The Lattice Boltzmann Method has developed for solving the Boltzmann equation in Cartesian domains containing immersed boundaries of arbitrary geometrical complexity moving with prescribed kinematics. When a immersed boundaries are sweeping the fixed fluid node, refilling the node information in a vicinity of fluid nodes is one of the important issues in Lattice Boltzmann Method. In this study, we propose a simple refill algorithm for the particle distribution function based on a proper velocity, density and strain rate to enhance accuracy and stability of the method. The refill scheme based on a asymptotic analysis of LBGK model has improved accuracy than interpolation schemes. The proposed scheme in this study is validated by the simulations of an impulsively started rotating circular cylinder to investigate adaptability for fluid-structure interaction (FSI) problem. This refill scheme has improved stability and accuracy especially at high Reynolds number region.