University of Tasmania
Browse
Brinkmann_whole_thesis_ex_pub_mat.pdf (1.12 MB)

State estimation of distribution networks

Download (1.12 MB)
thesis
posted on 2023-05-27, 09:15 authored by Bernd BrinkmannBernd Brinkmann
Power system state estimation has been introduced over four decades ago and since then has become an integral part of the day to day network operation. State estimators are implemented in almost every control center around the world and are used to continuously monitor the power system in a near real-time fashion. More specifically, the state estimation algorithm is used to obtain an estimate of the network state while acting as a filter for measurement errors by utilizing the redundant nature of the available measurement set. Over time, a large number of redundant measurement devices were installed throughout transmission networks, which makes it possible to estimate the state of a transmission system with a high degree of accuracy. However, this is not the case in distribution networks where only a very small number of real-time measurements are available. In absence of real-time measurements, pseudo-measurements are normally used in order to enable an application of the state estimation method. Pseudo-measurements are forecasted values for loads and/or generation connected at a specific point in the network which generally have large margins of error associated with them. This means that if a large number of pseudo-measurements are used to estimate a network state, the resulting state may be significantly different from the actual network state. This is the reason why state estimation has to this day not been widely implemented in distribution networks. In recent years, however, the growing amount of renewable generation connected at the distribution level has resulted in an increased risk that network constraints are violated. In order to perform a security assessment or take required control actions, the network operator must obtain reliable information about the state of the distribution network. This has given rise to the need of implementing state estimation in distribution networks in order to obtain the required real-time information about the networks state. Before the state estimation can be performed, it has to be determined if a unique estimate of the network state can be obtained from the available set of measurements. This is done by the observability analysis. However, traditional methods only determine if a state can be calculated and not if this result can provide practical information to the distribution network operator. Hence, a new probabilistic approach to observability is developed in order to overcome this limitation. The developed method assesses the network observability depending on the accuracy of the estimated network state and the proximity of the estimated parameters to their constraints under worst case consideration. In case the uncertainty in an estimated state is too large to be practical, additional measurement devices have to be placed in order to improve the accuracy. However, due to economic constraints, only a small number of buses can be equipped with real-time measurements. This makes optimal meter placement an important tool for the implementation of state estimation in distribution networks. As part of this thesis a new meter placement method is described which can potentially reduce the number of required measurement devices compared to conventional meter placement methods while providing a practical level of state estimation accuracy. The goal of this research is to develop new methods that support the application of state estimation in distribution networks in presence of large uncertainties in the state estimation inputs due to the lack of real-time measurements. For this purpose the focus of this thesis is on the accuracy of the state estimation. In particular the aspects of observability, uncertainty quantification, meter placement and the practical representation of state estimation results have been considered.

History

Usage metrics

    Thesis collection

    Categories

    No categories selected

    Exports

    RefWorks
    BibTeX
    Ref. manager
    Endnote
    DataCite
    NLM
    DC