啟用校正模式
應(yīng)用中CORRECTING和REPOSITION模式下默認(rèn)動(dòng)作可以按以下某一個(gè)值設(shè)定STCORRECT參數(shù)來配置：
?STCORRECT=-1（約束）（缺史逄隆）
當(dāng)應(yīng)用處于TEST或TEST-NOPULSE模式時(shí)不會(huì)自動(dòng)調(diào)用CORRECTING或REPOSITION模式。
與多測試(Multi-Test)模式一樣蠢终，當(dāng)CV違反約束后丁屎，必要的違反約束校正動(dòng)作將會(huì)自動(dòng)進(jìn)行抵知。然而叁熔，TEST或TESTNOPULSE將繼續(xù)。
?STCORRECT=0（復(fù)位）
當(dāng)CV違反約束時(shí)應(yīng)用將調(diào)用復(fù)位(REPOSITION)模式沸移。
?STCORRECT= 1（校正）
當(dāng)CV違反約束時(shí)應(yīng)用將調(diào)用校正(CORRECTING)模式痪伦。如果違反約束持續(xù)一定時(shí)間，即使運(yùn)行在校正模式后雹锣，它也會(huì)重新進(jìn)入復(fù)位模式网沾。

概要
CORRECTING模式用于驅(qū)動(dòng)過程在約束范圍內(nèi)，使得階躍測試可能得到恢復(fù)蕊爵。在TEST, TEST-NOPULSE或SUSPEND模式下辉哥，當(dāng)CV違反約束且最大數(shù)量校正動(dòng)作（STCMFLAG）已經(jīng)應(yīng)用后其將觸發(fā)。當(dāng)CORRECTING模式被觸發(fā)時(shí)攒射，當(dāng)前的MV階躍測試將被中斷醋旦，測試組的獨(dú)立DMCplus控制動(dòng)作將被激活。在CORRECTING模式時(shí)会放，MVs通過類階躍動(dòng)作將違反約束CVs帶回測試邊界限饲齐。系統(tǒng)將嘗試保持階躍狀態(tài)MV，使得即使有類似中斷當(dāng)前階躍依舊可以計(jì)數(shù)咧最。對(duì)于斜坡CV捂人，它將嘗試在最接近極限的里面甩骏，而不是測試極限中間返回一個(gè)測試裕度。測試組的CORRECTING模式將至少持續(xù)STREPOMIN × SSMINS分鐘先慷。但如果處于CORRECTING模式長于STREPOMAX× SSMINS而CVs依然處于違反約束狀態(tài)，則系統(tǒng)將激活REPOSITION模式（使用連續(xù)控制動(dòng)作）咨察。
CORRECTING模式遵循以下規(guī)則：
?當(dāng)CV超過測試限（STCVMIN和STCVMAX）外STCVDELTA時(shí)應(yīng)用將觸發(fā)CORRECTING模式论熙。
?至少經(jīng)過指定的STREPOMAX× SSMINS時(shí)間后應(yīng)用才將觸發(fā)CORRECTING模式。在STREPOMIN× SSMINS時(shí)間后如果所有的CVs都距測試限內(nèi)至少1/2STCVDELTA則該組將恢復(fù)測試摄狱。
?在STREPOMAX× SSMINS時(shí)間后脓诡，如果所有的CVs都在測試限內(nèi)，則該組將繼續(xù)測試媒役。對(duì)于斜坡CV祝谚，除了在測試限之內(nèi)，斜率也必須小于STRAMPRT酣衷。如果CVs不在測試限內(nèi)交惯，或斜坡變量不滿足斜率準(zhǔn)則，則將觸發(fā)REPOSITION模式穿仪。
當(dāng)處于CORRECTING模式時(shí)席爽，離散動(dòng)作解決方案是基于相關(guān)CV測試裕度（STCVDELTA）和模型增益（測試裕度越大，解決方案動(dòng)作越大）計(jì)算啊片。

避免CORRECTING和REPOSITION模式
盡管一些CORRECTING, REPOSTION和CONTROL模式的數(shù)據(jù)也可用于模型識(shí)別只锻，測試過程中一個(gè)主要目標(biāo)是限制應(yīng)用處于CORRECTING或REPOSITION模式的時(shí)間。在不違反最大MV階躍大小和CV測試限的前提下紫谷，智能向?qū)б嫠褂玫碾A躍大小算法將盡可能嘗試最大的階躍輸出齐饮。鑒于各種不確定性，若CV違反約束笤昨，測試人員依然可以進(jìn)入CORRECTING或REPOSITION模式祖驱。測試人員是不斷將過程推向測試限的，但由于可能使用較差的模型瞒窒，它會(huì)經(jīng)常對(duì)限制響應(yīng)過量或不足羹膳。

