Research On AC/DC Matrix Converter Applied In Battery Energy Storage Systems

As the technical support of renewable energy and smart grid, energy storage technology will play an increasingly important role in the power system. Power conversion system(PCS)—the bidirectional power converter is the interface between the energy storage device and the grid, and is the core component of an energy storage system. It is responsible for active power exchange with the grid and providing reactive power support in grid-connected mode, as well as feeding local load in stand-alone mode. AC/DC matrix converter is a novel converter, not only has the advantages of sinusoidal input current, controllable power factor and four quadrants operation capability, but also can realize single stage large range buck conversion. Therefore, in this dissertation, the AC/DC matrix converter is chosen as the PCS of battery energy storage system, and study is focused on the modulation strategy, the control strategy and the stability.The space vector modulation pattern has large influences on the waveform quality and the efficiency. Aimed at the problem that the analytical method is not intuitive to analyze the DC current ripple and the expression of the ripple is hard to derive, a graphical analysis method is proposed. The relationship between different modulation pattern and the DC current ripple is analyzed. An optimal modulation pattern which has the minimum DC current ripple under arbitrarily power factor angle condition by dynamically allocating the duty cycles of two zero vectors is proposed. The general expression of the input voltage ripple is derived, and the analysis indicates that the optimal pattern has the minimum input voltage ripple when the power factor angle is small and has suboptimal input voltage ripple with reduced switching when the power factor angle is very large. The probabilities of narrow pulse for different modulation patterns are analyzed, and it is indicated that the optimal pattern always has the minimum narrow pulses in a wide modulation range except very low modulation area. Considering the above factors, the optimal pattern has the best overall performance. Simulation and experiments are carried out to verify the correctness of the conclusions and the validity of the optimal pattern.For the grid-connected mode and the stand-alone mode of the battery energy storage PCS, the corresponding control strategies are proposed. The characteristics of the power factor angle and the reactive power are analyzed. For the grid power control in grid-connected mode, a grid current indirect control strategy with DC current loop plus grid power factor angle loop is proposed, and the strategy is easy to implement and has fast active power response. However, in this strategy the grid active power is controlled indirectly, and the reactive power control is dynamically coupled with the active power. In view of these issues, a grid current direct control strategy with dq axes grid currents loop plus DC current loop is proposed. For the stand-alone mode, a load voltage control strategy with dq axes load voltages loop plus DC current loop is proposed. The validity of the proposed control strategies are verified through lots of experiments.The finite-control-set model predictive control strategy is applied to the grid current control of the AC/DC matrix converter based PCS. The grid current and the DC current are chosen as the performance indicators. The DC current reference generation method based on steady state power balance is analyzed in-depth, and the disadvantages of this method that grid current overshoot generates easily due to the lack of consideration of the dynamic coupling of the currents and the reference error exists because of the dependency of accurate parameters are indicated, and then an improved method that based on the dynamic power relationship and using grid current feedback to correct the error is proposed. The Luenberger observer is utilized to obtain the input voltage, therefore the sensors are reduced, the cost is lowered, and the reliability is enhanced. A switching state combination preselection method based on the input current sector is proposed, and the calculation time, the switching losses, as well as the output dv/dt are reduced. The above methods are verified by experiments. At last, the different discretization methods of the predictive model are analyzed, and the influence of the model parameters mismatch on the control performance is discussed.The stability of the AC/DC matrix converter based PCS is the fundamental guarantee of the system. The input impedances of the matrix converter feeding constant power load is derived, and the mechanism of its impact on the system stability is analyzed. The stability of the AC/DC matrix converter open loop system using the input voltage and the grid voltage as the modulation reference voltage vector are analyzed respectively, and it is indicated that the system stability is poor when the input voltage is used as reference since the system is approximate constant power operation. Aimed at the method using the input voltage as the modulation reference, the entire small signal model of the PCS is built, the influence of the parameters of the input voltage amplitude filter, the phase locked loop, and the DC current loop as well as the power factor angle on the system stability is analyzed quantitatively. It is indicated that increasing the time constant of the input voltage amplitude filter and reducing the cutoff frequency of the DC current loop both can enhance the stability in charging mode, reducing the bandwidth of the phase locked loop can enhance the stability in discharging mode, and smaller power factor angle corresponds to more stable system in both charging and discharging modes. The correctness of the modeling and the analysis are verified through simulation.