Documentation

This is machine translation

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

Note: This page has been translated by MathWorks. Please click here
To view all translated materals including this page, select Japan from the country navigator on the bottom of this page.

How Online Parameter Estimation Differs from Offline Estimation

Online estimation algorithms estimate the parameters of a model when new data is available during the operation of the model. In offline estimation, you first collect all the input/output data and then estimate the model parameters . Parameter values estimated using online estimation can vary with time, but parameters estimated using offline estimation do not.

To perform offline estimation, use commands such as arx, pem, ssest, tfest, nlarx, and the System Identification app.

To perform online parameter estimation in Simulink®, use the Recursive Least Squares Estimator and Recursive Polynomial Model Estimator blocks. For online estimation at the command line, use commands such as recursiveARX to create a System object™, and then use step command to update the model parameters.

Online estimation differs from offline estimation in the following ways:

  • Model delays — You can estimate model delays in offline estimation using tools such as delayest (see Determining Model Order and Delay). Online estimation provides limited ability to estimate delays. For polynomial model estimation using the Recursive Polynomial Model Estimation block or the online estimation commands, you can specify a known value of the input delay (nk). If nk is unknown, choose a sufficiently large value for the number of coefficients of B (nb). The number of leading coefficients of the estimated B polynomial that are close to zero represent the input delay.

  • Data preprocessing — For offline estimation data preprocessing, you can use functions such as detrend, retrend, idfilt, and the System Identification app.

    For online estimation using Simulink, use the tools available in the Simulink environment. For more information, see Preprocess Online Parameter Estimation Data in Simulink.

    For online parameter estimation at the command line, you cannot use preprocessing tools in System Identification Toolbox™. These tools support only data specified as iddata objects. Implement preprocessing code as required by your application. To be able to generate C and C++ code, use commands supported by MATLAB® Coder™. For a list of these commands, see Functions and Objects Supported for C/C++ Code Generation — Category List.

  • Resetting of estimation — You cannot reset offline estimation. Online estimation lets you reset the estimation at a specific time step during estimation. For example, reset the estimation when the system changes modes or if you are not satisfied with the estimation. The reset operation sets the model states, estimated parameters, and estimated parameter covariance to their initial values.

    To reset online estimation in Simulink, in the online estimation block dialog box, on the Algorithm and Block Options tab, select the appropriate External reset option. At the command line, use the reset command.

  • Enabling or disabling of estimation — You cannot selectively enable or disable offline estimation. You can use preprocessing tools to remove or filter certain portions of the data before the estimation. Online estimation lets you enable or disable estimation for chosen time spans. For example, suppose that the measured data is especially noisy or faulty (contains many outliers) for a specific time interval. Disable online estimation for this interval.

    To enable or disable estimation in Simulink, in the online estimation block dialog box, on the Algorithm and Block Options tab, select the Add enable port check box.

    At the command line, use the EnableAdaptation property of the System object created using online estimation commands, such as recursiveARMAX and recursiveLS. Even if you set EnableAdaptation to false, execute the step command. Do not skip step to keep parameter values constant because parameter estimation depends on current and past input/output measurements. step ensures that past input-output data is stored, even when it does not update the parameters.

More About

Was this topic helpful?