Disturbance-estimator based dead-beat current controller for grid-connected single-phase power electronic converters
University of New Brunswick
Over the past few years, several standards and industrial codes have been introduced to address the power quality requirements in power systems. One of the main sources contributing to poor power quality is the voltage and/or current harmonic distortions that are introduced in distribution networks of power systems. In general, voltage and/or current harmonics are generated by non-linear and switched systems, in particular, gridconnected power electronic converters (PECs). Good examples of grid-connected PECs include reactive power and voltage compensators, and distributed generation units. The literature reports several designs for controllers developed to operate grid-connected PECs for meeting the standard power quality requirements. Among the popular designs for these controllers are: the proportional-integral (PI), fuzzy logic (FL), artificial neural networks (ANN), recursive repetitive controllers (RR), proportional resonant (PR), and predictive current controllers (PC). Predictive current controllers have demonstrated several advantages for applications in grid-connected PECs. These advantages include fast, dynamic, and accurate response, along with stable digital implementations. However, existing implementations of predictive current controllers suffer from degraded performance due to the time-delay introduced by the system requirements. Several methods have been proposed to overcome such a limitation and improve the overall performance of predictive current controllers. These methods can be classified based on their objective, which include reducing the sensitivity to parameter variations, compensating for the time-delay, and estimating and rejecting disturbances experienced by the controlled system. The employment of these methods have offered good improvements in the performance of predictive current controllers. Nonetheless, the complex implementation together with the requirements for additional measurements remain challenges for the applications of predictive current controllers in grid-connected PECs. This research work aims to analyze, develop, and test a new approach for improving the performance of predictive current controllers used in grid-connected dc-ac PECs. The proposed approach employs an observer feedback to minimize the controller sensitivity to variations in system parameters and a disturbance estimator in order to decouple, estimate, and compensate for disturbances, as well as to reduce the steady-state error. The design of the controller, observer and its disturbance estimator are carried using the pole-placement method. Its stability and performance analysis is conducted both on the simulation and experimental levels. The performance of the developed current controller is experimentally tested for a 10 kW interconnected system under different loading and disturbance conditions. In addition, other controllers are tested to highlight the advantages of the developed current controller. Performance and comparison results show accurate, fast, and robust responses that are initiated with negligible sensitivity to parameters variations and disturbances on the grid side.