Using Components of the Mahalanobis Squared Norm of the Residuals Vector for Quality Control in Local Engineering Networks

Abstract

ABSTRACT: Local engineering precision networks are currently used to monitor the displacements of large dams, with safety control purposes. These one, two or three-dimensional networks are local, be-cause they are tied to local reference frames, and they are precision networks, because they aim at millimetric, or even submillimetric, accuracy. The most important obstacles to such an accuracy are the instrumental and environmental systematic errors. The atmospheric refraction, which cau-ses curvature of the path of the electromagnetic waves and changes its propagation velocity, is the major source of systematic errors. Effective measurement quality control strategies are of utmost importance to guarantee the quality of the results (displacements). The paper presents a quality control strategy supported by random quadratic forms that result from the decomposition of the Mahalanobis squared norm of the residuals vector. The strategy is compared to the well known da-ta-snooping method with regard to the quality control and parameter estimation performances.