Observability and bad data analysis using augmented blocked matrices (power system analysis computing)

Abstract

Observability and bad data analysis in the context of the recently developed blocked sparse approach are discussed. The factorization-based observability analysis is extended to the new method. Important statistical derivations involving the blocked augmented matrices are presented. The computational aspects of performing bad data analysis as well as guidelines for computing residual variances and sensitivity matrices, are discussed. An efficient implementation of a bad data analysis algorithm based on normalized residuals and nonlinear residual computations is described.

Parallel domain decomposition based distributed state estimation for large-scale power systems

“…One of the popular method for BDI which is used in this paper is based on normalized residual test [18]: In this work the measurements having the largest normalized residual and larger than 3 were considered as bad data, with a 99. 7% confidence level.…”

Application of the undetectability index to design reliable metering systems for bad data processing

“…In order to be robust to gross errors, the WLS Estimator is endowed with the largest normalized residual technique (or largest normalized residual test) [1], [3], [4], [5], and the gross errors processing (or bad data processing) is performed only after the estimation process by analyzing the measurement residuals. The analysis is essentially based on the properties of these residuals, including their expected probability distribution [2], [6].…”

Observability analysis for state estimation using Hachtel’s augmented matrix method

“…The theory of numerical observability was extended to the blocked sparse approach in Ref. [16]. The numerical observability analysis algorithms extended to Hachtel's augmented matrix or blocked sparse approach, are mainly based on the algorithms developed in Refs.…”