Automatic detection of data manipulation in power systems

Abstract

Securing and protecting a power grid system is critical because a power grid transmits and distributes power to millions of people across a country. One of the significant topics in this field is having a real time model that monitors and controls power system grids. A robust monitoring system can be built based on State Estimation (SE) techniques especially when dealing with non-linear structures such as power systems. In order to have a secured power network, the data acquired by the Supervisory Control And Data Acquisition (SCADA) systems has to be reliable and consistent. This is achieved by enforcing false data detection methods where a malicious interference can be detected. Such methods can differentiate between rubbish data and an intrusion attacking the network. Neural Networks (NNs) are considered one of the widely used techniques in detecting false data injections. This work develops a strategy for automatic detection of data manipulation in a power system network. The main contribution is to introduce a Neural Network (NN) based system that can detect any data manipulation whether bypassed by the state estimators or not. The model is capable of detecting the intrusion with a minimum of three random meters in a grid being manipulated. This provides power system operators the ability to take the required decision before a large-scale attack can occur.