Localization of Open-Circuit Faults in GaN-Based Three-Phase Dual Active Bridge Converters with Reduced Sensing Requirements

The paper presents a cost-effective method for detecting and localizing single-switch open-circuit faults (OCFs) in three-phase dual active bridge converters using the αβ transformation. The proposed algorithm reduces the number of required current sensors and simplifies the detection and localization processes compared to state-of-the-art methods. The algorithm is verified using an experimental prototype operating at 400 V delivering 2 kW, using 650-V GaN devices switching at 300 kHz. Successful detection of all single-switch OCFs is demonstrated, with a worst-case detection time of 5 switching periods, verifying the effectiveness of the proposed algorithm. An auto-correction process to compensate for dc biases in the detected signals due to variations in system and component parameters is developed, which dynamically optimizes the detection time in different operating conditions. Fast OCF detection, as enabled by the proposed method, allows for the reconfiguration of the converter for post-fault operation.