Hybrid, adaptive, and nonlinear systems. Overview

Transcription

1 Hybrid, adaptive, and nonlinear systems Overview Introduction week DCSC September 4, 2018 Hybrid, adaptive, and nonlinear 1 / 22

2 Overview Hybrid, adaptive, and nonlinear systems Team members Main research topics & ongoing work My own work Related courses Ongoing research Selected MSc project proposals Hybrid, adaptive, and nonlinear 2 / 22

3 Team members Bart De Schutter Ton van den Boom Simone Baldi Erik Steur Sergio Grammatico Nathan van de Wouw Joris Sijs + 2 postdocs/reseachers + 13 PhD students Hybrid, adaptive, and nonlinear 3 / 22

4 Objectives and research area Development of systematic methods to analyze, monitor, and control complex systems, in particular nonlinear systems hybrid systems, i.e. systems with continuous and discrete-event behavior (switching) large-scale systems and networks consisting of interacting subsystems Multi-level control with coordination within and across all levels Adaptive solutions for control of uncertain systems Focus on both fundamental research and target applications: smart transportation and smart infrastructure in smart cities, biochemical circuits Hybrid, adaptive, and nonlinear 4 / 22

5 Main research topics Model predictive control Multi-level and multi-agent control Hybrid and discrete-event systems Adaptive and reconfigurable systems Nonlinear systems Big data Game theory... control inputs control actions system MPC controller optimization model measurements prediction objective, constraints Transportation networks (rail, road) Infrastructure networks (water, energy, logistics) Smart buildings Biochemical circuits Hybrid, adaptive, and nonlinear 5 / 22

6 Model predictive control Ton van den Boom, Bart De Schutter,... Hybrid, adaptive, and nonlinear 6 / 22

7 Multi-level and multi-agent control Bart De Schutter, Sergio Grammatico,... Divide system along various temporal and spatial scales Multiple control layers, intelligent control agents Objective: coordination within and across all layers Methods: MPC, game-based methods, ant colony optimization high level supervisor slow dynamics large region supervisor supervisor local controller local controller local controller fast dynamics small region Hybrid, adaptive, and nonlinear 7 / 22

8 Hybrid and discrete-event systems Bart De Schutter, Ton van den Boom,... Discrete-event systems Event-driven: state changes due to occurrence of event Examples: queuing lines in supermarket, manufacturing system, railway network max-plus algebra as main modeling framework max: synchronization, +: durations Focus on control (MPC) + analysis + stochastic systems Hybrid, adaptive, and nonlinear 8 / 22

9 Hybrid and discrete-event systems Bart De Schutter, Ton van den Boom,... Hybrid systems Combination of continuous and discrete-event dynamics (switching) Examples: electrical networks (switches, diodes), beer production, distillation column, flexible manufacturing systems, road management on mode. T=f on(t,w) T > T upp T < T low off mode. T=f (T,w) off Hybrid, adaptive, and nonlinear 9 / 22

10 Hybrid and discrete-event systems Bart De Schutter, Ton van den Boom,... Hybrid systems Combination of continuous and discrete-event dynamics (switching) Examples: electrical networks (switches, diodes), beer production, distillation column, flexible manufacturing systems, road management Various frameworks: piecewise affine, mixed-integer models, switching max- plus Focus on control (MPC) + analysis + stability + stochastic systems PWA( x 1, x 2 ) x x Hybrid, adaptive, and nonlinear 10 / 22

11 Adaptive and reconfigurable systems Simone Baldi,... Adaptation and reconfiguration capabilities in control systems Focus on problems where model-based approaches are at stake due to lack of knowledge (uncertainties in system and/or environment, faults,... ) adaptively drive the system toward desired behavior Reconfigurable control systems (detect faults and/or changes in operating conditions) automatic reconfiguration without human intervention, reduce maintenance costs Hybrid, adaptive, and nonlinear 11 / 22

