Power system observability and related functions Derivation of appropriate strategies and algorithms - - - Th. Van Cutsem Department of Electrical Engineering, University of zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA Liege, 6 4000 Likge, Belgium zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA Observability analysis deals with the adequacy of the avail- able measurement configurah’on for state estimation purposes. This paper considers methods to explore observ- ability and to face unobservable situations. M ore explicitly , first, the problem of unobservable networks is examined; strategies currently used are discussed and a more general method is proposed. to build a complete and reliable database. Second, topological concepts derived from observability are used to enhance the reliability of the bad data analysis under sningent identification conditions. Third, a concrete algorithm for observability analysis is proposed. Kqvwords: online control strategies, state estimation, data analysis I. Introduction The basic role of state estimation is to generate a complete, coherent and reliable database from information provided by signalling and measurements’. Observability analysis is aimed at determining whether the available measurements are sufficient in number and loca- tion to allow the state vector of the whole system to be estimated. If so, the network is said to be observable; otherwise it is unobservable. Important contributions in this field have been made by Clements, Krumpholz and Davis. These authors have laid sound theoretical foundations of the observability analysis by establishing and proving a necessary and sufficient con- dition for observability2. and by investigating several other related questions3-“. The need of an observability program arises first during the implementation of any state estimator: it is necessary to ensure that the power system is observable under various -- _ Received: 10 April 1984 normal operating conditions. At this stage, systematic sources of unobservability should be eliminated by metering reinforcement. This, however, does not guarantee that the power system will remain observable during online opera- tion: unobservability may arise because of telemetry failures or line outages. Another source of unobservability comes from the current trend in security analysis to repre- sent part of the external network by an unreduced load flow mode16. The external system model is then attached to the internal one and both systems are processed together by the state estimator. As complete data concerning external networks is generally not available in real time, these external networks will be systematic sources of un- observability. Sections II and III of this paper review briefly the funda- mental aspects of the observability theory of Clements et al. that present direct applications. Section IV reviews and develops notions and properties of critical measure- ments, which will be directly applied in the sequel. In particular, the notion of spread area of critical measure- ments is proposed. Section V addresses the treatment of unobservable networks. When a system is found to be unobservable. additional information is required to build a (as reliable as possible) database covering the whole net- work. Possible strategies are discussed and a general and flexible method is proposed. In Section VI, the notion of spread area of critical measurements is applied further to bad data analysis. This satellite function is also of major importance for a reliable state estimation, its two main objectives being to identify the erroneous measurements and to clear the final database from the errors induced by them. The paper proposes a post elimination procedure which improves the reliability of the bad data analysis in the presence of weak measurement configurations (or configurations weakened by undue measurement elimina- tions). I I. Problem statement I I.1 Fundamentals of static state estimation The weighted least square (WLS) state estimation of an Vol 7 No 3 July 1985 0142-0615/85/030175-13 803.000 1985 Butterworth & Co (Publishers) Ltd 175