MPC with frequency-domain disturbance models
Application of MPC to general vehicle control problems requires good models of the vehicle, of the driver and of the surrounding environment. Good models of the environment can substantially reduce the uncertainty due to the disturbances acting on the system under control. Often, environment models are more naturally derived from data than from physical principles. In some contexts, this is conveniently achieved using frequency domain models, i.e. spectral descriptions of the disturbances. At present, some frequency-domain formulations of MPC have been studied, mostly to enforce constraints on the spectra of some output variables. The objective of the activity is to study an MPC formulation where frequency domain descriptions of the disturbance inputs are available. Connections with stochastic MPC formulations are envisaged and will be investigated as well.