1、STATISTICAL PROCESS CONTROL,统计过程控制,What is Statistics ?A science which deals with Collection of data Summarisation of data Analysis of data Drawing information from data,什么是统计学 ?统计学是一门科学,具体涉及如下内容: 收集数据 总结数据 分析数据 从数据中找出信息,Why Statistics? It is more economic to assess a sample of product and use the r
2、esult to predict the properties of the whole lot. It leads to predictions with a high degree of precision.,Sample,Measurement,Judgement,Taking actions,STUDY,Information,Population-Sample Model,Base Population,Data,为何要统计?统计可以节约评估样品的费用,并且可以用该评估结构来预所有物品所需的费用 统计可以使预测具备极高的精确度,样品,测量,判断,采取行动,研究,信息,人口-样本模型,
3、基层群体,数据,Process Control Model,Voice of the process,SPC,ValueAddition,Customer,Changing Needs,Voice of the Customer,* * * Input Process Output,Man Machine Material Method Environment,Product Service,控制过程模型,过程声音,SPC,价值增加,客户,改变需要,客户的声音,* * * 输入 处理 输出,人 机器 材料 方法 环境,产品 服务,数据变量 属性(在连续的范围内进行测量) (只能接受离散值) 例
4、如: 重量, 例如:无缺陷硬度, 时间, 产出,DATAVARIABLE ATTRIBUTE(Measurements (Can take onlyon a continuous discrete value)scale)e.g. Weight, e.g. no defectives,Hardness, no. of defectsTime, Yield,WHAT IS SPC.,?,DATA COLLECTION METHODS TO INDICATE A PICTURE OF THE PROCESS AND TO PROVIDE A MEASUREMENT OF PROCESS PERFO
5、RMANCE. THE PICTURE WILL SHOW BOTH NATURAL AND UNNATURAL PROCESS BEHAVIOUR BASED UPON PROCESS INPUTS AND REFLECT THE PROCESS OUTPUT RESULTS.,何为 统计过程控制?,?,数据收集方法的目的是指出该过程的图片,并且提供 过程执行的测量方法。 该图片会显示以过程输入为基础的自然和非自然 的过程行为,并能反映输出结果的过程,WHAT SPC IS NOT,?,A SUBSTITUTE FOR NORMAL INSPECTION ROUTINES WHERE PRO
6、CESS CORRECTION IS USUALLY BASED ON A SAMPLE OF ONE. IT IS NOT INTENDED TO TAMPER WITH INHERENTLY STABLE PROCESSES.,IRRESPECTIVE OF SPECIFICATIONS.,统计过程控制不包括下例范围,?,不能代替正常检查 在进行修改时的常规程序 通常是以样本为依据的 并非有意损害稳定自然的过程,不考虑是否符合规格,WHY IMPLEMENT SPC,CUSTOMER INSISTS,KEEP AHEAD OF COMPETITION,LATEST CONTROL TECH
7、NIQUE,NEED TO IMPROVE,ISO 9000 REQUIREMENT,S.P.C.,执行 统计过程控制的原因,顾客的坚持,在竞争中领先,最新的控制技术,需要改进,ISO 9000 必备条件,统计过程控制,TARGETS IMPROVEMENT,REDUCES SCRAP,PROGRESS PERFORMANCE,PLANS SAMPLING FREQUENCIES,PREVENTS DEFECTS,STABILITY AND CAPABILITY,ASSIST,HOW CAN SPC HELP,?,目标改进,减少废品,过程执行,计划取样率,防止故障的出现,稳定性,协助,统计过程
8、控制有何帮助,?,M,C,B,WHERE DO YOU FIT IN,?,PREVENTION,DETECTION,PRO-ACTIVE,EFFICIENT,HIGH QUALITY,COST EFFECTIVE,RE-ACTIVE,INEFFICIENT,EXPENSIVE,LOW QUALITY,M,C,B,你符合哪一项,?,预防,发现问题,赞成活动,有效,高质量,有效花费,重新活动,效率低,价格昂贵,质量差,Tolerance,Avoid,Waste,Prevention,Defection,容忍,避免,浪费,预防问题,发现问题,WASTE,PEOPLE + FACILITIES + M
9、ATERIALS + TOOLS,R,S,Rs PRODUCE DEFECTIVE PRODUCTS.Rs FIND DEFECTIVE PRODUCTS.Rs REPLACE DEFECTIVE PRODUCTS.,浪费,人 + 厂房 + 原料 + 工具,R,S,Rs 生产不合格的产品.Rs 发现不合格的产品.Rs 更换不合格的产品.,OBJECTIVES,THEORYTARGETSREACTIONPRODUCTNUMBER GAME CONFORMITY,PRACTICALSIGNALSINTERPRETATION PROCESSRIGHT REASONSIMPROVEMENT,目标,原理
