Estimation data that contains deficiencies, such as drift, offset, missing samples, seasonalities, equilibrium behavior, and outliers, can adversely affect the quality of the estimation. Therefore, it is recommended that you preprocess your estimation data as needed.
Use the tools in the Simulink® software to preprocess data for online estimation. Common tools to perform data preprocessing in Simulink are:
Blocks in the Math Operations (Simulink) library. Use these blocks, for example, to subtract or add an offset or normalize a signal.
Rate Transition block, which allows you to handle the transfer of data between blocks operating at different rates. Use this block, for example, to resample your data from a source that is operating at a different sampling rate than the online estimation block.
MATLAB Function block, which allows you to include MATLAB® code in your model. Use this block, for example, to implement a custom preprocessing algorithm.
See Online ARMAX Polynomial Model Estimation for an example of how you can preprocess an estimation input signal by removing its mean.