A novel decentralized dynamic state estimation methodology for effective frequency monitoring in smart grids
J. Booma, B. Meenakshi Sundaram, S. Suresh, K. Karthikeyan
Abstract
Renewable energy and active, bi-directional smart grids have changed power networks. These developments create operational and control concerns, including uncertainties that inhibit monitoring. Due to lower system inertia, electronic power generation suffers frequency stability issues. Monitoring frequency helps discover stability problems and avert cascading failures and blackouts. A Smooth Variable Structure Filter (SVSF) and sophisticated filtering techniques are used to provide a robust Dynamic State Estimation (DSE) solution for frequency monitoring in high-renewable power systems. Establish a DSE framework that uses deep learning techniques, specifically feed-forward Artificial Neural Networks (ANN), to improve state estimation accuracy; test this DL-based methodology in detecting FDEs caused by line outages, load changes, and various faults; and evaluate the proposed methodology using a comprehensive simulation framework that replicates rea Comprehensive simulations show that the DSE improves system reliability and performance with an average Mean Absolute Error (MAE) of 0.05 compared to 0.15 for the Kalman Filter and 0.12 for the UKF. Our findings demonstrate enhanced FDE identification, frequency monitoring and control, and the potential of current machine learning in power system operations. This research addresses renewable energy integration issues and uses advanced estimate methods to propose a new power system stability paradigm.