Some standard content:
JCS03.120.30
National Standard of the People's Republic of China
GB/T 4091—2001
idt Is0 8258:1997
Conventional control charts
Shewbart control charts
2001-03-07 Issued
2001-09- 01 Implementation
State Administration of Quality and Technical Supervision
GB/T 4D91 -2001
1Product
2Symbols
3Nature of operating control charts
4Types of conventional grid charts
4.1Control charts for standard value determination situations
4.? 4.3 Types of control charts for the given standard value 5 Control charts for measurement control 5.1 Mean (R) or standard deviation (SD) control 5.3 Median (Me) control chart 6 Control procedures and interpretation of measurement control charts Tests for the cause of variation Process control and process efficiency 10.1 Preparations before establishing a control chart 10.1 Selection of rapidity Analysis of production performance Selection of management tools Problem and size of the process |I have prepared the effective data for the collection
the steps of the control plan
12 Graphic diagram of electrical control machinery
-2.1x chart R chart: the situation where the standard value is given
2.22 Interval and R chart: the situation where the marked value is given
12.3 Single force (X) and rotation difference control chart: the situation where the standard value is not determined.....Median chart: the situation where the standard value is not given.13 Examples of counting control charts
1.1.1 Force and multiple charts: the situation where the standard value is given in a constant technical path13.2 Force use: the situation where the standard value is given1.3 [Chart: $The standard is not provided for the tangible
13.4 Unit product unqualified micro chart 2.
I have given all the traps! Old document
GB/T4D91—2001
This standard is equivalent to the international standard IS0)8258:1991 Shewhart control chart 3 (5hcwh11controlcharts) and its 153 year 1 is static single,
This standard replaces GR/T2091,14091.9—1183. Compared with GB/T4091.1~4091.9—1983, the main changes in the technical specifications of this standard are:
1. The content arrangement format is different, and the original series of 9 standards are combined into one! The arrangement pattern of the plotting points of the standard judgment criteria is changed, such as the original 7-point chain is changed to a 9-point chain, the original 7-point trend is changed to an E-point trend, etc. Each type of conventional control chart is divided into two situations: standard value given and standard value not given; the "over-progress strategy" chart is added; it is clearly stated that in the joint application of the long chart and the long chart, the R chart should be established and analyzed first. Appendix A of this standard is a reminder appendix. This standard is proposed by the China Standards Research Center, and this standard is the National Technical Committee for the Application of Standardization Methods. The main originators of this standard are: China Standards Research Center, School of Economics and Management of Huazhong University of Science and Technology, Institute of Mathematics and Systems Science of Chinese Academy of Sciences, and Institute of Mechanical Science. The originators of this standard are: Liu Wensun, Ma Yihuai, Li Hua, and Xiao Hui. This standard is the first revision of the original control diagram 3 of GB/T4051.1~4091.91983. This standard was first published in 1983.
GB/T4091—2001
ISO is a joint committee of ISO member groups: the work of formulating national standards is completed through the technical requirements of the 190 member meeting. If the member group is interested in the work of its technical committee, it has the right to participate in the meeting. International groups (coordinated or non-coordinated) that maintain contact with ISO can also participate in the relevant work. In the field of electrotechnical standardization, ISO maintains close cooperation with the International Electrotechnical Commission (IEC). International standards adopted by technical committees need to be submitted to member groups for voting. Only when the member groups with the most votes can they be officially issued as international standards: The original international standard TSC) 258 was prepared by the Technical Committee on Application of Statistical Methods of ISO/TCG9. The appendix A of this international standard is only a reminder. GB/T4091—2001
The traditional method of manufacturing industry relies on the production of products, and then inspecting the final product and screening out the quality control of products that do not meet the specifications. This inspection strategy is usually wasteful and uneconomical because it is a matter of inspection after the defective products are produced. It is more effective to establish a strategy to avoid waste and not produce the defective products in the first place. This can be achieved by collecting and analyzing the process information and taking actions or measures from the process itself.
