A supervised learning computer-based algorithm to derive the amplitude of oscillations of structures using noisy GPS and Robotic Theodolites (RTS) records

摘要

A computer-based algorithm computing the mean amplitude of small-scale, high frequency oscillations of civil engineering structures using GPS and Robotic Theodolites (RTS) is presented. This algorithm is based on a peak-picking filter derived from supervised learning through independent experiments. This filter is a function of the oscillation frequency, computed even from noisy, apparent displacement records. The algorithm minimizes the noise of high-frequency GPS and RTS recordings and calculates the mean amplitude of the oscillations with millimeter accuracy. The application of this method to the Structural Health Monitoring is demonstrated in the case of measurement of displacements of two short-span bridges.