1993 年 59 巻 567 号 p. 2569-2574
This paper is concerned with elastodynamic inverse analysis in which unknown defects, cracks and/or boundary conditions should be identified from displacement responses on the boundary of an elastic body. In the previous paper, the authors proposed the inverse analysis method for identification of unknown parameters using the extended Kalman filter algorithm and the boundary element analysis for steady-state elastodynamics. In this study, computational aspects of the inverse analysis method are investigated with respect to the influence of errors in the measured data and to arrangement of the measuring points. Furthermore, computational efficiency and accuracy are compared between the inverse analyses using the extended Kalman filter algorithm and the optimization method. Numerical simulation is carried out for some example problems to identify an unknown defect and unknown boundary values as a parameter identification problem.