Abstract

The structural monitoring, which is becoming increasingly common, makes it necessary to choose one or more methods of processing and analyzing the obtained data. This choice is key to meeting the objectives of this monitoring. Considering that there are many methods with different capacities and requirements, this choice is a complex problem. This paper presents a methodology for the assessment and selection of methods for the identification of dynamic response properties of buildings using vibration monitoring. In this work, well-known methods based on time domain analysis (subspace-state) and non-parametric methods based on the frequency domain are studied when applied to seismic vibration records for a reinforced concrete building instrumented with a network of tri-axial force-balance accelerometers. For two and a half years, the sensor network has recorded close to 500 sensitive earthquakes. Nonetheless, it only includes those with a magnitude higher than 4.8 (ML or MW) and that occurred between September, 2015 and September, 2016. Using six versions of identification methods, modal response parameters are determined for each of the two main directions of the building, considering different levels of information acquired by the sensor network (floor information). An assessment of the performance of each method is conducted based on the number of identified modal frequencies and damping ratios; their variability (coefficient of variation); and the processing time of each method. As a result and conclusion regarding the use of this methodology, it is not only possible to choose the most convenient method for processing vibration records applied to structural health monitoring (SHM) according to its own objective, but also to progressively improve the design of the sensor network.