Voltage Balance Control Based on Dual Active Bridge DC/DC Converters in a Power Electronic Traction Transformer

In this paper, a voltage balance control strategy based on dual active bridge (DAB) dc/dc converters in a power electronic traction transformer (PETT) is proposed. Based on this strategy, the output-parallel DAB converters can be equivalent to an inputseries-output-parallel system. Furthermore, a PETT starting control method is put forward, which can effectively avoid risks of overcurrent and overvoltage in the PETT starting process. In order to carry out the controller design and system stability analysis, three different kinds of mathematical models of DAB converters are set up. The first model is related to a single DAB converter, the second model reflects the equivalent relation between an output-parallel DAB system and a single DAB converter in terms of the output-voltage control loop, and the third model indicates that the voltage balance control system based on DAB converters is a multiinput-multioutput system. Due to the nonzero off-diagonal elements of the controlled plant, there is a mutual effect between different control loops, which is defined as "interaction" in the multivariable feedback control theory. The stability of the voltage balance control system is made up of two parts, including the stability of each single-input-single-output (SISO) control loop and the influence of the interaction on the system stability. The research is carried out to measure the intensity of the interaction in this paper, and a criterion directly based on the controlled plant is proposed to predict the influence of the interaction, which can obviously simplify the system stability analysis. Considering the particular traction onboard application, a new control structure toward the voltage balance controller is introduced. Based on the new structure, the controller is designed and the stability of the SISO system is analyzed. Finally, a five-cell PETT prototype with rated power of 30 kW is taken to carry out further research, and the experiment results verify the effectiveness and correctness of the proposed algorithms.