Point-by-Point Model Tradeoffs

On this page…
What Is A Point-by-Point Model Tradeoff? Adding a Point-by-Point Model Tradeoff Calibrating Using a Point-by-Point Model Tradeoff

What Is A Point-by-Point Model Tradeoff?

There are two types of tradeoff that you can add to your session, a tradeoff of independent models, as described earlier (see Performing a Tradeoff Calibration), or a tradeoff of interconnected models: a point-by-point model (or multimodel) tradeoff.

A point-by-point model tradeoff is a specially built collection of models from the Model Browser.

You can build a series of models so that each operating point has a model associated with it. In the Model Browser, you can export models for a point-by-point model tradeoff from the test plan node. The models must be two-stage and must have exactly two global inputs. You can use the point-by-point test plan template to create these models. For more information see Set Up a Point-by-Point Model.

The procedure for calibrating by using a point-by-point model tradeoff follows:

1. Import your model and create tables from your point-by-point model. (See Creating Tables From a Model.)

2. Calibrate the tables. (See Calibrating Using a Point-by-Point Model Tradeoff.)

3. Export your calibration. (See Importing and Exporting Calibrations.)

The point-by-point model is only defined for certain cells in the tradeoff tables. These are the operating points that were modeled using the Model Browser part of the toolbox. These cells have model icons in the table. At each of these operating points, you can use the model to trade off, and by doing this you can adjust the value in the table. The point-by-point model is not defined for all other cells in the table and so you cannot use models to tradeoff. You can edit these cells and they can be filled by extrapolation. You trade off values at each of the model operating points in exactly the same way as when using independent models, as described in Choosing a Table Value at a Specific Operating Point. When you have determined table values at each of the model operating points, you can fill the whole table by extrapolation by clicking . See Filling Tables by Extrapolation.

Back to Top

Adding a Point-by-Point Model Tradeoff

The simplest way to create your point-by-opint tradeoff is to:

1. Import your model into CAGE. See CAGE Import Tool).

2. Create tables from your point-by-point model. See Creating Tables From a Model.

You can also:

• Create the tradeoff by exporting from the Model Browser test plan. See Exporting Point-by-Point Models to CAGE.

• In CAGE, create a new tradeoff and then import the point-by-point models from a file.

To create a new tradeoff from scratch:

1. Select File > New > Tradeoff. CAGE switches to the tradeoff view and creates a new empty tradeoff.

2. Select File > Import > Point-by-Point Model Tradeoff File.

   The file must have been exported from the Model Browser, from the test plan. See Exporting Point-by-Point Models to CAGE.

3. Select the correct file to import and click Open. This opens a dialog box.

   

4. In the left Model sites list, you can clear the check boxes for any models at operating points that you do not want to import.

   Notice that the operating points are displayed graphically at the top. If an operating point is deselected, it is displayed as gray here, rather than blue.

   CAGE will create tables for all the models and input variables, with breakpoints at all the model operating points. You can choose not to create all the tables; click Select Tables to choose from the list which tables you want.

5. Choose the normalizers (axes) of the tables by using the X- and Y-axis input drop-down menus.

6. You can adjust the number of breakpoints in the following ways:

   • Leave the Automatic breakpoint settings radio button selected and edit the relative tolerances around the model sites. Use the tolerance edit boxes in the model setup pane. You can observe the effects of altering the tolerances on the number of breakpoint dotted lines drawn on the top graphic. Initially each model site has a breakpoint. If operating points are close together, you can increase the tolerances to decrease the number of breakpoints.

     For example, if several close points may all have been intended to run at exactly the same point, you might want to adjust the tolerances until those model points (displayed as blue dots) only have one breakpoint line. The number of rows and columns that will be created is displayed in the edit boxes on the right.

   • Alternatively you can select the Manual breakpoint settings radio button and enter the number of rows and columns in the edit boxes, and you can directly edit the values of the breakpoints.

7. Click OK.

When you click OK, CAGE creates all the tables for the multimodel tradeoff, with breakpoints at the values you have selected.

Note   When you calibrate the tables, you can only use models to tradeoff at the operating points defined for the models. These cells have model icons in the table. You can edit other cells, but they have no models to tradeoff associated with them.

You can now calibrate your tables. See the next section, Calibrating Using a Point-by-Point Model Tradeoff.

Back to Top

Calibrating Using a Point-by-Point Model Tradeoff

Each editable operating point in your tables has a model icon in the cell, like this example cell.

These cells have a model defined at that point. You use the display of these models to help you trade off values at these points to fulfill your aims in exactly the same way as when using independent models in "ordinary" tradeoff mode, as described in Choosing a Table Value at a Specific Operating Point.

1. Change input values by dragging the red lines on the graphs or by typing directly into the edit boxes above the graphs. Use the context menu, toolbar or Inputs menu to find the maximum, minimum, or turning point of a model if appropriate.

2. Look at the model evaluation values (to the left of each row of graphs) and the input variable values (in the edit boxes below the graphs) to see if they meet your requirements.

   Remember that the green highlighted graphs indicate how the selected table is filled: if a row is green, the model evaluation value (to the left) fills the table at that operating point; if a column is green, the input variable value (in the edit box below) fills the table.

   See the example following; the SPK column of graphs is green, so the value of SPK in the edit box is entered in the table when you click the Apply Table Filling Values button ( ).

   

3. When you are satisfied with the tradeoff given by the value of your variable at this operating point, you apply this value to the table by pressing Ctrl+T, selecting Tables -> Apply Fill to Table, or clicking   in the toolbar.

4. When you have determined table values at each of the model operating points, you can fill the whole table by extrapolation by clicking    . See Filling Tables by Extrapolation.

You can then export your calibration; see Importing and Exporting Calibrations. An example point-by-point tradeoff is shown following.

Back to Top

Controlling Table Extrapolation Regions

Optimization Setup

© 1984-2012- The MathWorks, Inc.    -   Site Help   -   Patents   -   Trademarks   -   Privacy Policy   -   Preventing Piracy   -   RSS