10、对象反应产品数字游戏 一致,实际信号过程说明正确原因改进,WHERE WOULD YOU LIKE YOUR COMPANY TO BE,?,WORLD CLASS,BELIEVE YOU ARE THE BEST,SATISFIED WITH THE CURRENT POSITION,BEST IN THE COUNTRY,BEST IN THE AREA,?,你希望所属的公司以下处于哪个区域,?,世界级,相信自己是最好的,满足于现在的位置,国内最强,区域内最强,?,(A) Direct Effects1. Product quality becomes stabilised.2. Prod
11、uct quality is improved.3. Defective products decrease.4. Yields increase.5. Complaints and claims from users decrease.6. Production costs decrease.7. Production output increases.8. Unnecessary work disappears.9. Reliability of product increases.10. Costs for examination and inspection decrease.11.
12、Incorrect or false data disappears.,Effects of “Quality Control with Charts”,(A) 直接影响1. 产品质量变得稳定2. 产品质量获得改进.3. 产品不合格率降低.4. 产量增长.5. 来自用户的抱怨和索赔降低.6. 产品成本降低.7. 产量增加.8. 多余的工作消失.9. 产品的可靠性增强.10. 检查和检测费用降低.11. 错误数据消失.,质量控制图表的影响,(b) indirect Effects1. Systematic and overall understanding, analysing and chec
13、king of the production process will become possible without depending on personal hunches and experiences.2. Accurate judgements are deduced from obtained data.3. Anyone can always deduce the same judgement from the same data.4. Defects which have been overlooked become apparent.5. Defects can be di
14、scovered before they become serious.6. A regular trend is found from irregular data distribution.7. Not a part but the whole problem becomes clear.,Effects of “Quality Control with Charts”,(b) 间接影响 在不依赖个人感觉和经验的情况下,使系统和全局理解,分析和检查生产过程变得可能。 从取得的数据推断出的正确判断。任何人总能够从同样的数据的出相同的判断。 已被忽略的缺陷变得明显。 在缺陷变得严重前,及时发现
15、。 6. 从不合规律的数据分布中,找出符合规律的趋势。 7. 不仅可以发现一部分问题,也可以使整个问题变得明显。,质量控制图表的影响,Common Tools Check Sheet Histogram Pareto Analysis Cause and Effect Diagram Stratification Scatter Diagram Graphs and Control Charts,普通工具 检验单 直方图 柏拉图分析 原因和影响图表 层化 散点图 曲线图和控制图表,Check sheet for wedge hardnessRange Calvary mark Frequenc
16、y 169 183 197 211 225 239 253 267 281 / 2,楔子硬度的检验单范围 骑缝号 频率 169 183 197 211 225 239 253 267 281 / 2,Bushes Bore Diameter10.77 10.76 10.73 10.75 10.7810.76 10.79 10.75 10.75 10.7610.77 10.78 10.76 10.78 10.7510.80 10.77 10.74 10.79 10.7410.72 10.75 10.82 10.76 10.7310.76 10.77 10.79 10.76 10.7710.75
17、10.76 10.77 10.78 10.7510.81 10.74 10.81 10.74 10.7310.75 10.78 10.76 10.80 10.7710.74 10.79 10.78 10.77 10.80Spec :10.75+0.05,衬套内径10.77 10.76 10.73 10.75 10.7810.76 10.79 10.75 10.75 10.7610.77 10.78 10.76 10.78 10.7510.80 10.77 10.74 10.79 10.7410.72 10.75 10.82 10.76 10.7310.76 10.77 10.79 10.76
18、10.7710.75 10.76 10.77 10.78 10.7510.81 10.74 10.81 10.74 10.7310.75 10.78 10.76 10.80 10.7710.74 10.79 10.78 10.77 10.80规格:10.75+0.05,Histogram,20 18 16 14 12 10 8 6 4 2,LSL,USL,FREQUENCY,BORE DIA,10.70 .72 .74 .76 .78 .80 10.82,8,17,1,14,7,3,直方图,20 18 16 14 12 10 8 6 4 2,LSL,USL,频率,内孔直径,10.70 .72
