基于多时间尺度数字孪生的升压变换器健康与故障监测

Component degradation in power electronic converters poses a serious threat to system reliability. This degradation is linked to the health of individual components, potentially causing a butterfly effect that leads to reduced lifespan or faults in the power electronic system. This article presents a multitime scale digital twin (DT)-based condition monitoring approach, featuring enhanced pseudoheuristic health monitoring (HM) and pattern recognition-based fault monitoring (FM), demonstrated in a boost converter application. The proposed concept utilizes the application of condition monitoring using top switch/diode voltage and inductor current for improved health estimation and faster fault detection. It should be noted that the degradation of the deviceon-resistance R_ ds,{ {ON}} and inductor series resistance R_L similarly impacts inductor current and output voltage, making it challenging to monitor individual degradation. This article comprehensively analyzes different types of faults and experimentally validates on a low-voltage SiC-based converter prototype. The proposed FM algorithm is independent of operating conditions and sensor integrity issues, as the significant pulsating voltages across the device provide patterns with enough tolerance for robust FM, thereby enhancing fault detection compared with the inductor/switch current threshold-based fault detection algorithm in the existing literature. Finally, this article delves into design considerations and challenges related to adding a voltage sensor across the top device and associated control complexities.