Model predictive control and smart decision makers applications in uninterruptible power supplies

In this thesis, disturbance observer based model predictive controllers were designed to control the output voltage of a three phase uninterrupted power supply (UPS). The performances of the predictive controllers were evaluated both simulation and experimental model of the designed UPS system. First, a discrete time model of a three phase UPS device was designed, and the model uncertainties were modelled as a polytopic uncertain class using the discrete time of the UPS model. The disturbance effect of the load current and the model uncertainties were combined as lumped disturbances and a robust disturbance observer was proposed to estimate these lumped disturbances. CCS-MPC (Continuous Control Set Model Predictive Control) and FCS-MPC (Finite Control Set Model Predictive Control) controllers were designed based on the proposed disturbance observer and the output voltages of the UPS device were controlled using the proposed controllers. The evaluation results showed that the %THD values obtained with the proposed method were in the range of international standards and it was concluded that the proposed disturbance observer could be used with the model predictive controllers. The control algorithm was written in C language and implemented in simulation model of a UPS in PSIM software using dll (dynamic link library) function. The real-time application was carried out in a 6KW laboratory UPS prototype using TMS320F28377D microcontroller.