Identification of a discrete-time parametric model of a hyperfast switching Peltier cooling system

The paper addresses the problem of identifying a model of a hyperfast switching Peltier cooling system, which has been proposed in a special benchmark session in Cham et al. (2010). A parametric approach is adopted and the prediction-error method is utilised to determine structures and parameters of discrete-time time-invariant, multi-input single-output, black-box models. Some considerations on the sampling interval chosen are given, based on which low pass filters are designed to pretreat the data. It is concluded that, at least within the operating range under consideration, the overall system can be represented reasonably well via a linear time-invariant model of a relatively low order.