A New Framework for Detection and Identification of Network Parameter Errors

Normalized Lagrange multiplier test has been shown to be very effective for network parameter error identification, but its validation has so far been solely based on extensive simulations. This paper presents a new framework by which: 1) the normalized Lagrange multiplier test is re-formulated from the perspective of hypothesis testing, enabling proper handling of missing bad parameter cases; 2) formal proofs are given for the combined utilization of normalized Lagrange multiplier test and normalized residual test for simultaneous handling of measurement and parameter errors; and 3) the concepts of detectability and identifiability for measurement errors are extended to parameter errors, and a systematic approach for identifying critical parameters and critical k-tuples is provided.