The statistical method is a graphical method that applies the principle of statistical analysis to the control of the production process. It was first proposed by Dr. Linhardt in 1924. Control graph theory holds that there are various kinds of interactions. The first is that the "clock variation" is random variation, and the variation is caused by "chance causes" (also called "common causes") that are inherent and unidentifiable. The effect of each of these causes will only contribute a very small component of the total variation, and some will only contribute a significant component. However, the sum of the effects of all these unidentifiable chance causes is measurable and assumed to be inherent in the process. When this is eliminated, management policies are needed to free up resources to improve the process system and the actual changes in the process. This change is attributed to some identifiable cause that is inherent to the process and can be eliminated, at least in theory. These identifiable causes are called "accountable causes". They are materials inhomogeneity, tooling errors, process or operation errors, unstable performance of manufacturing or designed equipment, etc. The data obtained from the process-control chart can help detect abnormal patterns and provide a test criterion for statistical out-of-control. When the apparent cause of the variation is the process is in a state of control, the acceptable level of deviation is determined, and any deviation from the level of control is assumed to be caused by the identifiable cause. These identifiable causes should be identified, eliminated or reduced. The purpose of statistical process control is to establish and maintain an acceptable and stable level of control. To ensure that products and services meet specified requirements. To do this, the main statistical method used is the control chart. A control chart is a graphical method that provides information about the current state of a sample and compares this information with the established control results. Control charts are used to help assess whether a particular process has been achieved and whether a statistical state with an appropriate specified level has been achieved. They are then used to help maintain control and a high degree of consistency over the characteristics of the key product components by maintaining a continuous record of product quality during the production process. Use sample preparation and detailed analysis of control charts to better understand and improve the process. 1 Standard
National Standard of the People's Republic of China
Conventional Control Charts
Shewhart GB/T4091—2001:3/T 405:1.1~4351.9—1385 This standard provides a guide to the use and understanding of conventional control charts for statistical control of processes. This standard is only applicable to process control methods that use a system of control functions. Some supplementary information consistent with conventional control charts is introduced, such as the use of warning limits, the analysis of power and capacity, etc. In addition, there are some control methods, their characteristics can be found in GB/T7SS.
2 Symbols
Group size: the number of subgroup measurements in a single subgroup. Subgroup effect
Observation of quality characteristics [X,,X, can be used to represent a single observation), sometimes other symbols.For example, the median is the true value of the process mean of the ten groups of observations, for each group of observations arranged in ascending or descending order. When 1 is the median, the median is the median of the group. When 1 is the median, the median is the median of the two middle groups of observations. In the case of a single graph, the moving range is the difference between the largest and smallest values in the group of observations: 1, -X1. !X:, etc., the mean of the extreme values
Standard deviation of subgroups
1x,-X)
Average of subgroup standard deviation
Value of internal standard deviation
Estimated standard deviation of internal process
Unqualified product rate of factory group
Power=Unqualified product group size
Half mean of the unqualified product rate of all ten groups
Unqualified products in a subgroup
Number of unqualified products
Number of unqualified products
Average of all subgroup unqualified products
Number of unqualified products per unit of subgroup
GB/T 491—2001
The number of unqualified products in the subgroup is the number of unqualified products of all products/the total number of products inspected 3The nature of the control chart required
The control chart requires data to be drawn from the process at approximately equal intervals. The interval can be defined by time (e.g., per hour or number of devices (e.g., lead), avoid abnormal, this refers to the selection of subgroups in the group, the subgroups are composed of products or services with the same number of units and the same subgroup size, and one or more subgroup characteristics are obtained for each group, such as the group mean, group difference or standard deviation. The control chart is a kind of trap that corresponds to the given group characteristic value and the subgroup number. It contains a center line (CL) as the standard value of the point characteristic. In the assessment process, whether it is in In a state of statistical control, this benchmark is usually the mean value of the observed data: for well-regulated tests, this benchmark estimate is usually the long-term value of the characteristic specified in the product specification, or a nominal value based on the tested characteristic, or an implied standard value of the product or service. The control chart also contains two control limits determined by statistical methods, one at the center level called the upper limit (and the lower limit LC), and the lower limit (U
