Multiple model nonlinear control of synchronous generators

Abstract

Research has shown that conventional adaptive control and more sophisticated nonlinear approaches can suffer from a lack of robustness and transparency. In this paper, a multiple model technique is proposed that effectively incorporates established linear control methods within an inherently nonlinear framework. The resulting local model network (LMN) was found to describe the nonlinear dynamics of a laboratory turbogenerator excitation loop over its useful operating range. A simple nonlinear generalized minimum variance (GMV) controller was then designed for this system and compared with an adaptive GMV controller. The excellent control performance and transparent nature of the LMN illustrates the potential for this technique for a range of nonlinear control problems.