Advanced signal processing techniques for robust state estimation applications in smart grids
Since their inception, more than one century ago, electrical grids have played the role of a critical infrastructure. During the majority of this time, power systems have not faced radical changes. In contrast, over the last two decades this paradigm has rapidly changed. On the one hand, the environ...
| Autor: | |
|---|---|
| Tipo de recurso: | tesis doctoral |
| Estado: | Versión publicada |
| Fecha de publicación: | 2020 |
| País: | España |
| Institución: | CBUC, CESCA |
| Repositorio: | TDR. Tesis Doctorales en Red |
| OAI Identifier: | oai:www.tdx.cat:10803/670010 |
| Acceso en línea: | http://hdl.handle.net/10803/670010 https://dx.doi.org/10.5821/dissertation-2117-332753 |
| Access Level: | acceso abierto |
| Palabra clave: | Àrees temàtiques de la UPC::Enginyeria de la telecomunicació 621.3 |
| Sumario: | Since their inception, more than one century ago, electrical grids have played the role of a critical infrastructure. During the majority of this time, power systems have not faced radical changes. In contrast, over the last two decades this paradigm has rapidly changed. On the one hand, the environmental need for de-carbonization has stimulated the introduction of (i) green energy through renewable energy sources (RES); and (ii) Distributed Energy Resources (DER). On the other, the de-regulation of energy market has raised the necessity for substantial cooperation between the energy utilities. All the above implies that, to start with, power grids must be able to support bi-directional power flows. And, further, that variations in power generation and consumption must be timely and accurately monitored. To that aim, engineers and researchers can exploit recent innovations in measurement technology; advanced signal processing algorithms and optimization tools; and a widespread use of wired and wireless communication technologies. This, clearly, brings the notion of Smart Grids (SG) into play. The accomplishment of this modern paradigm requires the re-design of a number of classical management and control strategies running in the operation centers of traditional grids. Specifically, the main objective of this PhD dissertation is the re-formulation of a key functionality for the efficient monitoring, control and optimization of electrical networks: State Estimation (SE). Our research has been divided in two parts. In the first, the study is focused on the Transmission Grids (TG). The second, is dedicated to the (medium voltage) Distribution Grid (DG). With respect to TGs, we propose a hybrid SE scheme exploiting both PMU and legacy measurements. The problem suffers from an inherent non-convexity and, thus, we adopt a successive convex approximation framework (SCA-SE) to iteratively solve it. Our goal is to attain increased accuracy and faster convergence rate. Going one step beyond, we pose the SCA-SE problem in a decentralized setting. For the solution, we resort to the so-called Alternating Direction Method of Multipliers (ADMM). Finally, we take into consideration the presence of bad data in the measurement sets. In this case, we reformulate the problem in a Least Absolute Shrinkage and Selection Operator (LASSO) optimization framework and, we provide joint state estimation and bad data detection. In the second part of this dissertation, we address the problem of SE for the distribution grid. Our aim is to propose an algorithm capable of tracking the rapid variations over the voltage profile. To do so, we leverage on the recently introduced micro-PMUs (µPMUs) for distribution grids. Specifically, we present a regularized SE scheme operating at two different time-scales: (i) a robust state estimator that operates at the main time instants; and (ii) a regularized SE scheme for a number of intermediate time instants. For the former, we formulate the estimator as a regularized version of the Normal-Equations based SE solution (R-NESE). As for the latter, we present a Decomposed Weighted Total Variation State Estimation (D-WTVSE) scheme. In order to solve the D-WTVSE problem, we resort to the ADMM. Besides, we study the problem of µPMU placement (µPP). The problem is posed as a mixed integer semidefinite programming (MISDP) model and, thus, it can be efficiently solved. |
|---|