). The control limits of the conventional control chart are located at a convenient distance from the center. Wherein, is the total within-group standard deviation of the plotted statistic. Within-group variation is a measure of machine variation, which can be used to calculate an appropriate number of group variations. This measure does not include group variation. , and the control limits show that if the process is under control, then about 9 out of 7 of the subgroups will be within the control limits. In other words, when the package is under control, there is about a 3% risk that the plotted points will fall outside the control limit or the lower control limit on average 10 times. The word "about" is used here because any deviation from the basic assumptions (such as the assumptions about the data distribution) will affect this probability number. It should be noted that some professionals prefer to use 3.0 instead of 3. to make the nominal probability value 0.2%, or 2 bottom reports in 300 times, but Shewhart did not choose the coefficient of 3. Some professionals use real probability detection for control charts of non-stationary distributions, such as variation charts, grid product rate charts, etc. In order to emphasize the empirical solution, the control chart adopts two control limits instead of the rate control limit. The control limits are used to determine the probability of a loss. When a subgroup exceeds the control limit, some action is taken. Therefore, the control limit is sometimes called the "action limit". In many cases, it is helpful to add two control limits to the control chart. In this way, any subgroup value falling outside the two limits is used as a sign of the upcoming out-of-control state. Therefore, the control limit is also called the "caution limit". Two types of errors may occur when the control chart is used: The first type of error is called the first type of error. This is enough. The process involved is still based on the setting: bzxZ.net
1 Here 08250:5 is the international standard for the second virtual technology makeup difficulty history, and the state report is "GA/T". 2091—200
is in a state of flux, but at some point the process fails due to some accidental reason, and the conclusion that the process is in control is reached. The second type of error is called a Type I error. When the process is under control, but the point produced by the process still falls within the control limits due to some accidental reason, the wrong conclusion is reached that the process is still in control. The risk of a Type II error is a function of the following two factors: the width of the control limits, the number of process control points, and the size of the subgroups. The above three factors are continuous in nature, so the risk of a Type II error can only be estimated. The first type of error is considered to be a function of the control limits, the process control points, and the size of the subgroups. The first type of error is considered to be a function of the control limits, the process control points, and the size of the subgroups. The above three factors are continuous in nature, so the risk of a Type II error can only be estimated. The probability of error detection is 0% for the first type of error. Because of the high pricing situation, the loss of Type II error is not recognized! The concept of station metering is not practical, and carefully select a whole small group size: for example, or because of the convenience of the dispatch of the restrictions, the plan will focus on controlling and improving the performance of the vehicle body, the product is suitable and feasible.
When the process is in a state of potential control, the control rules provide a new method of continuous verification of the statistical principle of the morning secretion, which is generally set to maintain a state of statistical control. The specific situation of the daily value of the process is determined by the second American English language error risk, and the failure to meet the appropriate normal level of risk. The small action principle of the teaching belt is not studied in the sense of the proposed verification (see 1S) 724 (H/1799). The regulation control chart emphasizes that the control chart is used to identify the effective experience of the process "variable state" rather than the general solution. The author must carefully study the operating characteristics of the control chart and use it as a test and solution. When a full point value falls outside the limit or a series of points are extracted in reverse order as described in the first part of the test, the statistical state is not accepted. If this situation occurs, adjustments should be made to determine the cause when the error occurs. The cause can be confirmed and checked at a glance. The control status can be continued at any time. As we have proved above, for the first type of error, in the case of rare occurrences, we cannot understand the cause, and the card layer must When a control chart is initially established for a process, it is often found that the process is out of control. The control charts calculated based on the microcontrollers for this out-of-control process will lead to premature conclusions, as these control limits are too large. For this reason, it is always necessary to adjust the control state before establishing a control chart. The following sections will discuss methods for establishing control charts. 4 Types of Control Charts There are two types of control charts: measurement charts and efficiency charts. Each type of control chart has two different characteristics: ) Standard value given:
