FALSE DATA INJECTION ATTACK DETECTION AND EARLY INTERVENTION IN A POWER SYSTEM USING KULLBACK-LEIBLER DIVERGENCE AND MAXIMUM DIFFERENCE

Author

Prayush Khadka, Southern Illinois University CarbondaleFollow

Date of Award

5-1-2023

Degree Name

Master of Science

Department

Electrical and Computer Engineering

First Advisor

Kagaris, Dimitrios

Abstract

As the modern electric grid communicates with each other, the system’s inherent se-curity vulnerabilities start becoming a problem for efficient operation. Various attacks on the system for nefarious means diminish the system’s capabilities. False Data Injec- tion Attack (FDIA) manipulates the meters’ reading and shows inaccurate system val- ues. FDIA negatively affects the electricity grid’s usage, cost, and future planning for all stakeholders. This study seeks to find a way to efficiently detect the presence of false data attacks in the power system and presents a method of early detection of false data injection attacks. Two different processes are employed; Kullback-Leibler Divergence and Maximum Difference. Kullback-Leibler Divergence calculates the distance between two probability distributions, while Maximum Difference measures the largest possible differ- ence between two values. The analysis is done on the load data obtained from New York Independent System Operator (NYISO). The results indicate that the suggested investi- gation presents methods and mechanisms where false data injection attacks of different quantities are detected with high accuracy while having false positives under a reasonable value. The result points out that Maximum Difference is preferable to Kullback-Liebler Divergence for False Data Detection Attack detection. The analysis outcomes also sup- port the described process in early detection of the attack on the power system. Further study can be done to implement the methods in real-time attack detection scenarios.