选项窗口字段和按钮
Closed Loop Data(闭环数据):辨识要拟合的数据是在闭环条件下收集的。
Perform Compaction(执行压缩):若需要,允许FIR压缩。
Initial Conditions at Steady State(稳态初始条件):数据在中断前是否稳定。
Check Input Dependence(检查输入相关性):如果选中,它将在执行FIR拟合之前检查输入之间的线性相关性。如果存在共线输入,将显示警告消息及共线输入和受影响的输出列表。
Tag \ Attribute(位号/属性):列标题。
Ramp Flag(斜坡标签):输出是斜坡。
Drifting Flag(漂移标签):输出受非固定未测量干扰的影响。
No Immediate Response(无立即响应):归因于数据采样的零阶保持。
Prefilter Horizon(预滤波器时域):仅当漂移标签打开时适用。确定了预滤波器的时间常数。
Defaults(默认值):重置此窗口上所有按钮为默认值(斜坡标签除外)。
提示工具
提示工具在AIDAPro PCTP 2009 Build 1中作为AIDAPro 拟合和预测选项的简要指南。
在“Fit Group/Fit(拟合组/拟合)”节点上,可在“数据和选项”选项卡中使用提示工具。要访问该提示工具,请将鼠标指向特定的拟合选项,将显示描述性文本框。
数据选项卡的提示工具
提示工具具有以下适配选项:
• Resample Multiple(多个周期重采样)
选项卡选项的提示工具
适用于以下拟合选项
• Closed Loop Data(闭环数据)
• Perform compaction(执行压缩)
• Initial Conditions at Steady State(稳态下的初始条件)
• Check Input Dependence(检查输入相关性)
• Ramp Flag(斜坡标签)/Driftig Flag(漂移标签)/No Immediate response(无立即响应)/ Prefilter Horizon(预滤波器时域):描述这四个选项的提示工具。
FIR拟合
如Identification Concepts(概念识别)(参阅第27页的AIDAPro建模“Philosophy(原理)”)所述,模型识别的推荐原理是首先使用非参数多变量方法(第29页的“High-Order Model Identification(高阶模型辨识)”)预拟合数据。执行简化每个所得单输入单输出(SISO)关系模型到参数形式(第35页的“Model Reduction to Parametric Models(模型简化为参数模型)”)。
使用此标签可以:
•估计非参数多变量模型(使用FIR或ARX技术)。这将生成阶跃响应的非参数模型及其置信区间(如果您选择了计算置信选项)。
•对每个SISO关系,简化模型到参数形式。默认情况下,高阶结果将减为一阶模型。
•尝试不同的模型类型,以获得最佳的模型简化为参数形式。你还可以基于特定的过程知识设置模型参数约束。
FIR拟合选项卡有以下三个子选项卡:
• FIR Fit Setup (FIR拟合设定)
• FIR Fit Plots (FIR拟合图)
• FIR Fit Parameters(FIR拟合参数)
原文:
Options Window Fields and Buttons
Closed Loop Data: Identifies that the data being fit was collected under closed-loop conditions.
**Perform Compaction: **Permit compaction of FIR if needed.
Initial Conditions at Steady State: Whether or not the data was steady before any break.
Check Input Dependence: If checked, it will examine the linear dependencies among inputs before performing the FIR Fit. If collinear inputs exist, a warning message as well as a list of the collinear inputs and affected outputs will be displayed.
**Tag \ Attribute: **Column header.
**Ramp Flag: **Output is a ramp.
**Drifting Flag: **Output is affected by non-stationary unmeasured disturbances.
No Immediate Response: Zero order hold due to data sampling.
Prefilter Horizon: Applies only if drifting flag turned on. Determines the time constant of the prefilter.
**Defaults: **Button that resets all defaults on this window (except for ramp flag).
Tooltips
Tooltips are introduced in AIDAPro PCTP 2009 Build 1 as short guides to AIDAPro Fit and Prediction Options.
At Fit Group/Fit node, tooltips are available in Data and Options tabs. To access tooltips, direct your mouse to a specific fit option and a descriptive text box will appear.
Tooltips for Data tab
Tooltips are available the following Fit Options:
• Resample Multiple
Tooltips for Tab Options
Available for the following Fit Options
• Closed Loop Data
• Perform compaction
• Initial Conditions at Steady State
• Check Input Dependence
• Ramp Flag/Driftig Flag/No Immediate response/ Prefilter Horizon : one tooltip describes these four options
FIR Fit
As stated in Identification Concepts (see AIDAPro Modeling “Philosophy” on page 27), the recommended philosophy for model identification is to first pre-fit the data using a non-parametric multivariable approach (“High-Order Model Identification” on page 29). A model reduction on each of the resulting singleinput-single-output (SISO) relationships to parametric form is performed (“Model Reduction to Parametric Models” on page 35).
Use this tab to:
• Estimate a non-parametric multivariable model (using FIR or ARX techniques). This results in the step response of the non-parametric model and its confidence bounds (if you selected the compute confidence option).
• Perform a model reduction to parametric form on each SISO relationship. The high-order results are reduced to a first order model by default.
• Experiment with the different model types to get the best model reduction to parametric form.
You can also set constraints on model parameters based on specific process knowledge.
The FIR Fit tab has the following three subtabs:
• FIR Fit Setup
• FIR Fit Plots
• FIR Fit Parameters
2016.11.14