Journal of Materials Processing Technology 161 (2005) 121–127 Identification of analog and topological measurement errors in generalized state estimation George N. Korres a,∗ , Peter J. Katsikas a , George E. Chatzarakis b a School of Electrical and Computer Engineering, National Technical University, Athens, Greece b School of Pedagogical and Technological Education, Athens, Greece Abstract This paper presents a method for analog and topological measurement error identification in generalized state estimation. Errors of analog measurements depend on measurement equipment accuracy. Errors of topological measurements (switching device statuses) result in erroneous network models. In generalized state estimation, analog and topological data are processed as a single interacting information set. The conventional measurement model is extended by incorporating the breaker statuses (open or closed) and the active and reactive power flows as continuous state variables. Depending on breaker statuses, additional pseudomeasurements are introduced. The method treats the analog and status measurements as regular measurements. Hypothesis testing for detection and identification of bad analog and topological measurements is used (J (ˆ x) and ˆ r N , test respectively). Test results for the RTS substation system are reported. © 2004 Elsevier B.V. All rights reserved. Keywords: Generalized state estimation; Analog and topological measurement errors; Circuit breaker status; Bad data analysis 1. Introduction In the conventional state estimator (SE), the topology pro- cessor determines the bus/branch network model by using switching device statuses, and the analog data is processed next by the state estimator. The generalized state estimation treats the analog and topological data as a single interacting information set. Generalized state estimation models substa- tions in detail (bus-section/switching device level), making it possible to detect and identify topology errors, which would appear as interacting bad data. Methods for topology error identification can be found in [1–9]. In [10–15] detailed physical-level modeling of substa- tions is performed by incorporating the circuit breakers as zero impedance branches and their power flows as variables. The well known J (ˆ x)-test is extended to validate hypothe- sis about the status of switching devices [16]. Method [17] overcomes the drawbacks of [16] by introducing the breaker statuses as additional state variables. ∗ Corresponding author. Tel.: +30 210 7723621; fax: +30 210 7723659. E-mail address: gkorres@softlab.ece.ntua.gr (G.N. Korres). This paper presents a unified approach for identification of errors in topological and analog measurements. The con- ventional state estimator is extended by incorporating the active and reactive power flows and the statuses of the circuit breakers as continuous state variables. Hypothesis testing is used to detect and identify bad analog and topological mea- surements. The proposed method is illustrated with the RTS substation system. 2. The conventional state estimation model In conventional state estimation the breaker statuses are processed by the topology processor. By merging bus sections joined by closed breakers into “buses”, the bus/branch model is built. For a N-bus system, the state vector x =[δ T V T ] T , of dimension n =2N - 1, consists of the N - 1 bus voltage angles δ i , i = 2, 3, ..., N, with respect to a reference bus, and the N bus voltage magnitudes V i , i = 1, 2, ..., N. The measurement model is: z = h(x) + v (1a) E(v) = 0, E(vv T ) = R (1b) 0924-0136/$ – see front matter © 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.jmatprotec.2004.07.012