Unified PMU Placement for Observability and Bad Data Detection in State Estimation

This paper presents a unified algorithm for optimal phasor measurement unit (PMU) placement in power system state estimation. Considering the similarity of the cases that the system is unobservable and that the system becomes unobservable if a critical measurement or a critical pair is removed from the measurement set, a unified algorithm of PMU placement for state estimation combining observability analysis and bad data detection into a single iterative procedure is presented. In each iteration of the algorithm, PMU placement is performed to make the system observable, or make critical measurements become non-critical, or make critical pairs become non-critical pairs, respectively. Based on Theorem 3.1 in “A direct numerical method for observability analysis” (Gou and Abur, IEEE Trans. Power Syst., vol. 15, no. 2, pp. 625-630, May 2000), this paper presents an improved optimal PMU placement algorithm wherein each iteration is performed by a significantly smaller integer linear programming (ILP) model whose size is governed by the rank deficiency of the Jacobian matrix resulting from the configuration of conventional measurements. The proposed formulation thus leads to a far more smaller and simpler ILP problem compared to existing methods. Additional results are also presented to identify critical measurement pairs. The proposed approach is illustrated on the IEEE 14- and 118-bus systems.