Walter Scott, Jr. College of Engineering

Prajay Vora
M.S. Final
Jun 24, 2021, 2:15 pm - 4:15 pm
NA
Control design for generator of Nonlinear High Frequency Plasma System
Abstract: This document aims to develop control systems for a generator of a nonlinear high frequency plasma system. Initial modelling was done by Advanced Energy Industries, Inc. (AE) which was passed on to Colorado State University environment for further research into developing controllers for this special model. This thesis documents all the work done by Colorado State University till Summer of 2020. The first phase of the collaboration included finding metrics for the feedback system with the nonlinear load modelled by AE. The metrics serve for better understanding of the modelling and also to generate effective control criteria suited to AE requirements. AE required for the user defined wave-forms to be tracked in an average sense without significantly changing the real time tracking criteria. This tradeoff was also addressed while developing metrics. A preliminary approach for control design was a PID controller to study its effects in a nonlinear environment. A robust control approach called $H_\infty$ loop-shaping is the primary control design developed by CSU for this specific application. The nonlinear system was approximated with a transfer function and the controller developed for that approximation. The purpose of the approximation is to generate a controller that is highly robust considering the uncertainties in high frequency plasma loads. The metrics discussed above are used for confirming the efficiency of the controllers. Controller design was the second phase of the project. Finally, in phase three, Nelder-Mead optimization was used to generalize the $H_\infty$ controller for various generator and set-point specifications. A system identification processes was also developed consisting of curve fit models for the nonlinear load. This was done with a view to the future for classifying different loads and plasma to develop customised controllers.
Non-ECE Member: Mechanical Engineering, Prof. Dan Zimmerle
Member 3: Electrical Engineering, Prof. Jim Cale