Nonlinear Output-feedback Burn Control with an Optimal Reference Governor and Actuator Allocator

In future tokamaks like ITER, the plasma temperature and density will be regulated using burn control. Due to the hostile environment within the machine, diagnostics will be limited and noisy. Therefore, measurement of the whole plasma state, which is needed for feedback control, may not be possible. Moreover, the plasma response models used for control design may include parameters that are not precisely known. Using online optimization, ITER's suite of actuators can be managed more efficiently, and the plasma can be directed towards desirable operating points without violating operational constraints. The control scheme presented in this work is comprised of five components: a reference governor, a burn controller, an actuator allocator, and two observers. The reference governor optimizes the controller's reference such that the desired operating point can be achieved, and the controller responds by making requests for external heating and fueling. Then, the allocator manages the available actuators to ensure that these requests are satisfied. The two observers provide the controller with estimates of the plasma state and the uncertain model parameters. The performance of the entire burn control scheme is assessed in a simulation study for ITER.