19、.74 .76 .78 .80 10.82,8,17,1,14,7,3,Steps for Making Histogram1. Obtain a set of 50 to 100(N) observations. 2. Obtain the maximum value and minimum value. 3. Determine the number of classes KK = N -1 4. Class width c = Max - Min, rounded off to a convenient figure.K 5. Calculate the class boundaries
20、 as, min - 1/2 (Least-Count) to be the starting point. 6. Make a frequency table, i.e., number of data falling in each class. 7. Histogram : Erect rectangles over the class interval having area proportion to the frequencies. 8. Frequency polygon :Starting from zeroJoin the midpoints corresponding th
21、e top of each rectangle by straight lines.,制作直方图的步骤1. 获得一组5到100(N)的观察值。 2. 取得最大值和最小值。 3. 决定K级的数字K = N -1 4. 宽度等级 c = 最大 最小, 舍入一个附近的数值。K 5. 按照以下公式计算等级范围, 最小值 - 1/2 (最小计算) 到开试点。 6. 做一个频率表, 例如, 在各等级中下降的数据量。 7. 直方图: 在等级间隔上建立矩形 ,该矩阵区域和频率相称. 8. 频率多边形 :从零开始通过直线加入与各矩形顶部相应的中间点.,Exercise on HistogramNo. of Ob
22、servations (N) No. of classes (k)Under 50 5 750 100 6 10100 250 7 15Over 250 10 20Metal Box thickness (mm)Data Max. Min.3.56 3.46 3.48 3.50 3.42 3.43 3.52 3.49 3.44 3.50 3.56 3.423.48 3.56 3.50 3.52 3.47 3.48 3.46 3.50 3.56 3.38 3.56 3.383.41 3.37 3.47 3.49 3.45 3.44 3.50 3.49 3.46 3.46 3.50 3.373.5
23、5 3.52 3.44 3.50 3.45 3.44 3.48 3.46 3.2 3.46 3.55 3.443.48 3.48 3.32 3.40 3.52 3.34 3.46 3.43 3.30 3.46 3.52 3.303.59 3.63 3.59 3.47 3.38 3.52 3.45 3.48 3.31 3.46 3.63 3.313.40 3.54 3.46 3.51 3.48 3.50 3.68 3.60 3.46 3.2 3.68 3.403.48 3.50 3.56 3.52 3.46 3.48 3.46 3.52 3.56 3.56 3.56 3.46 3.52 3.48
24、 3.46 3.45 3.46 3.54 3.54 3.48 3.49 3.41 3.54 3.41 3.41 3.45 3.34 3.44 4.47 3.47 3.41 3.48 3.54 3.47 3.54 3.34 N = 100, Xl = 3.68 Xs = 3.30,有关直方图的练习观察号 (N) 等级号 (k)50以下 5 750 100 6 10100 250 7 15250以上 10 20金属箱厚度 (mm)数据 最大 最小3.56 3.46 3.48 3.50 3.42 3.43 3.52 3.49 3.44 3.50 3.56 3.423.48 3.56 3.50 3.5
25、2 3.47 3.48 3.46 3.50 3.56 3.38 3.56 3.383.41 3.37 3.47 3.49 3.45 3.44 3.50 3.49 3.46 3.46 3.50 3.373.55 3.52 3.44 3.50 3.45 3.44 3.48 3.46 3.2 3.46 3.55 3.443.48 3.48 3.32 3.40 3.52 3.34 3.46 3.43 3.30 3.46 3.52 3.303.59 3.63 3.59 3.47 3.38 3.52 3.45 3.48 3.31 3.46 3.63 3.313.40 3.54 3.46 3.51 3.48
26、 3.50 3.68 3.60 3.46 3.2 3.68 3.403.48 3.50 3.56 3.52 3.46 3.48 3.46 3.52 3.56 3.56 3.56 3.46 3.52 3.48 3.46 3.45 3.46 3.54 3.54 3.48 3.49 3.41 3.54 3.41 3.41 3.45 3.34 3.44 4.47 3.47 3.41 3.48 3.54 3.47 3.54 3.34 N = 100, Xl = 3.68 Xs = 3.30,Frequency Table Class Class Boundries Mid value Frequency
27、 Tally Frequency No. 1 3.275 3.325 3.3 / 3 2 3.325 3.375 3.35 / 3 3 3.375 3.425 3.4 / / 9 4 3.425 3.475 3.45 / / / / / / / 32 5 3.475 3.525 3.5 / / / / / / / / 38 6 3.525 3.575 3.53 / / 10 7 3.575 3.625 3.6 / 3 8 3.625 3.675 3.65 / 1 9 3.675 3.725 3.7 / 1,频率表 等级号 等级 界限 中间值 计数频率 频率1 3.275 3.325 3.3 /
