Efficient computation of covariance matrix entries

It was shown earlier that network parameter errors could be identified using their normalized Lagrange multipliers. However, normalization requires computation of their covariance matrix which is computationally very time-consuming, thus limiting its practical use for large-scale systems. This paper presents an effective algorithm to address this shortcoming. The proposed algorithm is described and simulation results are provided to illustrate the dramatic reduction in computational costs for large power systems.