Jacobian matrix-based observability analysis for state estimation

In this paper, numerical observability analysis is restudied. Algorithms to determine observable islands and to decide a minimal set of pseudo-measurements to make the unobservable system observable are presented. The algorithms make direct use of the measurement Jacobian matrix. Gaussian elimination, which makes the whole process of observability analysis simple and effective, is the only computation required by the algorithms. Numerical examples are used to illustrate the proposed algorithms. Comparison of computation expense on the Texas system among the proposed algorithm and the existing algorithms is performed.