28、 3 2 3.325 3.375 3.35 / 3 3 3.375 3.425 3.4 / / 9 4 3.425 3.475 3.45 / / / / / / / 32 5 3.475 3.525 3.5 / / / / / / / / 38 6 3.525 3.575 3.53 / / 10 7 3.575 3.625 3.6 / 3 8 3.625 3.675 3.65 / 1 9 3.675 3.725 3.7 / 1,Histogram,40 35 30 25 20 15 10 5 0,3.3 3.35 3.4 3.45 3.5 3.55 3.6 3.65 3.7,直方图,40 35
29、 30 25 20 15 10 5 0,3.3 3.35 3.4 3.45 3.5 3.55 3.6 3.65 3.7,Frequency distribution and specification limits,A. Variability is considerably less than required, and average level is well centered. No problems. B. Variability is again small, but average level has shifted close to the maximum limit. Pro
30、duction of defectives is imminent. C. The danger, foreseen in B, has remained unchecked, resulting in defective product.,分布频率和规格限制,A. 可变性基本上小于需要,并且平均水平居中。表示不存在问题。 B. 变化的可能性小了, 并且平均水平已接近最大限制。即将出现不合格产品。 C. 在中预见的危险还是没有被检测到, 因此造成了不合格品。,D. Product variability is still acceptable, but high. As a result, t
31、he distribution is “crowding“ both minimum and maximum limits. Investigation should be made as to whether (a) product variability can be reduced or (b) consumers requirements permit widening of the tolerance. E. The type of distribution in D has deteriorated into actual occurrence of defectives, due
32、 to slight shift in average level towards maximum limit. For remedy see D. F. In this case, the type of distribution in D has remained well centered, but a slight increase in variability has occurred, resulting in defectives. For remedy, see D. G. Double distribution, referred to as often as “bi-mod
33、al“. Probably two sets of tools, change in process or material during running of lots are responsible. H. Distribution is “cut off“ at the minimum limit, a typical sign is that vendor has screened out defectives below minimum limit prior to shipment. If distribution had been properly centered, the c
34、ost of screening and losses of product could have been avoided. I. The small distribution outside the maximum limit may represent limit may represent parts from tool and setup tryouts not set aside. By proper instruction of personnel, this situation should be readily corrected.,Frequency distributio
35、n and specification limits (contd),D.产品可变性依然可接受, 但是可变性衡高。 结果导致分布在最小和最大限制很拥挤。应该 做 关于 (a) 是否可以减少产品可变性 或者 (b) 应顾客要求,放宽公差。 E.由于在平均水平中朝最大限制的微小改变,在D中的分布类型已恶化,实际上导致不合格产品的出现,。补救措施见 D。 F.在这种情况下,D中的分布类型已保持居中。但是,可变性已发生了微小的增加。补救措施见D G.双重分布,通常指“双模式”。可能有两组工具在过程中发生了改变,或者说在磨削许多工件的期间所使用的原材料必须可靠。 H.在最小限度时,分布被切断。这是一个典
36、型的信号,说明卖主已经在装运前,筛选出了在最低限度下的不合格品。若分布位于正中,筛选的成本,并且可以避免产品的损失。 I.在最大限度以外的较小分布可能代表限度,也可能代表来源于工具的零件,并且不取消设置,试用。通过对人员的正确指导,这种情况应该很容易得到纠正。,频率分布和规格限制,Summarisation of variable data,Variable data,Central tendency,Mode,Dispersion,Range,Standard Deviation,Mean,Median,可变数据的总结,可变数据,中间趋势,模式,离中趋势,范围,标准误差,平均值,中值,Dat
