Identification of civil structural parameters using the extended Kalman filter

Abstract

In the context of civil and industrial structures, structural control and damage detection have recently become an area of great interest. The safety of a structure is always the most important issue for structural engineers, and to achieve this goal, the discipline of Structural Health Monitoring (SHM) was introduced. SHM records real-time information concerning structural conditions and performances. In order to evaluate the health conditions of structures, identifying the structural parameters is needed. Research activities of this area are increasing due to the availability of computation and wireless technologies. The objective of this thesis is to evaluate the tracking ability of the Kalman filter for identifying civil structural parameters based on measured vibration data which usually are earthquake accelerations. For linear elastic structures, the ordinary Kalman filter was used, but for nonlinear elastic structures, we implemented the extended Kalman filter.

(cont.) For simulating damage occurrence in structures, a sudden change of stiffness was introduced, and an adaptive extended Kalman filter was utilized to estimate the time-varying parameters. In this thesis, linear and nonlinear structures with single-degree-of-freedom and multi-degree-of-freedom were simulated. Measurements having different levels of white noise were considered in order to evaluate the effects of noise on parametric estimations. In addition, the impacts of different levels of noise covariance were also discussed. Simulation results from different structural models were presented to demonstrate the effectiveness of the Kalman filter.