Fast Model Predictive Control for Magnetic Plasma Control - FMPCFMPC

Principal investigator:

Dr. Samo Gerkšič

Duration:

1.1.2015 - 31.12.2017

Funding:

Horizon 2020 / Euratom / EUROfusion / Work Programme 2015 Enabling Research

Abstract:

The challenge of magnetic shape control of unstable elongated plasma in tokamak fusion reactors is to maintain the prescribed plasma shape subject to large-scale disturbances, such as vertical displacement events (VDE), H-L transitions and edge localised mode (ELM) perturbations, and to considerable changes of local dynamics in different operating points.

Model Predictive Control (MPC) is a control methodology in which future control actions are determined by optimisation of a performance criterion defined over a future horizon in which control signals are predicted using dynamic process models. Variants of MPC are considered the advanced process control technology of choice in many industries involving multivariable processes. MPC possesses many features desirable in plasma control systems. However, MPC is currently not applicable to large-scale multivariable systems with fast dynamics requiring sub-second sampling rates as required in plasma magnetic control. Our intention is to develop a new implementation of MPC, where the key part of the optimisation problem involved is solved parametrically or using recently emerged fast MPC techniques. The approach will be applied to the plasma shape and current (PSC) controller and for the control of resistive wall modes (RWM).

Project partners: