Supervisor:
Professor Frede Blaabjerg

Co-Supervisor:
Assistant Professor Ariya Sangwongwanich
Yongheng Yang (Zhejiang University)

Assessment Committee:
Associate Professor Szymon Bęczkowski, AAU Energy (Chair)
Associate Professor Sergio Busquets-Monge, Universitat Politècnica de Catalunya (UPC), Spain
Thiwanka Wijekoon, Huawei Technologies – Nuremberg Research Centre, Germany

Moderator:
Assistant Professor Subham Sahoo

Abstract:

Medium-voltage DC (MVDC) networks have been attracting more and more attention in various power supply systems due to the advantages of high efficiency, reliability and flexibility. DC-DC converters with high voltage-blocking capability and high step-up ratio to be employed to the MVDC systems are key devices. The two-three (2/3)-level dual-active-bridge (DAB) DC-DC converter with a neutral-point-clamped (NPC) bridge is a promising solution. However, the research for multi-level DAB converters has not been fully explored. Therefore, this Ph.D. project was proposed to improve the efficiency and reliability of the 2/3-level DAB converter by exploring several control strategies.

Three main challenges in terms of efficient and reliable control for the 2/3-level DAB converter are considered in this Ph.D. project: 1) In order to improve the efficiency of the 2/3-level DAB converter by control strategies, the optimum combinations of the control variables should be obtained. Analytical solutions are preferred for automatically regulating the control variables online along with the transferred power or DC-link voltages, especially when the operating conditions vary in wide ranges. However, with the increased number of the control variables in multi-level DAB converters, the difficulty for obtaining the analytical solutions increases and the previous methods for the two-level DAB converter may fail to solve the optimization problem; 2) The previous capacitor voltage balancing methods depend on the identification of the transformer current polarity, which is challenging to be identified with various operating conditions or parameters. In addition, current and power fluctuation will occur along with the changed transformer terminal voltage, which will result in overshoot current during the balancing; 3) The previous Open-circuit-fault (OCF) diagnosis methods for the two-level DAB converter fail to locate the exact faulty switch due to the increased number of switches. In addition, the three-level NPC bridge provides more potential for improving the post-fault performance of the DAB converter, e.g., power-transfer capability, which should be utilized in the fault-tolerant method.

To tackle the above challenges, this Ph.D. project proposes several control strategies: 1) To obtain a generic optimal control strategy for improving the efficiency, a minimum-current-stress control method based on analytical solutions is proposed. In order to simplify the complexity for calculating the analytical solutions, the numerical solutions are first analyzed to determine certain operating modes where the optimal solutions appear. Then, the analytical solutions are obtained by the Karush-Kuhn-Tucker (KKT) conditions. Furthermore, a simplified closed-loop control system is designed based on the analytical solutions to achieve online regulation; 2) Two capacitor voltage balancing control strategies are developed to address the above two issues during the voltage balancing, i.e., current fluctuation and model-based feature. For the conditions where the transformer current can be easily determined, the complementary-switching-state (CSS) method is suitably applied to improve the dynamics during the balancing. During the CSS method, the adverse switching states are replaced by their CSSs. By doing so, the neutral-point charges can be controlled without voltage change during the balancing. Thus, the current and power fluctuation can be effectively suppressed. On the other hand, if the transformer current polarity is challenging to be determined when the operating conditions/parameters change in a wide range, the model-free fixed-switching-state (FSS) method can be applied, where two additional switching states are utilized to decouple the voltage balancing control from the transformer current polarity; 3) To locate the faulty switch, a fault diagnosis method is proposed, where the midpoint voltages of each bridge leg are applied as the diagnosis signals, and the exact faulty switch can be located based on the mean values and duty cycles of the post-fault midpoint voltages. Subsequently, a fault-tolerant control strategy based on the complementary-switch-blocking (CSB) method is proposed to suppress the DC bias and overshoot current resulting from the OCF and bring the post-fault DAB converter back to safe operation. In addition, compared to the traditional fault-tolerant method, the proposed CSB method can improve the post-fault power-transfer capability.