A Robust Controller Design Methodology Addressing Challenges Under System Uncertainty

Yunpeng Si, Nikhil Korada, Qin Lei, Raja Ayyanar

Research output: Contribution to journalArticlepeer-review

Abstract

This paper proposes a generalized design methodology of a robust controller to mitigate the impact of system uncertainty on controller stability and performance which includes steady-state error, disturbance rejection, high-frequency noise attenuation and speed of dynamic response. The first step is to select the weighting functions that bound the transfer functions for the entire range of uncertainty. The second step is to form mathematical representation for both robust stability and robust performance. The third step is to conduct the robust H-infinity controller synthesis to generate the full-order controller, and then carry out order reduction and recheck of the design objectives. The last step is to select an optimized controller based on the multi-dimensional Pareto Front algorithm. The proposed method has been firstly applied to the current controller design of a grid-connected inverter with variable grid impedance, and secondly to the voltage controller design of an LLC resonant DC/DC converter with variable resonant capacitance. The results indicate that the selected optimal H-infinity controller has an overall more satisfactory performance in terms of stability, steady-state error, disturbance/noise rejection capability and dynamic performance, compared with conventional PI and PR controllers when there is a large variation of system parameters.

Original languageEnglish (US)
Pages (from-to)1-16
Number of pages16
JournalIEEE Open Journal of Power Electronics
DOIs
StateAccepted/In press - 2022

Keywords

  • Design Methodology
  • Grid-Connected Inverters
  • LLC Converters
  • Pareto Front Optimization
  • Robust H-Infinity Control

ASJC Scopus subject areas

  • Electrical and Electronic Engineering

Fingerprint

Dive into the research topics of 'A Robust Controller Design Methodology Addressing Challenges Under System Uncertainty'. Together they form a unique fingerprint.

Cite this