Computationally Efficient State Estimation for Power Systems with Conventional and Synchrophasor Measurements

Abstract

State Estimation (SE) is a mathematical algorithm used to process a redundant set of measurement data, thereby filtering out measurement errors and providing an estimate of the most likely state of a power system. Due to their great importance for reliable system monitoring, SE algorithms have been researched and refined for decades. As a result, SE formulations that are based on conventional measurements provided by Remote Terminal Units (RTU) are well established. However, the increased penetration of Phasor Measurement Units (PMU) into transmission systems in recent years has introduced a new potential for improvements in the SE area, but also new numerical challenges. Therefore, novel computationally efficient hybrid SE methods that are able to utilize both RTU and PMU measurements are required. The main scope of this thesis is the provision of novel algorithms for static state estimation of transmission systems that can treat both conventional RTU measurements and the synchrophasor data provided by PMUs simultaneously, i.e. within a single-stage estimation problem. The ultimate goal is the derivation of algorithms with low computational complexity, high accuracy, and robustness against gross measurement errors. Furthermore, the thesis also investigates new linear approaches for parameter estimation, as well as efficient methods that can take into account all available synchrophasor measurements, which are updated in practice much more frequently compared to the RTU data. The first part of the thesis is focused on leveraging the recently proposed Equivalent Circuit Formulation (ECF) of the power flow problem to derive new SE methods based on the concepts of circuit theory. More precisely, two ECF-based methods that can treat RTU and PMU data simultaneously are derived. The first one, although being nonlinear, significantly reduces the level of inherent nonlinearity compared to conventional power flow-based estimators, while the second one is a fully linear algorithm, and is essentially an improved version of the first proposed estimator. While both methods map the state estimation problem to the equivalent circuit framework by introducing appropriate circuit models for different types of measurement sets, they differ from each other primarily in the approaches used for the circuit modelling of RTU measurements, as well as in the methods utilized to address the effect of measurement uncertainties. The goal of the second part of the thesis is the derivation of novel SE methods within conventional linear Weighted Least Squares (WLS) and Least Absolute Value (LAV) frameworks. The aim is again to ensure the simultaneous treatment of RTU and PMU data, as well as to derive linear measurement functions for all types of measurements. The proposed modelling approach for the entire network, as well as for all measurements, is formulated with respect to voltages and currents in rectangular coordinates and is fully linear, which renders the computational complexity of the proposed estimators low. The proposed WLS-based SE algorithm utilizes the Largest Normalized Residual (LNR) test to identify bad measurements in the post-processing stage. On the other hand, the introduced LAV-based estimator is inherently robust and is formulated and solved as a single Linear Program (LP). Additionally, an alternative inherently robust estimator is also discussed. A novel method for computationally efficient parameter estimation is developed as well in the second part of the thesis, based on the previously mentioned linear WLS-based SE algorithm, thus allowing for the simultaneous estimation of states and network parameters. The entire algorithm is linear, and it can be used to estimate parameters of transmission lines, transformers, and shunts. Finally, three methodologies based on a static state estimation framework are proposed to address the discrepancy in refresh rates of PMU and RTU data. Hence, these algorithms ensure that all available PMU measurements are leveraged, which ultimately yields an improved tracking of the system's dynamic behavior.