On-line adaptive observer for buildings based on wave propagation approach

A novel on-line system identification method for shear beam building models, based on a wave propagation approach, is developed as an alternative solution to modal analysis methods for the health assessment of multi-story buildings. A discrete shear beam model is introduced that is used to design an adaptive observer, allowing for the estimation of displacements and velocities, as well as the unknown shear wave velocities and damping coefficients in real-time. The adaptive observer design is based only on acceleration measurements, does not need a coordinate transformation, and uses the normalized recursive least squares method with forgetting factor and a parameter projection scheme to achieve stronger convergence. Moreover, the proposed identification scheme employs a novel parameterization based on linear integral filters, which eliminates constant disturbances and attenuates measurement noise. The algorithm efficiency is demonstrated through experimental results on a reduced scale five-story building.