AIB-4 "AI-based Security Analytics for Smart Grid Systems"
Abstract
ABSTRACT
Security is a key challenge in the deployment and running of Cyber-Physical System (CPS) based Smart Grid (SG) networks. Nowadays, different smart Internet of Things (IoT) sensors and transducers are tightly integrated with the grid-tied inverter and converter. This integration generates many complicated problems of unknown threats that could come in unconventional ways. Developing a combined cyber and physical security strategy to better protect the increasing attack surface is a key research challenge. A cyber-attack on a SG would have devastating effects on the reliability of widespread infrastructure given the potential cascade effects of shutting down the power grid. The use of Artificial Intelligence (AI) and related technologies in the power sector allows for communication between SGs, smart meters, and Hardware-in-the-Loop (HIL) devices. In this project, we proposed to use a long short-term memory (LSTM) recurrent neural network model to detect cyber and physical threats in a smart grid system. We built a 3-bus and an 8-bus Simulink model in MATLAB and collected data for simulated fault scenarios. To test the validity and long-term feasibility of our proposed model, we tested eight different fault scenarios and observed their voltage, current, and power variations with threats.
Keywords
Cybersecurity, AI, IoT, CPS, MATLAB.
AIB-4 "AI-based Security Analytics for Smart Grid Systems"
University Readiness Center Greatroom
ABSTRACT
Security is a key challenge in the deployment and running of Cyber-Physical System (CPS) based Smart Grid (SG) networks. Nowadays, different smart Internet of Things (IoT) sensors and transducers are tightly integrated with the grid-tied inverter and converter. This integration generates many complicated problems of unknown threats that could come in unconventional ways. Developing a combined cyber and physical security strategy to better protect the increasing attack surface is a key research challenge. A cyber-attack on a SG would have devastating effects on the reliability of widespread infrastructure given the potential cascade effects of shutting down the power grid. The use of Artificial Intelligence (AI) and related technologies in the power sector allows for communication between SGs, smart meters, and Hardware-in-the-Loop (HIL) devices. In this project, we proposed to use a long short-term memory (LSTM) recurrent neural network model to detect cyber and physical threats in a smart grid system. We built a 3-bus and an 8-bus Simulink model in MATLAB and collected data for simulated fault scenarios. To test the validity and long-term feasibility of our proposed model, we tested eight different fault scenarios and observed their voltage, current, and power variations with threats.