Structural Health Monitoring using Adaptive LMS Filters

A structure´s level of damage is determined using a non-linear model-based method utilizing a Bouc-Wen hysteretic model. It employs adaptive least mean squares (LMS) filtering theory in real time to identify changes in stiffness due to modeling error damage, as well as permanent displacements, which are critical to determining ongoing safety and use. The structural health monitoring (SHM) method is validated on a 4-story shear structure model undergoing seismic excitation with 10% uniform noise added. The method identifies stiffness changes within 0.5-1.0% inside 0.2-1.0 seconds at different sampling frequencies. Permanent deflections are identified to within 10% of the true value in 1.0 second, converging further over the remainder of the record.