EFFICIENT SYSTEM IDENTIFICATION ALGORITHM USING MONTE CARLO FILTER AND ITS APPLICATION

Abstract

In the Monte Carlo Filter (MCF), the probability density function of the state vector is approximated by many particles. In the classical MCF, however, as the degree of freedom of structural mode lincreases, the number of particles to be generated is increased in exponential order. This results in extreme increase of computaion time. In this paper, we develop the Relaxation MCF, in which we improve the filtering process of the classical MCF. Moreover, we develop the GA-RMCF, which combines the Genetic Algorithm (GA) and the RMCF. We apply the GA-RMCF to identifying the dynamic parameters of a five-story model building using observed data obtained through the shaking table tests.