Hybrid Machine Learning based False Data Injection Attack Detection and Mitigation Model for Waste Water Treatment Plant

• Published in: 2022 International Conference on Automation, Computing and Renewable Systems (ICACRS) pp. 674 - 680
• Main Authors: Parvathy, A., Leela Kasyap, G., Venkata Abhinav, D., Surya Sai, A.N.V., Sriranjani, R., Hemavathi, N.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 13.12.2022
• Subjects: Anomaly Detection; Auto Encoder; Cyber-attacks; Data models; False Data Injection Attack; Gradient Descent; K-Means; Linear regression; Machine learning; Machine learning algorithms; Neural networks; Ordinary Least Square; Prediction algorithms; Renewable energy sources
• DOI: 10.1109/ICACRS55517.2022.10029198

The industries are deploying Internet of Things (IoT) for complex applications in their respective domains which employ different sensors to collect data from different plants and pass it over signal conditioning units to obtain desired results. Now-a-days, these plants are vulnerable to cyber-attacks and one of prominent attacks is False Data Injection Attack (FDIA) which means injecting false data into the sensor readings, which has considerable losses that may include loss of human lives, equipment, etc. Hence, the proposal aims to develop a Machine Learning Algorithm which can protect the clean data from corrupted data and are useful for Anomaly Detection and Data Cleaning. False data injection is detected through three machine learning algorithms namely, Linear Regression with Least squares method (OLS) and Gradient descent and K-Means clustering and the suitable detection model is obtained through comparison of performance indices. Further, Auto Encoder neural network is applied to reconstruct the true data from the false data.