| KEYWORD |
MORE Lab _ http://www.morenergylab.polito.it
Model reduction by least-squares moment-matching for control of wave energy conversion systems
keywords CONTROL SYSTEMS, MODEL REDUCTION, MODELLING, SYSTEM DYNAMICS, WAVE ENERGY - HYDRODYNAMICS - NUMERICAL MODELLING
Reference persons NICOLAS EZEQUIEL FAEDO
Research Groups MORE Lab _ http://www.morenergylab.polito.it
Thesis type MASTER THESIS
Description Wave energy conversion devices, commonly referred to as wave energy converters (WECs), need to be
controlled in order to maximise the energy extraction from the ocean wave resource, hence directly
lowering the associated levelised cost of energy.
Control for WEC systems departs from standard regulation/tracking objectives, commonly employed in
control engineering: The objective is that of maximising energy extraction, and not that of
following/tracking a given set-point/reference. As such, the vast majority of the WEC control techniques
employ lie within the field of optimal control theory, where an associated optimal control problem
(OCP) is solved in real-time to compute the corresponding control action.
OCPs are virtually always model-based: That is, a dynamical model of the WEC system is required in
order to predict future motion, enforce constraints, and maximise the energy objective. These models
need to be parsimonious in terms of both computational and analytical complexity, in order to facilitate
real-time calculations, i.e. to be implementable.
Nonetheless, being the Navier-Stokes equations the starting point for WEC modelling, computing
control-oriented models can be a daunting task. As such, tools from the field of model reduction have
been recently applied, to provide dynamical structures which feature a ``simplified'' form compatible
with real-time control requirements, while still representing the main dynamics of the WEC system.
This project will explore the use of model reduction techniques by moment-matching, in a least squares
sense. Moment-matching-based models are essentially interpolating structures, which can match the
steady-state output of a given target model, for a defined class of input signals. These reduced models
have the capabilities of representing both short- and long-term WEC behaviour, being ideal for control
and performance assessment purposes, respectively.
See also main.pdf
Required skills Linear algebra, fundamentals of physical modelling, transfer functions, state-space systems.
Deadline 28/08/2024
PROPONI LA TUA CANDIDATURA