累積和控制圖在企業(yè)質(zhì)量控制中的應(yīng)用(更新版)
【正文】 ct: Shewhart chart is a tool for distincting between normal and abnormal fluctuations in the process and determing whether the process is in control. However, The judgement of the Shewhart chart for whether the production process is normal, is based on the isolated result of the observation. Its Cpacity for detecting large shift is extremely high, but for the small shift is very low. Accumulative sum chart use averagelinked to control the two types of error, whose judgement for the process is based on the results of previous is very useful in detecting the sudden tiny changes of the process average. Therefore, according to the data from Hejian Hengtai Die Casting factory ,this paper will use Shewhart chart and Accumulative sum chart to analyze whether the property value of the quality in the production process is result will show that Accumulative sum chart has higher sensitivity in the abnormal process, thus demonstrate its superiority.Key words: Shewhart chart，Accumulative sum chart ， Sequential probability ratio test ，averagelinked一、引言在質(zhì)量管理過(guò)程中，可以應(yīng)用很多技術(shù)和方法，其中統(tǒng)計(jì)技術(shù)是很有價(jià)值的一類(lèi)方法。因此，本文將通過(guò)對(duì)河北省河間市恒泰壓鑄件廠生產(chǎn)的鑄件數(shù)據(jù)分別應(yīng)用休哈特控制圖分析和累積和控制圖來(lái)判斷生產(chǎn)過(guò)程質(zhì)量特性值是否受控, 從而表明累計(jì)和控制圖對(duì)過(guò)程異常有更高的靈敏度，從而顯示出它的優(yōu)越性。二者都是ISO9000標(biāo)準(zhǔn)中推薦的樹(shù)立統(tǒng)計(jì)方法，然而累計(jì)和控制圖對(duì)過(guò)程異常有更高的靈敏度，具有更高的優(yōu)越性。若有異常波動(dòng),則認(rèn)為過(guò)程失控,這時(shí)就需要查找原因,采取適當(dāng)行動(dòng)及時(shí)糾正,使過(guò)程恢復(fù)到受控狀態(tài)。CUSUM控制圖的設(shè)計(jì)思想是將樣本信息加以累積，將過(guò)程的小偏移累加起來(lái)，以達(dá)到放大的結(jié)果，提高檢驗(yàn)小偏移的靈敏度。Thomas ， Luceno amp。如果兩種假設(shè)下的概率相差不大，則繼續(xù)抽檢下一個(gè)樣本，再重復(fù)上述過(guò)程，直到能明確做出接受H或H的假設(shè)為止。如果一個(gè)時(shí)間序列的獨(dú)立樣本觀測(cè)值X，X，…，X，服從方差為σ (均值未知)的正態(tài)分布，那么上式可以由以下式表示: ①其中Δ=μμ，是兩個(gè)假定的均值之差或過(guò)程均值偏離μ的程度,S是對(duì)樣本數(shù)據(jù)偏差的累積和 S=則式①確定了上下控制線。當(dāng)樣本累積和落到任何三個(gè)區(qū)域之一時(shí)，就停止抽樣。由于累積和控制圖與序貫概率比檢驗(yàn)有著緊密的聯(lián)系，當(dāng)隨機(jī)變量x服從正態(tài)分布、二項(xiàng)分布、幾何分布、指數(shù)分布、泊松分布時(shí)，基于序貫概率比檢驗(yàn)下，累積和控制圖參數(shù)k和h的選取方法各不相同。為及時(shí)發(fā)現(xiàn)失控L越小越好，1－1∕L即為犯第二類(lèi)錯(cuò)誤的概率。％，故令p＝%,確定p時(shí)，應(yīng)使p∕p的值不致于過(guò)小，根據(jù)過(guò)去的生產(chǎn)所能達(dá)到的較好水平，取p＝%。檢測(cè)到一種趨勢(shì)，或持續(xù)上移或下移。由于該累積和是從20號(hào)樣本開(kāi)始，因此20號(hào)樣本可能是過(guò)程均值增大的起始
環(huán)評(píng)公示相關(guān)推薦
鄂ICP備17016276號(hào)-1