This is machine translation

Translated by Microsoft
Mouseover text to see original. Click the button below to return to the English version of the page.

Note: This page has been translated by MathWorks. Click here to see
To view all translated materials including this page, select Country from the country navigator on the bottom of this page.

Validate Estimation Results

This topic shows how to validate estimation results in the Parameter Estimation tool. After you estimate parameters as described in Estimate Parameters and States, validate the estimation results using another data set.

Configure and Perform Validation

To validate a model using the Parameter Estimation tool:

  1. Load the validation data set.

    At the MATLAB® prompt, load the validation data into the MATLAB workspace.

    load iodataval;
  2. Add a new experiment for validation.

    In the Parameter Estimation tool, in the Validation tab, click New Experiment. A new experiment with default name appears in the Experiments area of the tool. To rename the experiment ValidationData, right-click the experiment and select Rename from the drop-down menu.

  3. Import the validation data set into the validation experiment.

    Right click the experiment name, and select Edit. Specify [time,iodataval(:,1)] in Inputs , and [time,iodataval(:,2)] in Outputs.

  4. Specify the experiment for validation.

    When you create an experiment, it is by default selected for estimation. To select another experiment for validation, in the Validation tab, click Select Experiments.

    In the Select Experiments dialog box, clear Estimation and select Validation for the validation experiment.

  5. Specify the estimation results to use in the validation.

    After you import validation data, select the estimated parameter values to validate.

    The estimation results are saved in the EstimatedParams variable in the Results area of the tool. In the Validation tab, click Select Results to Validate.

    To validate the estimated parameters, in the Select Results dialog box, select EstimatedParams and clear Use current parameter values.

  6. Select the plots to display at the end of validation.

    Parameter Estimation tool can display the measured and simulated responses and the residuals plot at the end of validation. Select the plots to display by checking the corresponding box on the Validation Tab.

  7. Validate the estimation results.

    On the Validation tab, click Validate. The Validation Progress Report shows the status of the validation.

Compare Measured and Simulated Responses

  1. Compare measured validation data against the model output simulated with the estimated parameters.

    The tool displays the experiment plot for each experiment selected for validation. Each experiment plot shows the measured data, and data from simulation using each set of results selected. For example, the following figure shows the experiment plot for the ValidationExp data.

  2. Examine the Residuals Plot.

    The residuals plot shows the difference between simulated response and measured data. When there is a good fit between the simulated output and measured data, the residuals show the following behavior:

    • Lie within a small percent of the maximum output variation.

    • Do not display any systematic patterns.

    For example, you can see from the following figure that the residuals for ValidationExp data satisfy both criteria.

Related Examples

More About