Fast fault detection schemes developed for dual active bridge converters

Abstract

Dual active bridge isolated converter is receiving increased attention as one of the core technologies in electric vehicle charging and battery storage systems. Consequently, both academia and industry are interested in addressing their existing challenges, such as reliability concerns.

One of the main reasons for converter failure is switch faults, which can be classified into open circuit faults (OCFs) and short circuit faults (SCFs). Although many detection schemes have been developed, existing detection methods often have high implementation costs, complex sensing, and computational burdens, which make real-time processing costly and impractical. Furthermore, most schemes assume switch gate drivers have the capability to detect SCFs and system-level protection schemes for SCFs are not needed, which is not valid for all of the drivers used in the industry.

This thesis proposes a new fault detection and isolation scheme for both OCF and SCF. After a detailed analysis of dual active bridge converter operating principle, midpoint voltages are selected as the fault diagnosis signal. By unifying the fault diagnosis for both schemes, the overall protection cost and implementation complexity are significantly reduced. In the OCF detection scheme, dynamic thresholds are used, which enhance protection sensitivity and effectiveness. The SCF scheme can detect and isolate faults quickly enough to prevent failure and catastrophic consequences. The schemes are validated through hardware-in-the-loop tests. OCF is detected within a switching cycle of 3.33 µs, and SCF is detected in $235$ ns which is significantly shorter than the typical short circuit withstand capability of current industrial MOSFETs. Therefore, by detecting and isolating the faults promptly, the reliability of the dual active bridge converter is improved.