Contents lists available at ScienceDirect Electrical Power and Energy Systems journal homepage: www.elsevier.com/locate/ijepes Improving topology error identification through considering parameter and measurement errors Mehdi Kabiri ⁎ , Nima Amjady Electrical Engineering Department, Semnan University, Semnan, Iran ARTICLE INFO Keywords: Energy management system Weighted least squares method Topology and parameter error identification ABSTRACT In this paper, a new approach for simultaneous identification of incorrect branch status and erroneous para- meters is proposed. This approach consists of a three-stage algorithm based on the properties of parameter estimation models. It only requires the results of a conventional state estimator to identify the errors. Different statistical analysis and comprehensive numerical experiments are carried out to illustrate the effectiveness of the proposed algorithm. 1. Introduction 1.1. Background and motivation The correct performance of energy management system (EMS) ap- plications highly depends on the accuracy of the state estimation (SE) function. The SE function uses the redundant analog measurements gathered by the SCADA system as well as the network topology ob- tained from status measurements of the SCADA system. Bad measure- ments, and incorrect network topology and parameters are factors that influence the correctness of the SE results. Error in metering devices and noise in communication equipment may lead to bad measurements and incorrect status of some circuit breakers (CBs). Additionally, in- accurate manufacturing data and out-of-date data bases are common reasons of network parameters' errors. Network topology and para- meters' errors may have significant impact on the convergence and accuracy of the SE. Furthermore, they may exacerbate bad data inter- action issues, complicating bad data identification. 1.2. Aim In this paper, a new approach for detecting and identifying network topology errors is presented. The proposed topology error identification approach has the ability of identifying the incorrect branch status in the presence of bad data and inexact network parameters. It uses the nor- malized Lagrange multipliers of the constraints added for modeling CBs and parameter errors to identify the incorrect status of branches through estimating the parameters of suspicious branches pertaining to topology or parameters' errors. The parameter error identification comes into the problem as a subsidiary procedure to improve the per- formance of topology error identification. Errors in network parameters are either permanent or dynamic. The permanent errors remain in the network database until they are even- tually spotted and corrected. In the long-run, it should be expected that most permanent network parameters are properly identified. On the other hand, dynamic errors pertain to parameters that change con- tinuously. For instance, tap positions of transformers have dynamic nature and can affect the parameter errors of transformer branches. Phase shifting transformers have similar impact on the parameter er- rors. Thus, parameter errors in addition to topology errors should be checked regularly. 1.3. Literature review The methods implemented to detect and identify topology errors are generally based on a classical SE or a generalized SE. In the classical SE, the conventional bus-branch model, generated from the topology pro- cessor, is used to identify the incorrect status of branches. For example, in [1] the topology errors are identified by normalized residual tests. In [2], the state vector is augmented by introducing a binary variable per branch. Then, every binary variable is estimated to determine the connected/disconnected status of the associated branch. The generalized SE, unlike the classical SE, incorporates an explicit model of each CB into the SE formulation [3]. Modeling CBs as zero impedance branches has been presented in [4,5]. In [3], the concepts of pocketing and zooming have been presented for topology error detec- tion where the state estimation and bad data detection are conducted on the network pockets and then an incorrect status is identified when https://doi.org/10.1016/j.ijepes.2017.11.011 Received 20 July 2017; Received in revised form 28 September 2017; Accepted 8 November 2017 ⁎ Corresponding author. E-mail address: kabiri@semnan.ac.ir (M. Kabiri). Electrical Power and Energy Systems 97 (2018) 309–318 Available online 21 November 2017 0142-0615/ © 2017 Elsevier Ltd. All rights reserved. T