불평형 그리드전압 상태에서 교류-직류 변환 매트릭스컨버터의 고성능 제어기법

Abstract

In recent years, AC-DC Matrix Converters (MCs) have received much interest in research for AC-DC power conversion due to its significant advantages such as bidirectional power flow, sinusoidal grid current, controllable input power factor, tight DC voltage regulation over a wide range, compact design, and long life. All of these advantages make AC-DC MCs promising for industrial applications such as energy storage systems, AC-DC microgrids, and vehicle-to-grid systems. In practical applications, the three-phase grid voltage are often under unbalanced conditions, which deteriorate the performance of the AC-DC MCs due to the ripple on the DC side and low-order harmonics in the grid current. This thesis presents the development of various advanced control strategies based on space vector modulation (SVM) and model predictive control (MPC) to drive the AC-DC MCs under unbalanced conditions with high performance in term of the output current and voltage ripples, grid current harmonics, and grid power factor. The first approach is based on the SVM technique; an enhanced control strategy is developed based on an independent control scheme for the active and reactive powers. The input current reference is directly synthesized from instantaneous power analysis in a stationary frame, and a constant DC voltage and current and a sinusoidal grid current are obtained without any input current controller. Additionally, a near unity power factor is simply achieved by using an average grid reactive power-based PI controller. The average grid reactive power is obtained through a notch filter without using the grid voltage sequence components, so the proposed control strategy is implemented easily without large storage requirement. The second approach is based on the MPC technique. MPC is applied to AC-DC MCs to achieve fast dynamic response and simple implementation. In spite of these advantages, the conventional MPC (C-MPC) for an AC-DC MC still has various issues to be solved. This thesis present three advanced MPC schemes to deal with various issues of the C-MPC such as computational burden, active damping function, steady state performance improvement, and voltage sensor reduction. First, the thesis presents a simplified MPC (S-MPC) scheme with active damping function. The grid current reference is generated by analyzing the instantaneous power model in a stationary frame without information about the grid voltage sequence components. The prediction process is simplified by computing only one required input current vector (RICV) instead of nine grid current predictions. In addition, a novel method is proposed to realize the active damping function without increasing the computational burden by integrating the damping current into the RICV. Thus, the execution time of the S-MPC is significantly reduced compared with that of the C-MPC, and the system performance is easily improved by shortening the sampling period. Second, a virtual-flux-based MPC (VF-MPC) scheme is developed to remove the grid voltage sensor and enhance the grid current performance. By modeling the power flow in terms of the virtual flux (VF) and its 90o lagging signal, the grid current reference and grid voltage are obtained simply without the extraction of the VF positive and negative sequence components. Furthermore, the number of current vectors is increased from 9 to 39 vectors by generating 30 virtual current vectors to minimize the grid current tracking error. Especially, to reduce the computational burden due to the increased number of current vectors, a preselection method is presented to reduce the number of candidate current vectors from 39 to 8. Third, the thesis develops an improved MPC (I-MPC) scheme to simultaneously compensate the power ripple in the input filter and reduce the grid current distortion by taking into account the power ripple in the input filter under unbalanced grid voltage conditions, which is normally ignored in the MPC of an AC-DC MC. The power ripple is calculated based on the grid voltage and its 90o lagging signal, which makes the implementation simple without grid voltage components extraction or digital filter design. Furthermore, a closed-loop current controller is developed based on a resonant controller (RC) to reduce the grid current harmonic caused by power ripple under imbalanced grid voltage conditions. All control strategies are successfully tested by simulation in PSIM and experiments on prototype AC-DC MC. Proper comparisons are also given to verify the effectiveness of the proposed control strategies.