b> Standard plot,
standard value is the specified requirement or daily value (Notes to Table 1, Table 5 and Table 5) 4.1 Standard detection of given characteristic to
This kind of control chart is to find out whether the characteristic of the plot is significant or not; the variation caused by accidental reasons is completely based on the data, and the variation caused by non-accidental reasons is not significant. 4.2 Standard control plan
This kind of control chart can determine whether the difference between the standard value of a group of data and the corresponding standard value is significantly greater than the expected variation caused by natural reasons, and the value of each sub-year is the same. Standard The difference between control charts for certain situations and control charts for certain situations is that the requirements for the process center position are different. Standard values can be determined based on experience gained through the use of test information or control charts without specified standard values, or they can be determined by considering the needs of the service and the economic values of the production process, and then the nominal values specified in the product specification. More appropriate standards are considered to represent all the data within the specified range. In order to control the effective operation of the standard position, the data should be collected together with the solid material, and the chart based on such standard values should be used for the calculation of the control, especially for the purpose of using three-dimensional control, so as to achieve the balance of production performance. 3
4.3 Types of control charts for recording and counting control charts
Central measurement control charts
TB/T4091—2001
1) Chart of card value (X) and slow trend (R) or standard deviation) chart: 2) Single dependence (X and moving range ();
3) Digit (M) chart and maintenance () chart
b Counting control charts
1) Chart of unqualified products (,) chart of number of unqualified products (chart: 2) Number of unqualified products (chart or number of unqualified units of products (&chart. 5 Measurement control charts
Measurement data refers to the observed values obtained by measuring and recording the numerical size of the characteristics of a single product in the case group, such as length expressed in m, resistance expressed in ohms, sound expressed in decibels (), etc. Measurement control (especially its most commonly used types The types and specifications of these control functions represent typical applications of process control. Counting control functions are particularly useful for the following reasons: Most processes have a countable nature, so gate control charts are widely used. 1) A metric is more suitable than a screened process because the expression contains the required information. It is not necessary to consider the performance of the process. The chart can be used to give an independent description of the performance of the process. In the long run, some controls can be compared with the specification, while others cannot. Although it is usually more expensive to obtain counted data than to obtain valid data, the subgroup size of the count data is almost always much smaller than that of the count data, so it is more effective. In some cases, this can help reduce the cost of general inspection and shorten the time interval between the production of parts and the final product. This standard specifies all metric data. The control chart assumes that the group variability follows a positive (Gaussian) distribution, and this assumption will allow the control chart to use the normality assumption to derive the coefficients. Since most control limits are used to make decisions, it is believed that the normality should not have a significant impact on the control chart. In general, due to the extreme limit theorem, the mean will always tend to be positively distributed, even when the individual observations do not follow a positive distribution, so for charting purposes, it is reasonable to assume that the control chart is normal even if the group variability is 1 or 2. When dealing with individual observations, the distribution of the observations will reshape, and the validity of the normality assumption should be checked regularly. When using a single summary of data, it should be noted that the distribution of the extreme limit theorem is not always normal, although it is useful in estimating control limits. When calculating constants, an assumption of approximate stability is made for the distribution of the extreme values of the standard deviation, which is satisfactory for empirical decision-making procedures. 5.1 Mean (X) chart and extreme R or standard deviation (area) Metrology control charts can be used to describe the data of both deviation (product variation) and location (process average). Because of this, measurement control charts are always used to control the mean and position, and are not the control charts of the process. The most commonly used control charts are the K charts: Table 1 and Table 2 give the control limit formulas and coefficients of the control charts respectively. The control limit formulas of the control charts are usually given. The actual standard values of DR and DR are determined. Note that XR is initially a test value, st or r is the standard value, and the initial determination of UCL is performed. I LG.
