Large-scale water resources systems are formed by interacting sub-systems, whose different spatial and temporal characteristics make them strongly heterogeneous. Additional complexity is posed by their large dimensionality, the presence of non-linear phenomena, the uncertainty associated to disturbances and the presence of multiple water users. Modelling and control of these systems thus calls for the development and adoption of innovative methods.
The complex sensors networks and/or the process-based models typically adopted in the water-resources community often provide large amounts of data that cannot be analysed and properly employed with traditional system identification techniques. This calls for the adoption of efficient methods, as feature extraction, variable selection or emulation modelling techniques. Moreover, the distributed nature of such systems is a further reason for the increasing interest in multi-agent systems modelling. On the control side, the large dimensionality of these systems makes the traditional, centralized optimal control approaches hardly usable, thus requiring for the adoption of novel techniques, as reinforcement learning and distributed control.
The aim of this session is to provide an active forum in which to discuss the integration and appropriate application of novel methods, algorithms and methodologies for efficient modelling and control of large-scale water resources systems. The proposed topic will include a large spectrum of methodological and practical activities.
The main topics will focus on the following areas:
Applications can belong to any area of large-scale water resources systems: water quality modelling and control, rainfall-runoff modelling, control of large-scale systems, optimal planning of water resources, analysis of hydro-meteorological data sets, etc.