Nonlinear Adaptive Burn Control and Optimal Control Allocation of Over-Actuated Two-Temperature Plasmas
V. Graber, E. Schuster
American Control Conference
Denver, Colorado, USA, July 1-3, 2020.
Abstract
|
|
Tokamaks are reactors that produce energy from the fusion, or merging,
of atomic particles. A suitable reaction rate is achieved by heating a
gas of charged particles (free ions and electrons), or plasma, to extreme
temperatures. From the fusion of deuterium and tritium ions, a burning
plasma produces alpha particles that contribute to the heating of the
plasma. Burning plasmas are highly nonlinear systems that require
careful regulation of temperature and density, or burn control, to
reach desirable operating points. Once constructed, ITER will be the
first tokamak designed for burning plasmas. In this work, a Lyapunov-based
burn controller is developed using a full zero-dimensional nonlinear
model. An adaptive estimator manages the presence of uncertain model
parameters. The control objective is to stabilize equilibria despite
model nonlinearity and uncertainty. Density is regulated through the
injection of fuel pellets. For ITER, the temperature of the ions may
differ significantly from that of the electrons in the plasma core.
Therefore, the proposed controller considers separate response models
for ion and electron energies. For energy control, the controller
commands two virtual control efforts: the external ion heating and the
external electron heating. To satisfy these two virtual control efforts,
ITER will have access to ion cyclotron heating, electron cyclotron
heating and two neutral beam injectors. With more actuators than virtual
control efforts, the two-temperature plasma system is over- actuated.
Actuator redundancy is resolved by constructing an optimal control
allocator that considers actuator saturation and rate limits. A simulation
study demonstrates the capability of the adaptive control and control
allocation algorithms.