Gain Scheduling With Classification Trees for Robust Centralized Control of PSSs

摘要

This paper takes advantage of wide-area measurements to propose a centralized nonlinear controller that acts on power system stabilizers (PSSs), to cooperatively increase the damping of problematic small-signal oscillations all over the system. Details are given to utilize existing stabilizers while causing minimum changes to the equipment and warranting improvement, or at least no detriment, of current system behavior. A set of gains is designed by linear state feedback control to respond to representative system conditions, and these are coordinated with an equal number of decision trees in charge of scheduling by switching. The resulting nonlinear controller is implemented on a model that resembles the characteristics of the Central American System, exemplifying its feasibility to be applied in large systems. The effect of time delay in the control loop is assessed, and the limits for tolerance to latency in this application are determined.