전압형 다상 인버터에서 복잡하지 않는 고성능 모델예측 전류제어기법

Abstract

Over the last two decades, multiphase drives have been developed for various applications, such as naval propulsion systems, more-electric aircraft, electric vehicle traction drives, and offshore turbines due to their advantages over their three-phase counterparts: enhanced fault tolerance, lower torque pulsation, higher torque density, reduced voltage and current per phase for the same power, and lower dc-link current harmonics. Typically, multiphase drives are supplied from a multiphase voltage source inverter (VSI). Some modulation and control methods have been developed for multiphase VSIs, such as space vector pulse width modulation (PWM), carrier-based PWM, and model predictive current control (MPCC). Among these, MPCC is the most effective and simplest current control method for multiphase VSIs due to its simple principle, fast dynamic response, straightforward handling of nonlinearities and constraints, and the capability of simultaneously tackling multiple control objectives.

Despite of these advantages, the MPCC for multiphase VSIs still suffers some problems, namely high computational burden, large low-order current harmonics, time consuming weighting factor tuning process, common-mode voltage (CMV), and neutral-point voltage imbalance. To solve these problems, this thesis presents the development of various low complexity MPCC strategies to drive the multiphase VSIs with high performance in term of current harmonics, the CMV, and the neutral-point voltage.

Firstly, an effective MPCC strategy to simultaneously reduce the CMV and current harmonics for a two-level five-phase VSI is proposed. In the proposed strategy, 10 virtual voltage vectors are utilized as the input control set to reduce the CMV and low-order current harmonics. To reduce current ripples, two virtual voltage vectors among these virtual voltage vectors are applied throughout the whole sampling period to drive the five-phase VSI. The two selected virtual voltage vectors are determined based on the location information of the reference voltage vector, and their duration times are calculated using a simple algorithm. As a result, the computational burden is significantly reduced.

Secondly, a simple MPCC strategy to reduce both the CMV and current harmonics for a two-level seven-phase VSI presented. In the proposed simple MPCC strategy, 14 virtual voltage vectors are proposed, and these virtual voltage vectors are employed as input control set to reduce CMV and low-order current harmonics simultaneously. In each sampling period, the optimal voltage vector is directly selected from the 14 virtual voltage vectors to achieve the best output current performance without a cost function. Therefore, time-consuming and complex selection of suitable weighting factors is avoided, and the computational burden is dramatically reduced compared to that of the conventional MPCC strategy.

Thirdly, this thesis proposes a simplified MPCC strategy to eliminate the CMV and reduce current harmonics for a dual five-phase VSI-fed open-end load supplied with a single DC source. 21 virtual voltage vectors were developed as an input control set for the proposed MPCC strategy to eliminate CMV and reduce the current harmonics. In each sampling period, 5 virtual voltage vectors among 21 virtual voltage vectors are selected based on the location information of the desired voltage vector and evaluated by a new cost function to find the optimal voltage vector. Thus, the computation time is significantly reduced because the current predictions calculation is eliminated, and the cost function evaluation calculations are reduced to 5.

Fourthly, this thesis proposes a simple MPCC strategy for a three-level sparse five-phase VSI (TLSF-VSI) to reduce current harmonics as well as balance the neutral-point voltage. 21 new virtual voltage vectors were introduced, and the optimal voltage vector is determined among these 21 virtual voltage vectors to reduce output current harmonics. The switching states to achieve the balanced neutral-point voltage are selected from the optimal voltage vector without weighting factors and capacitor voltages prediction. Therefore, the proposed MPCC strategy can be implemented simply, and the computation time is significantly reduced.

Simulation and experimental results are given to verify the effectiveness of the proposed strategies.