PhD Thesis

Centralized and distributed predictive control of inland navigation networks, including fault diagnosis capabilities

Work default illustration



  • Started: 01/12/2015
  • Thesis project read: 05/12/2016


This thesis regards the problem of optimal management of water resources in an increasingly constrained context. In particular, the attention is focused on inland waterways, which are large-scale systems, generally composed of several interconnected diversion bays, and characterized by slow dynamics and large time delays.
To achieve the optimal management, the efficient control of the structures must be ensured. Since centralized approaches are often impractical to implement in the case of large-scale systems, a distributed model predictive control approach is proposed to solve the problem. This control strategy is characterized by a number of features that make it very suitable to use for inland waterways. In parallel, the problem of state estimation must be also tackled in order to design the controller. To this end, Moving Horizon Estimation, which can be regarded as the dual technique of MPC, is applied. Since both rely on the use of a model, it is necessary to develop a modeling approach that characterizes the dynamic behavior of the system with accuracy.
These techniques will lead to a satisfactory system performance provided that sensors provide reliable measurements and actuators are not impacted by faults. Thus, supervisory methods that allow diagnosing the occurrence of faults in the system are also carried out.