State estimator based on the least median method

Author(s)

Marcelo Nanni

• Document type: Master's Dissertation
• Press: São Carlos.
• Institution: Universidade de São Paulo (USP). Escola de Engenharia de São Carlos (EESC/SBD)
• Defense date: 2009-04-01

Examining board members

João Bosco Augusto London Júnior; Alexandre Cláudio Botazzo Delbem; José Roberto Sanches Mantovani

• Advisor: João Bosco Augusto London Júnior

Abstract

In the last decades several state estimators were developed and applied to power systems. Among these estimators, the one based on the least median method is of interest to us. This because the weighted least median of squares (WLMS) estimator is capable of filtering gross errors corrupting redundant measurements called leverage points. Leverage points are highly influential measurements that attract the state estimations solution towards them. However, some of the WLMS estimator tasks require excessive computing time, making that estimator impracticable to large-scale power systems in real-time environment. The WLMS estimator tasks requiring excessive computing time are: (i) selection, among all available measurements, of several samples with N non-redundant measurements for which the system is observable, where N is the number of system states to be estimated; (ii) determination of the minimum redundancy of the available measurements set; and (iii) the solution of several load flows (one for each selected samples of N measurements). This work proposes a robust state estimator based on the least median method applicable to large-scale power systems in real-time environment. In order to do this, new methodologies were proposed to perform the tasks mentioned above. The proposed methodology to perform tasks (i) and (ii) is based on the analysis of the H\'delta\' matrix structure (this analysis enables both observability and redundancy analysis in a straightforward manner). To perform task (ii), it was developed a load flow methodology based on a forward/backward sweep power flow method. Finally, in order to increase the computational efficiency of the proposed estimator in real-time environment, the tasks that do not depend on real-time information will be conducted by an off-line process. As the main results of this work we could enumerate: (i) a robust state estimator; and (ii) an efficient methodology to determine both the minimum redundancy of the available measurement set and the observable samples with N non-redundant measurements. (AU)