CAREER: Intelligent Control Systems for Active Magnetic Bearings: An Enabling Technology for Flywheel Energy Storage Systems

North Carolina State University, Raleigh NC

Investigators

Abstract

This project will explore the use of novel learning-based intelligent control to meet the challenge of controlling Active Magnetic Bearings (AMBs). If they can be controlled more effectively, AMBs have the potential to make it feasible to use advanced flywheel energy storage systems to signigicantly improve the efficiency of hybrid electric vehicles; this, in turn, could measurably ease the world's transition to higher oil prices. The challenge of controlling AMBs more effectively is well-known to be the main obstacle which needs to be addressed, in order to make this all happen. The development of brain-like learning systems-systems which output contols or actions, as organic brains do-is general fundamental scientific importance. AMB control will be used as a testbed for benchmarking the capabilities of more brain-like designs compared with more traditional designs, and for studying the characteristics of existing brain-like design with an eye to possible improvements. Because the AMB problem is very nonlinear and difficult for traditional methods, there is a significant possiblity that novel methods will perform better (after considerable analysis) and that results of this project could assist in the difficult paradigm shift to learning-based nonlinear approaches.