Search

In this article, a low complexity finite-control-set model predictive control (FCS-MPC) based on the discrete space vector modulation (DSVM) is proposed for T-type three-phase three-level (3P-3L) converters. Different from the conventional FCS-MPC, 48 virtual voltage vectors (VVs) of the converter are constructed by real VVs based on the DSVM. Thus, the performance of 3P-3L converters is significantly improved and the peak amplitude of high-order harmonics concentrates at the sampling frequency. Furthermore, two-stage FCS-MPC based on virtual VVs is proposed to reduce the computation burden. Its first stage selects one of six virtual VVs that minimizes the current tracking error. Then, these candidate VVs located in the same sector as the optimal virtual VV selected in the first stage are evaluated in the second-stage optimization. Thus, the computational efficiency has been greatly improved. To verify the validity of the proposed control method and show its superiority over the conventional FCS-MPC, experimental results are presented. [ABSTRACT FROM AUTHOR]

The presence of dc offsets results in fundamental frequency oscillations in the frequency estimation and it is a challenging problem in single-phase grids. In this article, a four consecutive samples (4CSs) based method is proposed to estimate grid frequency with strong dc offset rejection capability. In details, a 4CS model is established based on the relationship among the fundamental grid voltage, dc offsets, and their consecutive sample signals. Different from the traditional 3CS model, the proposed 4CS model can reject the negative effects of dc offsets completely. Based on the obtained model, a frequency estimation is proposed and stability analysis is provided according to the adaptive control theory. Unlike the traditional 3CSs relying on prefilters, the proposed method is filter-free and it still has strong suppression capability against dc offsets. The theoretical sound and experimental results indicate that the proposed method achieves accurate frequency estimation with good dc offset rejection capability. [ABSTRACT FROM AUTHOR]

This paper presents a novel model predictive control (MPC) for the T-type single-phase three-level inverters. The cumbersome procedure in tuning of weighting factors is eliminated so that the implementation of the proposed MPC becomes easy. The constant switching frequency is achieved during the control implementation in order to facilitate the filter design. In order to optimize the distribution of the output current harmonics, multiple voltage vectors are employed in each control cycle with their application times setting inversely proportion to the respective cost function. The redundant small voltage vectors are utilized to balance the neutral point voltage of the inverter by directly regulating the upper and lower dc-link capacitors voltages. Finally, the proposed MPC algorithm is experimentally evaluated and compared with other two unfixed switching frequency MPC algorithms in terms of the steady state, transient performance, and parameter sensitivity. [ABSTRACT FROM AUTHOR]

Model predictive control (MPC) suffers from high computational burden and cumbersome tuning of weighting factors especially for three-level three-phase inverters. Here, a fast finite-switching-state MPC (FSS-MPC) algorithm without weighting factors is proposed. First, the deadbeat control is used to construct the voltage vector reference. Then, the T-type three-level three-phase inverter topology is selected and voltage vectors are carefully tuned in order to minimize the neutral-point (NP) voltage fluctuation. The voltage vectors that are far from the desired voltage vector reference and go against balancing the NP voltage will not participate in the cost function optimization. Thus, only 3 effective voltage vectors instead of totally 27 vectors are required in the implementation of the proposed FSS-MPC algorithm, which saves computation time up to 53.3%. Furthermore, the proposed algorithm makes the tuning of the weighting factor unnecessary, which simplifies the practical implementation and improves the portability of the algorithm. Finally, an experimental prototype was established and main results including the steady-state and dynamic performance were presented to validate the effectiveness of the proposed algorithm. [ABSTRACT FROM AUTHOR]