Fault-Aware Compression for High Sampling Rate Data Acquisition in Smart Grids.

High rate sampling of voltage and current signals is required for data acquisition in smart distribution networks to enable fault diagnosis at the center. On the one hand, it is challenging for hardware-constrained sensor platforms to handle and process high volumes of sampled data. On the other hand, the requirement of high sampling rate is only imposed to capture higher harmonics arising during faults, and a much lower sampling rate will suffice for recording data during normal operation. We propose fault aware compressive sampling, FACS for short, which is an algorithm for enabling high rate sampling of data in smart distribution systems. FACS takes advantage of dichotomy in sampling rate requirement and seamlessly switches to a lower resolution for fault-free cycles. This poster reports performance of a simple realization of our general algorithm implemented on a low-cost Parallela board. Our implementation could handle 32kHz sampling rate while losslessly compressing real-time, streaming measurement data using less than 10ms per cycle. We have benchmarked the compression and fault-detection performance of our algorithm for synthetic fault data.