Design method for optimized feedforward+feedback shape control

Despite the wealth of experience in plasma shape control, difficulties still persist in pushing the plasma to operational limits and scenarios. We present a design methodology for obtaining feedforward trajectories for a given scenario and combining this with robust feedback control. Feedforward is calculated from the target shape evolution and serves the purpose of identifying a 'global' solution for the full-shot trajectory, balancing priorities throughout the shot, and also compensates for plasma resistivity and current profile evolution. Feedback control is used to compensate for 'local' disturbances and inevitable modeling errors. We use Model Predictive Control (MPC) for the feedback method, and shape transitions are guided by the time-dependent MPC weights. By comparison, many operational control systems use different controllers for each shape type (diverted vs limited etc). Results are validated with the Gsevolve flight simulator and NSTX-U data and performance improvements are demonstrated such as eliminating undesired upper-single-null / lower-single-null switching and operation closer to hardware limits.