x24;
"Group
Number of stable values
1.880:2.659
:1-732
1.023,1.95-
GA/T 4091—2001
Table non-control chart calculation line coefficient
Control limit coefficient
0, 000° 2. 606 0. 000 3. 585c.coq3.257
0. 090 2. 2rh U. U0u 2, 354
Center end system effect
737 51-253 3 1. 13E5. SE6 5 | 71. 3420. 5771. 427
2. 08s0, ovo 1. 61 U. Uu 1. 913 3. U/ o,nca2 959 41. 42 3 2. 754 0. 360 61.0s9,0. .85
0. 17611.F51 0.388
.sus .J3%1.t:
0. 95 1. 5 3.#7 5. 51 7
01.7910.232/1. 702 0. 547 5. 393 0. 184 1. 8130. 969 31.0517 2.676 5. 3541.332n.230
1. o. 337
0.949 a. 3c93
.2841-716.26.66 U.687 5.16 *2 1.7775.975
0.903 .285
n.xR 0. 26
0. 332 . 2+9
.X0: 235 | |tt | 31. 63r 0. 81t.52 . 256,1. 744jD. 975 41.n25 23. 173 5. 315 22.927
n. 3611. 6460. 2.61.610 0. 922 5. 594,3. 2831.710.977 G1.0289
3.2585.3659
0.382..-6190.371.81.025.54
4c.3951.5631.1.65.595
n. 466:1. 428 1. 5724. 4211. 644 1. 2usS.TE9
c.4:c1. 526,1 282 5. 7821, 33.416-552
D. 4Fk -.534
c.4541. 5111. 356 3. 820 9. 3730.4821-518
.356 3.391
3. 356 5. 250 8
J.407 > 263 5
2. 92 21. 6r× 03. 472 . 2F4 01. 6:3
c. yx ajl. u' . x
.934 5;1.0.6 7
. 9,5 4:1. ( 4 0.4 03. 689
0.180:
0. 519,--490c. 3041. 1761. E49.22t U. 1.. 1. 5850. 946 91. ( 4 R. 7556. 655 0.173 0. 563||t t||C. 624ju.167 (.647 522
177 0. 5161.4.91. E9: : H1 u. 12 1. 575.9x 61.
0. 9xR 21. 0:1 93. 613
1. 4860.5281.4481. e56 6. 679 6. 434 1. 56505, 06 U. 415 1.13.
0. 162 0. 63: 3. 5+3
1,157 n, 6*9 ( . 555
1--3a1.716
0-2787
71. u-1 4 3. R55
.4291.7596
6. 031, 0. 451 1. 543j0. 99 21. 0,0 9: 3. 555 0. 25 75n.549!1.
0 . 153 0. 606 0. 5651. 5350. 559 1- 1201.836. t 4: 1. 41 0 99 6 Data: ASTFilh, .SA
5.2 Single value (X) control chart
E.010 F 3. I11 . 754 4
In some process control situations, it is either impossible or impractical to obtain a complete subgroup. The above typical disappearance occurs when the measured value is > or when the output of any one of the two programs is relatively uniform. In other cases, there is only one possible value. In these cases, the program should focus on individual data for process control, such as collecting meter readings or batching raw materials. In the single control case, since there is no reasonable set of constraints to provide an estimate of the within-batch variation, the calculation is based on the variation provided by the two observations. The moving average is the absolute difference between two adjacent observations in a sequence, that is, the absolute difference between the first observation and the second independent measurement, and then the absolute difference between the first observation and the third observation. The range of the observed values is the difference, and so on: from the moving range the semi-mean spherical dynamic potential () is calculated, and then the control plan is generated. Similarly, the total semi-mean value can be calculated from the entire number (X) Table 3 gives the control limit formula for the single value control chart. For the single value control chart, the control limit formula for each point is as follows: If the process is not correctly understood, the interpretation of the unit control should be particularly cautious. 5
GB/T4091---2001
C) Unit system The analysis of the interannual reproducibility of the process is not correct, so in this case, a smaller group size (2 to 4) and an R control chart may give you a better Well, even if a longer interval is required between subgroups, it is called: Table 3 Control limit formula for single value control chart
Statistic A
Unit x
Table dynamic pole stem
Central Road
1..and the standard position of the governance,
mark is difficult to detect and has not yet been fully
indicates that the pool position has dropped by half and the average moving extreme difference, Buying effect and - from Table 2 = 2, we get:
5.3 Currency position stimulus (control chart
center line
R:x dsr
standard blue age determination
UCL and L.CL
can only be calculated by numerical control, the median can be used as a substitute for text and R Control charts within the chart, the control obtained from the median chart, have certain advantages over the X and X charts. They are easier to use and require less calculations, which can increase the acceptance of control charts by field workers. Spend. In the figure, the control limits of the median effect plan can be used to describe the degree of dispersion of the scientific output and to give a dynamic description of the operation of the median effect plan. This method is used to calculate the reverse direction: the median of the group is used to make the difference between the median and the median of the group. The latter method is easy to calculate and the speed is fast. If this method is used in this standard, the calculation plan of the control limit is as follows:
5.3.1 The center line of the digit chart
= the average digit of a group of digits ||tt| |[.CI...=MI AR
ILM-MeAR
The inverse of the pull-in effect, the method is the same as 5.1 The same as the R chart. The properties of the constant 41. are shown in Table 4.
