Realtime Detection of PMU Bad Data and Sequential Bad Data Classifications in Cyber-Physical Testbed

Modern Smart Grids incorporate physical power grids and cyber systems, creating a cyber-physical system. phasor measurement units (PMUs) transmit time synchronized data from grid to the The System Operator (SO) in layer analyzes both online offline format ensures reliability security of by sending necessary command back PMUs. However, various events such as line ground faults, frequency events, transformer well cyberattacks can cause deviation measurements received SO, which be termed "bad data". These bad turn SO take wrong restorative/ mitigating strategy. Therefore accurate detection identification correct type is ensure grid’s safety optimal performance. In this work we proposed realtime sequential classification At first, have exploited low rank property Hankel-matrix detect occurrence realtime. Secondly, classify into two categories: cyberattacks. algorithm utilizes difference approximation error multi-channel before after random column permutations during events. If identified cyberattack, our proceeds identify cyberattack. We considered possible cyberattack types: false injection attack (FDIA) GPS-spoofing (GSA). observes rank-1 single-channel Hankel matrix containing unwrapped phase angle distinguish FDIA GSA. Finally, implemented testbed PMU simulator openECA. Results using IEEE 13 node test feeder show that choosing optimum parameters Hankel-matrix, detected correctly within less than 1 sec. event or shows 100% accuracy for data-window greater 140. Bad classified either with perfect length 73 threshold 0.1. A between 80 120 GSA FDIA, while varying shift 0.1° 0.5°. model also verified 118 bus system simulated SIEMENS PSS/E. Due more complicated structure, requires computational type, however still 2 sec, perform small 40.