Robust decentralized controller design via ai to enhance power system dynamic performance

Power systems are modeled as large-scale systems composed of a set of small-interconnected subsystems. It is generally impossible to incorporate many feed backloops into the controller design for large scale interconnected systems and is also toocostly even if they can be implemented. These motivate the development ofdecentralized control theory where each subsystem is controlled independently on itslocal available information.On the other hand, the operating conditions of power systems are always varying tosatisfy different load demands. Control systems are therefore required to have theability to damp the system oscillations that might threaten the system stability as theload demand increases. However, as power systems are large-scale nonlinear systemsin nature, the applications of conventional power system stabilizer (PSS) are limited.There is thus a need for controllers, which are robust to changes in the systemoperating condition. Robust controllers based on HeY;) control theory are particularlysuited for this purpose.This thesis proposes two robust decentralized controllers for multimachine powersystem instead of using a complex centralized controller. The first one is based onH theory, and results in high order controller. The second controller is a0proportional integral (PI) type, and is tuned by a novel robust performance as the firstone, but it is more appealing from an implementation point of view. In more detail,the second control design is first cast into the robust H control design in terms of0linear matrix inequalities (LMI) in order to obtain robustness against system operatingconditions. An additional constraint is that the structure of the controller is predefinedas a PI type, which is ideally practical for industry. In order to obtain the optimalcontroller parameters with regards to the H and controller structure constraints,. . 00genetic algorithms (GAs), a powerful probabilistic search technique is used to find thecontrol parameters of the PI controller.