REAL-TIME IMPLEMENTATION OF A MIMO ADAPTIVE POWER-SYSTEM STABILIZER

Power systems are non-linear and they are often subjected to random disturbances. Therefore stochastic controllers with on-line system identification are ideally suited to power system control problems. Experience with a real time implementation of an adaptive power system stabiliser to damp the dynamic oscillations of a power system is presented. A multi-input multi-output (MIMO) pole shifting control algorithm together with a least-square system identification is used. The system identification is improved using a variable forgetting factor in the recursive least-squares algorithm. The computation time was greatly reduced by streamlining the identification algorithm using the sparse nature of the matrices associated with the computation and by using parallel processing techniques. The controller was tested in real time using a physical model of a power system. The results show that the damping of the power system dynamic oscillations can be improved by using this controller.