整定以避免CORRECTING或REPOSITION模式包括以下內(nèi)容：
?設(shè)置適當(dāng)?shù)臏y試限(STCVMIN和STCVMAX)虽界。限制應(yīng)設(shè)置成假設(shè)它們很可能被侵犯柴罐。
?CV安全裕度(STCVDELTA)是一個(gè)指示觸發(fā)REPOSITION模式前超出限制容忍量的整定參數(shù)灌诅。
?在測試模式中狈蚤，STSTEPFRAC是一個(gè)影響所有MV階躍大小的全局參數(shù)廊移。它反映了你對(duì)模型的信心日丹，以及可以用來避免過多的違反約束剿另。
在測試中一個(gè)好的策略是將界限和安全裕度設(shè)置在合理的值刁赖，然后使用STSTEPFRAC參數(shù)盡快動(dòng)作壳炎。隨著新的泞歉，改進(jìn)的模型被投用逼侦，測試中可以通過增加STSTEPFRAC來最大化階躍尺度。

也可以看看
智能向?qū)?yīng)用中的REPOSITION(復(fù)位)模式
智能向?qū)?yīng)用中的校正動(dòng)作
智能向?qū)?yīng)用中的參數(shù)整定

附原文：
Enabling CORRECTING mode
The default CORRECTING and REPOSITION mode behavior for an application can be configured by setting the STCORRECT parameter to one of the following values:
?STCORRECT = – 1 (CONSTRAINT) (default)
No CORRECTING or REPOSITION mode will be invoked automatically when the application is in TEST or TEST-NOPULSE mode.
The necessary violation correction moves will be made automatically whenever CV violates, just like in Multi-Test mode. However, the TEST or TESTNOPULSE will continue on.
?STCORRECT = 0 (REPOSITION)
The application will invoke REPOSITION mode when CV violation happens.
?STCORRECT = 1 (CORRECTING)
The application will invoke CORRECTING mode when CV violation happens. If the violation persists for a certain time period, even after running in CORRECTING mode, it will fall into REPOSITION mode.

**Overview **
CORRECTING mode is used to drive the process within constraints so that step testing may resume. It is triggered from TEST, TEST-NOPULSE or SUSPEND modes when a CV violates the constraint and the maximum number of correcting moves (STCMFLAG) has already been applied. When CORRECTING mode is triggered, the current MV step test is interrupted, and a discrete DMCplus control action is activated for the test group. During CORRECTING mode, the MVs make step-like moves to bring violating CV(s) back one test margin inside the limits. An attempt is made to keep the stepping MV where it was, so the current step can still be counted even with this interruption. For a ramp CV, it attempts to return it one test margin inside the closest limit, instead of the middle of the test limits. A test group stays in CORRECTING mode for at least STREPOMIN × SSMINS minutes long. But if it is in CORRECTING mode longer than STREPOMAX × SSMINS with CV(s) still violating their constraints, then REPOSITION mode is activated (using continuous control action).

CORRECTING mode is governed by the following rules:
?CORRECTING mode is triggered by a CV that is more than STCVDELTA outside of its test limits (STCVMIN and STCVMAX).
?CORRECTING mode is triggered for a minimum of time specified by STREPOMAX×SSMINS. After STREPOMIN×SSMINS if all of the CVs are at least ? STCVDELTA inside the test limits then the group will resume testing.
?After STREPOMAX×SSMINS, if all the CVs are inside the test limits then the group will resume testing. In addition to being inside test limits ramp CVs must have a ramp rate that is less than STRAMPRT. If the CVs are not inside the test limits or the ramps don’t satisfy the rate criteria then REPOSITION mode will be triggered.
When in CORRECTING mode, the discrete move resolution is calculated based on the associated CV test margin (STCVDELTA) and the model gains (the bigger the test margin, the bigger the move resolution).

Avoiding CORRECTING and REPOSITION mode
Although some CORRECTING, REPOSTION and CONTROL mode data can be used for identification, a major goal during testing is to limit the amount of time the application spends in CORRECTING or REPOSITION mode. The step size algorithms used by the SmartStep engine will try to make the largest step that does not violate either the maximum MV step size or any CV test limits. Because of various uncertainties, the tester may still enter CORRECTING or REPOSITION mode due to CV limit violations. The tester is constantly pushing the process toward test limits, but since it is using potentially poor models, it will frequently overshoot or undershoot the limits.

Tuning to avoid CORRECTING or REPOSITION mode includes the following:
?Setting the test limits appropriately (STCVMIN and STCVMAX). The limits should be set with the assumption that they will likely be violated.
?The CV Safety Margin (STCVDELTA) is the tuning parameter that indicates how much overshooting of the limits will be tolerated before triggering REPOSITION mode.
?STSTEPFRAC is the overall parameter that affects all MV step sizes in test mode. It reflects a confidence in your models, and it can be used to avoid excessive constraint violation.
A good strategy is to set the limits and the safety margins to reasonable values, and then use the STSTEPFRAC parameters to temper moves early in the test. As new, improved models are brought online, STSTEPFRAC can be increased to maximize the step size used during the test.

See Also
REPOSITION mode in a SmartStep application
Correcting moves in a SmartStep application
Tuning parameters in a SmartStep application

2015/10/27