Proactive fault-tolerant economic MPC
Event details
| Date | 23.11.2015 |
| Hour | 15:00 › 16:00 |
| Speaker | Brage Rugstad Knudsen |
| Location | |
| Category | Conferences - Seminars |
Abstract:
Fault tolerance in model predictive control (MPC) of dynamic systems is important for both safety and economic optimality of operations. Among possible fault scenarios, a partial or full loss of actuation may be detrimental on the control performance or in worst case destabilize the system. In this talk, I will present a proactive approach for accommodating incipient actuator faults in linear control systems operated with economic MPC, by means of invariant safety sets and L1 exact penalty functions. I will further present an extension of the scheme to nonlinear control systems and discuss related challenges, as well as connections between the proposed approach and minimum-time control.
Bio:
Brage Rugstad Knudsen is a postdoctoral researcher at the Department of Engineering Cybernetics, Norwegian University of Science and Technology, working on fault-tolerant model predictive control. He received a PhD from the same department in 2014. His main research interests are optimization, receding horizon control, and mixed integer programming with applications to natural gas and energy systems.
Fault tolerance in model predictive control (MPC) of dynamic systems is important for both safety and economic optimality of operations. Among possible fault scenarios, a partial or full loss of actuation may be detrimental on the control performance or in worst case destabilize the system. In this talk, I will present a proactive approach for accommodating incipient actuator faults in linear control systems operated with economic MPC, by means of invariant safety sets and L1 exact penalty functions. I will further present an extension of the scheme to nonlinear control systems and discuss related challenges, as well as connections between the proposed approach and minimum-time control.
Bio:
Brage Rugstad Knudsen is a postdoctoral researcher at the Department of Engineering Cybernetics, Norwegian University of Science and Technology, working on fault-tolerant model predictive control. He received a PhD from the same department in 2014. His main research interests are optimization, receding horizon control, and mixed integer programming with applications to natural gas and energy systems.
Practical information
- Expert
- Free