Advanced signal processing methodology of vibration response data toward Structural Health Monitoring purposes

Abstract

Abstract This paper outlines a comprehensive and consistent methodology for signal processing analysis of vibration response data, applicable for final structural monitoring and identification purposes. The methodology combines classical and advanced techniques, including, in its pre-processing phase, the adoption of a Time Domain Compression (TDC) technique and the application of an AutoRegressive Moving Average (ARMA) modeling approach. The TDC technique removes lower-quality subsamples from the full data set, resulting in a higher-quality modified signal that may display a weakly stationary character. The ARMA modeling approach enhances the understanding of the response signals by modeling unknown source inputs; as a peculiarity, the inherent polynomial function applied to a white noise source in the model is interpreted as a filtering term that transforms the source into a non-white noise configuration, enabling the effective deciphering of the structure transfer function features. The research is part of a more comprehensive case study concerning the structural evaluation of a historical reinforced concrete arched bridge over the Adda river in Lombardy, Italy. The focus of this paper is specifically on the application of the TDC and ARMA techniques to the signal response data collected from the bridge under operational conditions.