Scale with common range（縮放統(tǒng)一范圍）：為FIR锐涯，F(xiàn)2P和PAR仿真數(shù)據(jù)（包括取決于選中活動單選按鈕的仿真褒搔、誤差浑吟，差分和去趨勢數(shù)據(jù)）設(shè)置相同的范圍（范圍=最大值—最小值）帮匾。若要激活此選項奉件，請右鍵單擊與FIR宵蛀，F(xiàn)2P或PAR數(shù)據(jù)對應(yīng)的圖例行。
Trend Properties（趨勢屬性）：自定義繪圖屬性（如背景顏色县貌，字體术陶，樣式）。
Create Tag From Prediction（從預(yù)測創(chuàng)建位號）：從來自FIR煤痕，F(xiàn)2P和PAR模型的仿真梧宫、誤差、差分和去趨勢數(shù)據(jù)創(chuàng)建位號杭攻。若要激活此選項祟敛，請右鍵單擊與FIR，F(xiàn)2P或PAR數(shù)據(jù)對應(yīng)的圖例行兆解」萏可使用此功能檢查模型的質(zhì)量。根據(jù)誤差或殘差創(chuàng)建位號锅睛，并檢查Tag Groups （位號組）> Trend（趨勢）菜單的自相關(guān)圖埠巨。
理想的情況是在非零的滯后時間，模型殘差的自相關(guān)圖應(yīng)該顯示零或非常小的相關(guān)系數(shù)现拒。類似的圖顯示殘差具有隨機(jī)（白噪聲）特性辣垒，并且模型已經(jīng)充分捕獲來自輸入變量的非隨機(jī)（確定性）影響。
Make Prediction Tag User Visible（使預(yù)測位號用戶可見）：將仿真數(shù)據(jù)（包括取決于選中活動單選按鈕的仿真印蔬、誤差勋桶，差分和去趨勢數(shù)據(jù)）填充到工作區(qū)中的位號列表。
如下圖所示侥猬，通常對于斜坡模型例驹，使用simulation（仿真）選項將產(chǎn)生預(yù)測漂移：

在數(shù)據(jù)段起始時，擬合預(yù)測（青色）開始匹配原始時間序列（藍(lán)色）退唠，但隨后似乎隨著時間開始漂移鹃锈。這是AIDAPro在數(shù)據(jù)段的時間零點處將仿真預(yù)測與實際數(shù)據(jù)相匹配，而過程在該時間點處于非穩(wěn)定狀態(tài)的結(jié)果瞧预。在實際生產(chǎn)數(shù)據(jù)集中屎债，這種情況會非常頻繁地發(fā)生仅政。
為了更好地評估這種情況下的模型擬合，請檢查不同的顯示框盆驹。當(dāng)是斜坡輸出變量時圆丹，為了匹配AIDAPro實際擬合，曲線會更新以顯示輸出變量和仿真預(yù)測的導(dǎo)數(shù)召娜。如下圖所示运褪，設(shè)置差分標(biāo)志可以更容易地驗證斜坡輸出變量擬合的質(zhì)量：

**Scale with common range: **Set the same range (range=max values-min values) for the simulated FIR, F2P, and PAR data (this includes the simulation, error, difference, and detrend data of depending on the activated radio button). To activate this option, right click on a legend row that corresponds to FIR, F2P, or PAR data.
**Trend Properties: **Customize the plot properties (e.g background color, font, styles).
Create Tag From Prediction：Creates a tag from simulation, error, difference, and detrend data from FIR, F2P, and PAR models. To activate this option, right click on a particular legend row that corresponds to FIR, F2P, or PAR data. Use this functionality to check the quality of your model. Create a tag from errors or residuals and check the autocorrelation plots from Tag Groups > Trend menu.
The autocorrelation plots of modeling residuals ideally should show zero or very small correlation coefficients at time lags other than zero. Such plot implies that the residual has random (white noise) characteristics and non-random (deterministic) effects from input variables have been sufficiently captured in the model.
Make Prediction Tag User Visible：Populates the simulated data (simulation, error, difference, detrend data depending on the activated radio button) to the list of tags in the workspace.
Often for ramp models, using the simulation option results in drifting predictions as seen in the following graph:
The fit prediction (cyan) matches the original time series (blue) well at the beginning of the segment, but then seems to drift away over time. This is a result of AIDAPro matching the simulation prediction to the actual data at time zero of the segment, and the process not being at steady state at that point in time. This situation can happen very frequently in real-world data sets.
To better assess model fits in this type of situation, check the difference box on the display. The plot is updated to show the derivatives of the output variable and the simulation prediction, which matches the actual fit done by AIDAPro in the case of ramp output variables. As seen in the following figure, visualizing the quality of a ramp output variable fit is much easier with the difference flag set:

2016.11.25