12 Transportation networks Bart De Schutter, Ton van den Boom,... Freeway and urban traffic networks traffic jams & congestion time losses, costs, incidents dynamic traffic management integration of various control measures (speed limits, ramp metering, route guidance,... ) Hybrid, adaptive, and nonlinear 12 / 22

13 Transportation networks Bart De Schutter, Ton van den Boom,... Freeway and urban traffic networks integration of various control measures (speed limits, ramp metering, route guidance,... ) integration of freeway & urban sustainable mobility: reduction of emissions and fuel consumption multiple objectives balance between user & system optimum large-scale traffic networks Hybrid, adaptive, and nonlinear 13 / 22

14 Transportation networks Bart De Schutter, Ton van den Boom,... Intelligent vehicles automated highway systems hierarchical control cooperative intelligent vehicle highway systems + cybercars distributed and multi-level control Railway networks operational management (re)scheduling preventive maintenance service contracting Hybrid, adaptive, and nonlinear 14 / 22

15 Infrastructure networks Bart De Schutter,... Water networks flood prevention irrigation maintain water levels within bounds Electricity networks smart grids energy hubs (gas/electricity) Logistic systems baggage handling container terminals routing and scheduling Hybrid, adaptive, and nonlinear 15 / 22

16 Smart buildings Simone Baldi,... Energy efficiency: climate control Building automation: monitor and manage loads Optimized maintenance: detection and identification of faults... Challenges: address occupants behavior, time-varying loads, weather conditions, uncertain building parameters,... Hybrid, adaptive, and nonlinear 16 / 22

17 Ongoing work Distributed and multi-level control of large-scale hybrid and discreteevent systems Keep on increasing speed and performance of analysis and control methods Increasing emphasis on mixed-integer optimization Bridging gap computer sciences systems and control Smart cities Hybrid, adaptive, and nonlinear 17 / 22

18 Recommended courses Systems & Control courses: optimization in systems and control (SC42055) modeling and control of hybrid systems (SC42075) adaptive control (SC42120) model predictive control (SC42125) knowledge based control systems (SC42050) networked and distributed control systems (SC42100)... Application courses (see list on DCSC website), e.g.: traffic & transportation (MSc Transport, Infrastructure & Logistics) Profile Transportation Networks optimization, stochastic systems (MSc Mathematics)... Hybrid, adaptive, and nonlinear 18 / 22

19 Ongoing research PhD students and postdocs Traffic and transportation José Ramón Domínguez Frejo: Efficient traffic control with variable speed limits Anahita Jamshidnejad: Multi-level predictive traffic control for large-scale urban networks Xiaojie Luan: Train scheduling and maintenance planning Energy Farid Alavi: Robust control of fuel-cell-car-based smart energy systems Jesus Garcia Lago: Development of non-intrusive and intrusive energy management algorithms Miguel Picallo Cruz: Advanced monitoring and control of the electrical distribution grid Tomas Pippia: Robust management and control of smart multi-carrier energy systems Jiangeng Fu: Big data methods for maintenance of smart energy systems Hybrid, adaptive, and nonlinear 19 / 22

20 Ongoing research PhD students and postdocs Fundamentals Erwin de Gelder: Big data approach for scenario-based assessment of automated driving systems Amir Firooznia (*): Integrated distributed control of cyber-physical systems Zhou Su (*): Game-theoretic approaches for service contracting in railway infrastructure maintenance Jia Xu (*): Model predictive control for hybrid systems Hybrid, adaptive, and nonlinear 20 / 22

21 Cooperation with companies Some companies you can do your MSc project with/at: TNO Infraspeed ProRail Océ Technolution Mobile Water Management ORTEC Ministry of Transportation DVS... Hybrid, adaptive, and nonlinear 21 / 22

22 For more information... See web site: Research Contact PhD students and other researchers & professors involved (see slides 6 16 and 19 20) Hybrid, adaptive, and nonlinear 22 / 22