Multi-input multi-output APSS for rotor angle stability

A Multi-Input Multi-Output (MIMO) Adaptive Power System stabilizer (APSS) has been developed in this dissertation to improve the dynamic performance of the power system. In this multivariable application, state space model is employed to represent the generator. Online identification of the model parameters and states estimation help to track the dynamics of the low frequency oscillations in the power system. The Generalized Predictive Controller (GPC) is ideal to produce the control signals for its ability to handle the multivariable system with online constraints in a systematic way. Two control signals from the MIMO APSS, V_pss and V_gov, are fed to the excitation loop and governor loop, respectively. Input mechanical torque and output electrical torque are coordinated together to damp the oscillations.; State space model is identified according to a subspace method, MIMO Output Error State sPace (MOESP). Re-decomposing system outputs into orthogonal subspaces under the help of Instrument Variable (IV) helps to get accurate identification results in noise interference environment. Online update of the model parameters is realized through Projection Approximation Subspace Tracking (PAST) algorithm. Multi-step ahead optimization is used in the control signal calculation procedure which considers the dynamic impact of the control signals on the future states of the system. A closed loop predictor is constructed to improve numerical conditions of the optimization problem.; Detailed governor and turbine models are used in the simulation studies compared with the simple governor model in most PSS studies. Comparison studies between the proposed MIMO APSS and the Conventional PSS (CPSS) are conducted on both a single machine infinite bus system and a multi-machine system. The results demonstrate the effectiveness of the MIMO APSS in damping oscillations in the power system. The performance of the MIMO APSS is superior to that of the SISO APSS which ignores the transient dynamics in the governor loop of the generator.; Due to the physical condition limits in the lab, only the SISO version of the APSS is tested on the physical model of a single machine infinite bus system. The SISO APSS has the same model structure and control signal calculation procedure as that of the MIMO APSS except the dimensions of matrices B and C are reduced. The APSS is implemented on a personal computer with a data acquisition card combined with the Real-Time Windows Target and Real-Time Workshop toolboxes from Matlab Simulink software environment. The laboratory test results also show that the prototype APSS based on online updated state space model and predictive controller can provide a better performance than the CPSS.