Learn more about Model-Based Calibration

New Response Models and Datum Models

On this page…
Adding New Response Models Datum Models

Adding New Response Models

When you first set up a test plan, the Data Wizard automatically contains the response model setup after the data matching functions.

The following applies when you return to a previously setup test plan to add a new response model, or when you click the New button at test plan level to add new response models to an existing test plan.

Double-click the Responses outport of the block diagram or use the New Response Model item in the File menu, or use the toolbar icon. (None of these is available unless you are in the Test Plan view, that is, have the test plan node selected in the model tree. This should be obvious; you can only see the test plan block diagram with the test plan node selected.)

The Response Model Setup dialog box has a list box containing all the variables in the selected data set except the inputs to the local and global models; you cannot use an input also as a response.

You can reach the Local Model Setup and Global Model Setup dialog boxes using the Set Up buttons to change the local and global models also, and you can add Datum Models (maximum or minimum) if the local model supports this. See below for more information.

You can return to the local or global setup options individually at any time by double-clicking the block in the test plan diagram.

Back to Top

Datum Models

Under Datum you can choose a datum model when setting up a new response model, but only for some local models — polysplines and polynomials (but see Linked Datum Models following). Other local models cannot have a datum model, as they do not necessarily have a unique turning point.

The datum model tracks the maximum or minimum of the local models. This is equivalent to adding the maximum or minimum as a response feature, which can be useful for analysis if those points are interesting from an engineering point of view.

The Datum options are

• None

• Maximum — This can be useful in cases using polyspline modeling of torque against spark. The maximum is often a point of engineering interest.

• Minimum —- This can be useful for cases where the object is to minimize factors such as fuel consumption or emissions.

• Linked datum model — This is only available to subsequent two-stage models within a test plan in which the first two-stage model has a datum model defined. In this case you can make use of that datum model. The linked datum option carries the name of the response of the first two-stage model, where it originated.

If the maximum and minimum are at points of engineering interest, like MBT or minimum fuel consumption, you can add other response features later using the datum model (for example, MBT plus or minus 10 degrees of spark angle) and track these across local models too. It can be useful to know the value of MBT when modeling exhaust temperature, so you can use a linked datum model from a previous torque/spark model. Having responses relative to datum can also be a good thing as it means the response features are more likely to relate to a feature within the range of the data points.

You can also export the datum model along with local, global, and response models if required. See Exporting Models.

Fitting Process for Polynomial Splines With a Datum Model

The fitting process for a polynomial spline with a maximum datum is:

1. The toolbox fits a quadratic polynomial to the data.

2. The toolbox finds the x-location of the maximum of this polynomial (if it doesn't have a maximum, then the model will not be fitted).

3. The toolbox uses this x-value as a starting point in an optimization to find the best knot position for the polynomial spline. Note this optimization does not have any constraint that Bhigh2 stays negative.

4. The toolbox checks the result to see if the new knot position is still at the maximum of the curve. If so, then finish.

   If this is not the case, then the algorithm returns to the quadratic polynomial fitted in step 1, which has the required maximum.

Back to Top

Global Model Setup

Build Models Dialog Box

© 1984-2009- The MathWorks, Inc.    -   Site Help   -   Patents   -   Trademarks   -   Privacy Policy   -   Preventing Piracy   -   RSS