The stable and reliable operation of grid-integrated renewable energy systems requires advanced control and coordination of grid-side converters (GSCs), utilizing the feedback measurements of voltage and current sensors from both the direct current (DC) and alternating current (AC) sides of the converter. However, the effective operation of the converter is susceptible to sensor failures or divergence from their proper operation. Although sensor fault detection algorithms are usually effective under abrupt faults, the fault propagation effect caused by the physical interconnection between the DC and AC sides of the converter may limit the performance of the sensor fault isolation process in revealing the exact location of a potential faulty sensor. Therefore, this work proposes a robust, model-based fault isolation and accommodation scheme. Specifically, a synergistic sensor fault isolation framework based on adaptive estimation schemes is proposed for both single and multiple faults in the DC voltage and AC current sensors, considering modeling uncertainty and measurement noise. The performance analysis in terms of stability, learning capability, and fault isolability is rigorously examined. An accommodation scheme based on a virtual sensor utilizing dynamic sensor fault estimation with real-time learning capabilities is applied to a GSC. Finally, the performance of the proposed fault isolation and accommodation scheme is evaluated through simulation analysis under several scenarios involving single and multiple sensor faults.
Read full abstract