It should be noted that some restricted median control charts respond slower to the control situation than the literature. At the value of A| |tt||5.3.2 Photograph
Heart-Night-The average value of R values of all sub-passes UCL=D,R
G1=F,R
English number, Table 2.
6 Control procedures and interpretation of metrological control charts
Product parts with process naming through the chart system are processed and under the current water level (respectively , the station calculations remain unchanged. The individual subgroup differences (only the mean value and the X-value) are only caused by natural factors. They rarely exceed the control limit. In other words, the knee may stop at the natural cause. The data will not show some obvious trends or patterns except for the changes caused by the cow. It also controls the central position of the product and shows the stability of the process: the figure reveals the instability between groups from the perspective of the average The control chart shows that the process is not expected to deteriorate. It is an indicator of the size of the process under consideration. 4091:--2001
If the within-group variation is not true, the scale chart shows that the control status is not correct. In this case, all This will only occur when the subgroups receive the same treatment. If the process does not maintain statistical control, the adult value will increase, which may indicate that different subgroups have received different treatments or that different The winding factor is acting on the process. The closed control state of the control chart will also affect the X chart. Since both the extreme difference migration layer and the new release energy of the particle dry part are the estimates of the variation between the particles , so the first thing to do is to analyze the scale drawing. The following review procedures should be followed: 6.1 Project acceptance and analysis Analyze the data and calculate the half-mean and range: 6.2 First, point to R. Compare with the control data to check if there are any out-of-control points or any unusual patterns or trends. If the cause is a symptom, analyze the operation of the process in order to identify it, make corrections, and prevent it from occurring again.
6.3 Eliminate all traceable causes that are affected by the request Then recalculate and assign new control limits to the points. When comparing with the new control limits, make sure that all points are in unified control. It is necessary to correct the "fraudulent burial" and re-explain the sequence.
6.4 If the source of the identified error is not found, then any error will be removed from the graph and the corrected value will be calculated. Solution and estimation of the control limit of Wuzhou-A,R Ren 2: Exclude the display of out-of-control decision-making, the group does not have a wide taste of "dismantling the bad effect", more evenly planned electricity said, the pass certificate is subject to the second Knowing the exact point of the moon's shadow, we can estimate the area where the background becomes uneven. 6.5 When a control chart indicates that a process is in statistical control, the discrete process is considered to be in statistical control. If the process is stable, the average value can be analyzed to determine whether the position of the process changes with time. Draw an X control chart at 6.5 points and compare it with the integer limit. After checking the data points There are out-of-control points, or abnormal patterns or trends. - column, analyze any out-of-tune conditions, and then take corrective and preventive measures. Remove any large control points that have been fixed: recalculate and plot the new process half value (text) When the new control limits are compared, you need to confirm that all data points are in statistical control. If necessary, repeat the "Create New Calculation" procedure. If the initial data of the technical standard values are all within the test control limits, then the current control limits will be used within the technical period: these control limits will be used to control the current process, and the person in charge (operator or supervisor) will 1. Integrate the signal of any control situation in the figure or figure and seek immediate action: 7. Pattern test of identifiable causes of deterioration
Figure 2 shows a set of explanations for the common The eight pattern tests for the control chart are shown in the figure. The history of these tests is fully discussed in references [ and [6].
Although the above pattern tests can be used as a correction test,However, the analyst should also consider the possible influence of special causes on the process. Therefore, when the market shows the source of the problem, the peak should only be measured according to the actual situation. The occurrence of any of the conditions specified in these tests indicates that the main source of the problem has occurred. The control limits are located above and below the center. In order to consider the above tests, I will divide the system into 6 zones, each with 1 zone. The labels of these 6 zones are A, HB, B, and C. The three centers are symmetrical. This test is applicable to X-charts and single-value X-charts: This single-value X-chart assumes that the value of the X-chart is viscous! The distribution needs to be
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.