CC Kick-Off CC Kick-Off MeetingMeeting
Grenoble 24-25/1/2002Grenoble 24-25/1/2002
CC: PartnersCC: Partners
• VERIMAG (Oded Maler)• ETH Zurich (Manfred Morari)• Lund (Anders Rantzer)• PARADES (Alberto SV)• CWI (Jan van Schuppen)• ABB (Eduardo Gallestey)• EDF (C.-M. Fallinower)• Siena (Alberto Bemporad)
CC: GoalsCC: Goals• Develop new techniques for
(computer-aided) control systems design.
• Complex, heterogeneous systems, hybrid dynamics.
• Case-studies (engine control, power production, power transmission).
• Tools (optimization based, reachability based).
MotivationMotivation• Modern complex systems have different
components• They cannot be all described using the
same mathematical formalism• Nor solved using the same techniques• This is already done by practitioners
without theoretical coverage• Combination of control and verification
techniques
Work-PackagesWork-Packages• SC: Survey of Hybrid Control• CH: Control of Hybrid systems• RM: Reachability-based Methods• TL: Tools• AA: Automotive Applications• PP: Power Production and
Transmission
Package inter-dependencePackage inter-dependence
SC: Survey of Control (1)SC: Survey of Control (1)Many different application domains of control,
each characterized by:• Economic importance• Type of models and techniques (speed,
coupling of variables, precision of models)• Dominating culture (CS, control, specific
engineering)• Relevance of theoretical results• Use of computers in design and
implementation
SC: Survey of Control (2)SC: Survey of Control (2)• Investigating questions related to the
implementation of controllers by complex hardware/software system
• Unified models to describe both physical systems and their digital controllers (themselves expressed in a variety of formalisms).
CH: Hybrid Control (1)CH: Hybrid Control (1)In general, extensions of various
results and techniques from control to treat systems with discontinuities (switching systems, piecewise-linear systems, etc.)
Covers mostly the intentions of partners ETH, Lund and CWI
CH: Hybrid Control (2)CH: Hybrid Control (2)• System-theoretic foundations (minimal
realization of piecewise-linear systems)• Identification of piecewise-affine systems• Controller synthesis via (mixed integer-
linear) optimization, model-predictive control
• Control on polytopes• Optimization of switching for discrete-time
systems
RM: Reachability Methods RM: Reachability Methods (1)(1)
Adaptation of discrete verification techniques to continuous and hybrid systems. Essentially computing everything that can happen to a system under all admissible disturbances and controls.
The main effort of Verimag in the project
RM: Reachability Methods RM: Reachability Methods (2)(2)
• New techniques for non-linear systems• Representation of non-convex polyhedra• Goal reachability and optimality• Strategies for differential games• Combination of simulation and verification• Abstraction techniques for treating large
systems
AA: Automotive ApplicationsAA: Automotive ApplicationsA case-study concerning the design of
engine controllers. Models have inherent non-linear and hybrid aspects (sparks, driver commands) and need new design methods.
Current models are too complex for current tools and will motivate their improvement.
Provided by PARADES (affiliated with MM)
Power GridsPower GridsA case-study about the modeling and
analysis of power grids that need to transport power according to varying demand.
Provided by ABB in collaboration with ETH
Power PlantsPower PlantsTwo case-studies provided by EDF:
• Water level controller for a nuclear plant
• Optimization of combustion in a fossil power plant. Essentially switching between different choices of actutors.
TL: ToolsTL: Tools• Implementing the developed
algorithms• Hysdel: the ETH tool for optimization-
based verification and synthesis• d/dt : Verimag’s tool for reachability
analysis• Connection with Simulink (?)
SummarySummary
It will be very nice if we fulfill some percentage of these promises and this will constitute a real contribution to control system design.