地震応答のARモデルと融合粒子フィルタを用いた構造同定

抄録

In the field of structural identification, Kalman filter and Particle filter have been one of the most widely used tools. These are called parameter identification and data assimilation methods and are methods for updating sim-ulation models one by one while introducing measurement data to the simulation model. In the updating process, the dynamic characteristics of the observed structure such as the stiffness and damping properties are estimated. However, in order to ensure the identification accuracy, it is extremely important to set the process noise. This study attempts to develop a SHM system that can accurately evaluate the integrity for the structure that has not been destroyed due to an earthquake. The first attempt is made to detect the damage occurrence from the re-sponse data by using AR model. In evaluation of response, AR coefficients are used to detect a sudden change of structural condition caused by seismic load. The second attempt is made to identify the structural parameter for the purpose of damage quantification by using merging particle filter. Finally, the performance of the proposed system is discussed through numerical simulations and a shaking table test.