BC...)

test again twice and measure the experimental error. The results of these three experiments are as follows: yg1, YG2, YG3 and YB1, YB2, Yb3. If we think these three good results are better than the three bad results, the probability of catching the strong and weak factors is 95%

(4) check the reproducibility. That is, the inspection is done, d= [(yg1+yg2) - (yb1+yb2)]/2, d= [(yg1-yg2) + (yb1-yb2)]/2 (d represents the difference between the average value of good results and the average value of bad results; D represents the average value of the difference between two good results and two bad results)

in fact, we have done three experiments, so there are three groups of D ∶ D, namely D1 ∶ D1, D2 ∶ D2 and D3 ∶ D3. The reproducibility requires that their absolute values are greater than or equal to 5 ∶ 1

(5) variable retrieval is only applicable to the situation of reproducibility, otherwise this road is blocked. That is, the strong and weak factors are not in a, B, C, D, e, F, or GA and Ba are reversed, gf and BF are reversed, and so on

after the reproducibility is confirmed, the specific methods of variable retrieval are as follows:

step 1: do an experiment, use abrg and agrb to get the following results y (abrg) and Y (agrb). There are several possible results: (RG indicates that the remaining variables are good characteristic parameters, Rb indicates that the remaining variables are bad characteristic parameters)

(1) y (abrg) is not much different from the previous YG, then a is not the main, that is, a is not a strong factor, Nor is it a weak factor or a sub weak factor

(2) y (abrg) is similar to Yb, then a is a strong factor

(3) sometimes there is a partial difference between Y (abrg) and YG of the hydraulic pump suction system, then a is a weak factor or sub weak factor, and there must be other weak factors. This experiment can be proved by Y (agrb)

(a) y (agrb) is similar to Yb, then a is irrelevant

(b) y (agrb) is similar to YG, then a is a strong factor

(c) y (agrb) is partially different from Yb. A is a weak factor or a sub weak factor, and there must be other weak factors

step 2: after the first step is completed, use B variables to search. In this way, according to the requirements of some product standards and the needs of users and enterprises, the variables a, B, C, D, e, f are checked and searched. After that, we must grasp the strong factors, weak factors and sub weak factors. If there are two factors A and B, we will finally make a verification exchange

step 3: conduct verification exchange, set a and B as two factors, and conduct abbbrg and agbgrb tests

the above steps are the steps of variable retrieval. If you know that a, B, and C are non main variables, the indicators of a, B, and C can be relaxed, while the main variables D, e can fully control their target values and changes, which can improve the quality of products. 2 when applying a product to a performance test, there is an average error of 42 Units (far greater than the index requirements), which is the main reason for the performance index exceeding the standard. After discussion, the scientific researchers believe that there are six variables a, B, C, D, e and F

take 1 as the unit to measure the error value of the variable affecting this performance index, and the quantitative unit is given in this paper

y (agbgcgdgefg) =16

y (abbbcbdbebfb) =40

disassemble and reassemble the product, and then make it

y (agbgcgdgefg) =15

y (abbbcbdbebfb) =45

so d=27, d=3, d:d=9:1. This working process is reproducible, so the strong factors and weak factors are among these six factors, so we can use variable retrieval method to experiment, The experimental situation is shown in the following table: test sequence number combination result conclusion 1

2A Jinan assay cantilever impact tester experimental method; Brg


42 variable a is not important 3



37.8 variable B is not important 5



38 variable C is not important 7



36d is a weak factor, but it is not

Copyright © 2011 JIN SHI