Non-Parametric Identification Beyond the Nyquist Frequency

Fast-dynamics models are essential for control design, specifically to address intersample behavior. Traditionally, the inputs and outputs of a system are sampled equidistantly at a reduced rate, and consequently identified at the reduced rate. This poster presents non-parametric time and frequency domain approaches to identify the system beyond the Nyquist frequency of the output, if the input is controlled at a higher rate than the output. Examples of such systems include vision-in-the-loop and chemical systems. The methods utilize respectively kernel regularization and local models, resulting in accurate single experiment identification. Both methods demonstrate their effectiveness on an experimental setup.