Analysis of the Largest Normalized Residual Test Robustness for Measurements Gross Errors Processing in the WLS State Estimator

Abstract

This paper purpose is to implement a computational program to estimate the states (complex nodal voltages) of a power system and showing that the largest normalized residual (LNR) test fails many times. The chosen solution method was the Weighted Least Squares (WLS). Once the states are estimated a gross error analysis is made with the purpose to detect and identify the measurements that may contain gross errors (GEs), which can interfere in the estimated states, leading the process to an erroneous state estimation. If a measure is identified as having error, it is discarded of the measurement set and the whole process is remade until all measures are within an acceptable error threshold. To validate the implemented software there have been done several computer simulations in the IEEE´s systems of 6 and 14 buses, where satisfactory results were obtained. Another purpose is to show that even a widespread method as the LNR test is subjected to serious conceptual flaws, probably due to a lack of mathematical foundation attendance in the methodology. The paper highlights the need for continuous improvement of the employed techniques and a critical view, on the part of the researchers, to see those types of failures.

Keywords

• State Estimation
• systems
• Normalized Residual
• Gross Errors