Project description: Modern control systems put new demands on control theory. Many of the modelling, analysis
and design methods available do not scale well with increasing complexity. Applications and/or industrial practice
often relies on distributed control structures, and there is a strong need for more systematic approaches to design
and analysis of such structures and the corresponding information interfaces, especially with the development of
“internet of things” and the so-called “smart society”.
An important challenge for control and optimization is industrial robots where the task is to plan and carry out
an operation as fast as possible given a number of constraints in terms of accelerations, loads on the mechanical
structure, energy consumption, etc. The constraints in combination with dynamical models of very high complexity
imply a strong need for efficient optimization methods. There are several challenges. One is that the dynamics is
nonlinear making the optimization problem highly non-convex. Another is that re-planning of operations in real
time due to obstacles makes the need for efficient optimization methods much more relevant than before. Current
industrial standard does not allow for re-planning. Optimization for industrial robots has not been considered in
previous ELLIIT projects. The vision is to within 5 years have online optimization routines performing planning
and re-planning of optimal robot trajectories in real time.
Another important challenge for control and optimization is robustness analysis of large-scale interconnected systems
such as power grids. The introduction of renewables in the power grid requires high-fidelity models, which
also imply a strong need for more efficient optimization methods. In this project we will investigate and develop new
optimization methods and software for modelling, analysis and design of large-scale control systems that scale well
with problem size. Within ELLIIT we have previously developed scalable robustness analysis methods assuming
that suitable models where available. For systems like power grids, this is not the case. A major challenge is to
in a distributed manner obtain linearized models for power grids, and to in a distributed manner build so-called
LPV models which capture the uncertainties of the power grid. The vision for 5 years is to have efficient tools for
modelling power-grids based on the Modelica modelling language which admits efficient analysis of robustness of
the grid. This work will be carried out in collaboration with ABB Corporate Research in Switzerland.