Handling uncertainty with application to indoor climate control and resource allocation planning

摘要: The thesis focuses on the stochastic control, ie control of systems involving uncertainty and whose states are represented by a stochastic process. The basic idea in stochastic control is how to incorporate uncertainty in the control action, ie how to model it. Several approaches are applicable, whose performances mainly depend on the specific application. In particular, the thesis deals with two attractive approaches to handling uncertainty in the control problem: the stochastic approach and the robust approach. Both these methods are illustrated through an industrial application. The robust approach is applied to classical problem of multistorage system control. An external uncertain demand is satisfied by using items stored in buffers. The demand, whose probability distribution is unknown, is assumed to be unknown but bounded inside given constraint sets. The Robust Invariant Set Theory, a specific theory broadly used in robust control, is exploited to find a control law and the initial buffer levels guaranteeing that an unknown bounded demand is always satisfied. The proposed control method is applied to a manufacturing plant located in South Italy: simulation results prove its efficacy and the reliability. The stochastic approach is employed in the building climate control. The goal is to develop predictive control strategies to save energy in indoor climate control while maintaining high user comfort. The disturbances mainly consist of the weather and the people occupying the building. Uncertainties are modeled in a stochastic fashion: the disturbances probability distribution can be approximated well as a normal distribution and then identified. One …