A Study on Distributed State Estimation and Data Processing for Monitoring Operations of Large-Scale Power Systems

Abstract

The power industry is undergoing significant changes in response to climate change and an increasing complexity in the power system. The shift involves a paradigm shift towards an increased penetration of renewable energy resources and demand response (DR), aiming to achieve net-zero carbon emissions. The rise of distributed generations (DGs), such as photovoltaics (PVs), wind turbines (WTs), and hydrogen energy storages (HESs), contribute to increased uncertainty and volatility, posing risk of operational failures and power outages. Simultaneusly, the growing complexity of the power system challenges in centralized infrastructure due to computational burdens, high data storage costs, and communication bottlenecks. To cope with these challenges, the growing importance of distributed state estimation (DSE) is essential for mitigating violations of operating constraints, including line capacity and bus voltage limits. In recent years, the widespread deployment of phasor measurement units (PMUs) has facilitated the establishment of a comprehensive wide-area monitoring system (WAMS), capable of directly measuring time-synchronized states. However, unlike conventional remote terminal units (RTUs) in supervisory control and data acquisition systems (SCADA), the limited availability of PMUs for state estimation (SE) persists due to high installation costs. Consequently, there is a primary research focus on hybrid DSE, where operating states are estimated using both PMU and SCADA measurements. However, previous hybrid DSE studies necessitate further improvements, particularly in refining bad data processing (BDP) methods, considering practical characteristics such as time skewness resulting from differing sampling rates between SCADA and PMU. In addition, it is also imperative to investigation of DSE convergence under varying operating conditions, attributed to increased uncertainty and variability in power grids. Therefore, this dissertation introduces a novel strategy for decentralized phasor- aided state estimation (DPHASE) to enhance BD processing performance and, consequently, the accuracy of DSE results. Developed using a parallel multi-stage hybrid approach suitable for the recent availability of PMUs, the DPHASE strategy effectively addresses practical challenges associated with differences between sets of buses for SCADA- and PMU-based state estimates in sub-areas. Comparative case studies validate that the proposed DPHASE strategy improves accuracy and robustness under various test network and data measurement conditions. Furthermore, an equality-constrained PMU-based DSE strategy is proposed to enhance numerical stability and computational efficiency. Investigating network conditions inducing the numerical instability of DSE, we define a new local set of measurements and form equality constraints, significantly reducing the condition number and improving the numerical stability of DSE. An accelerated ADMM method is adopted to enhance time efficiency, with simulation results demonstrating the effectiveness of the method under various measurement conditions.