Predictive control is practically the most relevant advanced control method for multivariable dynamical systems with relatively slow dynamics, however challenges remain with implementation for processes with fast dynamics, due to the computation time of on-line optimization. So far, we have developed a fast predictive controller for plasma magnetic current and shape control for the ITER tokamak, where the on-line optimization problem is being solved in the millisecond range. With the improvement of the optimisation algorithm for quadratic programming and its implementation we aim to speed up its execution, so that such controller could be experimentally tested in dynamically faster medium-sized tokamak reactors or in other multivariable control problems with fast dynamics.
Dr. Samo Gerkšič
Samo Gerkšič works at Department of Systems and Control. He is involved with the development of advanced control algorithms, mostly predictive control, and with control of challenging dynamical systems, where recently the emphasis was on control of nuclear fusion and fission reactors and hydrogen systems.
Research programme: Systems and Control
Training topic: Fast predictive control