37、a summarisationCentral Value : MeanX = X1 + X2 +XNN,数据总结代表值 : 平均数X = X1 + X2 +XNN,Data X 1. 1 15 15 15 102. 1 6 11 15 193. 19 24 6 3 294. 21 3 8 12 15,Calculate X,数据 X 1. 1 15 15 15 102. 1 6 11 15 193. 19 24 6 3 294. 21 3 8 12 15,计算 X,DispersionA) Range =Xmax - XminSet Data Range s1 1 15 15 15 15 15
38、 292 1 6 11 15 19 24 29Limitation of Range : Affected only by two extreme valuesB) Standard deviation(Deviation measured from mean)s = (Xi - X )2N - 1,离中趋势A) 范围 =Xmax - Xmin套 数据 范围 s1 1 15 15 15 15 15 292 1 6 11 15 19 24 29范围限制 : 仅仅受两个极限值的影响B) 标准偏差(根据平均值测量的偏差)s = (Xi - X )2N - 1,Section IIConcept of
39、 variation & Need for process control,第二部分变分的概念 & 需要的过程控制,Variation Is the cause of all quality problems Is a natural phenomenon- in nature- in industry Its detection depends upon- Precision of measurement- Skill of personnel It can be reduced but not eliminated Lower variation means higher possibil
40、ity of conformance,变化 是引起所有质量问题的原因 是一个自然现象- 其本质- 在行业中 能否发现变化,主要取决于- 测量的精确- 人员的技术 可以减少变化,但不能消除变化 较低的变化意味着较高的一致可能性,Variation No two product or characteristics are ever alike. Difference may be large, or they may be unmeasurably small, but is always present. Variations can be due to chance causes or as
41、signable.Example :Dimension of a machined part would be affected by :- Machine (clearance, bearings wear)- Tool (hardness, strength, geometry)- Material (size, hardness)- Operator (setting, accuracy of location)- Maintenance (lubrication, temperature)- Environment (temperature, voltage fluctuation),
42、变化 是上没有任何两种完全相同的产品或相同的特性 差异可能大,也可能小到无法测量,但是 变化通常是一直存在的 变化可能是偶然发生的,也可能是必然发生的举例:一加工零件的尺寸可能受以下因素的影响 :- 设备 (游隙, 轴承损耗)- 工具 (硬度, 强度, 几何特性)- 材料(尺寸, 硬度)- 操作者 (调试, 准确定位)- 维护 (润滑, 温度)- 环境(温度, 电伏波动),Chance vs. Assignable causes of VariationCHANCE (COMMON) CAUSES ASSIGNABLE (SPECIAL) CAUSES Consists of many ind
43、ividual causes Consists of just one or few individual causes Any one cause results in only a Any cause can result in large minute amount of variation variation Example : Example:- Slight variation in raw material - Batch of defective material- Lack of human perfection in reading - Untrained operator
44、instruments and setting controls - Faulty set-up Cannot be economically eliminated Easy to detect and generally economical to eliminate Process follows a predictable No specific pattern(statistical) patternGOK HAK(GOD ONLY KNOWS) (HUMAN ALSO KNOWS),偶然变化原因 vs. 必然变化原因偶然原因 必然原因 包括许多个人原因 包括一个或者很少一部分人的原因
45、 任何一个造成一两分钟量的变化的原因 任何造成较大变动的原因 举例 : 举例:- 原料中的微小变化 - 一批不合格原材料- 在阅读说明书,控制设置方面,人员的素质并非最佳 - 未经培训的操作者- 调试不正确 不能采取较为经济的手段消除 易于发现,并可以采取较为经济的手段消除 工艺流程允许采用可预测的统计模式 无特定模式GOK HAK(只有上帝才知道) (只有人类才知道),M,C,B,Associates,COMMON,CAUSE VARIATION,* NATURAL BEHAVIOUR. * SAME PERFORMANCE OVER TIME. * PREDICTABLE PATTERN - RANDOM. * IRRESPECTIVE OF SPECIFICATION. * STABLE.,