440 / 2022-09-30 15:24:11

NUMERICAL COMPARISON OF ADAPTIVE FILTERS IN STIFFNESS PARAMETER IDENTIFICATION: EXTENDED KALMAN FILTER (EKF) AND RECURSIVE LEAST SQUARES (RLS)

recursive least squares, extended Kalman filter, damage detection, structural health monitoring, time-varying parameter identification

Structural Health Monitoring

In this work, structural damage detection methods that outperform conventional visual inspection methods will be examined. The approaches in question make use of time-based System Identification methods. In the research of system identification and damage detection using EKF in building structures, the user chooses arbitrary beginning parameter settings for EKF execution or substitutes different values to specifically specify and apply the initial parameter values exhibiting the highest convergence performance. In a similar vein, we will compare the RLS method to the EKF method for damage identification. The main distinction between these two approaches is that while RLS only allows us to identify parameters that remain constant over time, EKF allows us to forecast both time-variant parameters like displacement, velocity, and acceleration as well as time-invariant quantities like stiffness. The effectiveness of these strategies for damage identification was demonstrated and contrasted in an experimental investigation on a shear building.