Generality of nonparametric nonlinearity identification approach with improved extended Kalman filter using different polynomial models

Severe dynamic excitation frequently gives rise to local nonlinear structural behavior in structures. Therefore, due to the diversity of structural nonlinearities, a novel method is proposed for identifying the nonlinear restoring force (NRF) and dynamic loading in a nonparametric manner for multi -degree -of -freedom (MDOF) structures, which utilizes an improved extended Kalman filter with unknown input (IEKF-UI) using partial degree -of -freedoms (DOFs) acceleration measurements. The first Chebyshev polynomial model (FCPM) and Legendre polynomial model (LPM) are used as the nonparametric models for NRF, respectively. To investigate the generality of the proposed method, numerical simulations and experimental test with MDOF frame structures fitted with different parametric magnetorheological (MR) dampers mimicking diverse nonlinearities are conducted. Results considering different noise levels imply that the nonparametric identification method using different polynomial models is general for the identification of nonlinear structures involving different parametric MR dampers, where identified energy dissipation curves of MR damper are also compared.