Publication: Dynamic line rating system based on machine learning and optimal sensor placement method for cyber-physical power networks
Loading...
Date
2024-04
Authors
Olatunji, Ahmed Lawal
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Global access to clean and affordable electricity can be improved with an efficient and reliable transmission network using the Dynamic Line Rating (DLR) system rather than the conventional, fixed line rating known as the Static Line Rating. This research investigates critical reliability challenges in implementing DLR for efficient transmission. These challenges are DLR forecasting, False Data Injection Attack (FDIA), and sensor placements. A new DLR system forecasting algorithm that leverages k-means clustering and Monte Carlo simulation was developed. It surpasses existing algorithms with forecast skill, mostly at 98%, and provides probabilistic ratings for system operators and planners with the risk and benefits attached to each rating. A data-driven classification algorithm based on feature ranking, selection, and optimized Binary Generalized Linear Model-Logistic Regression significantly mitigates FDIA, achieving an accuracy of 0.917 and a perfect false negative rate. In addition, a comprehensive framework for assessing cyber-physical power system (CPS) reliability incorporating the DLR system is established. This Markov process-based framework includes a DLR sensor placement hierarchical clustering algorithm, simulation, computation, and optimization of failure rates through reduced latency significantly improves the CPS steady-state availability. The IEEE 24-bus and IEEE 72-bus network availabilities improved from 58.28% to 99.74% and 52.14% to 99.61%, respectively. These solutions are critical to DLR systems to ensure CPS security and grant more access to clean and affordable electricity.