Premium
An UKF‐based nonlinear system identification method using interpolation models and backward integration
Author(s) -
Li Luyu,
Qin Han
Publication year - 2018
Publication title -
structural control and health monitoring
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.587
H-Index - 62
eISSN - 1545-2263
pISSN - 1545-2255
DOI - 10.1002/stc.2129
Subject(s) - nonlinear system , interpolation (computer graphics) , kalman filter , control theory (sociology) , identification (biology) , system identification , nonlinear system identification , linear interpolation , computer science , algorithm , mathematics , artificial intelligence , data modeling , pattern recognition (psychology) , quantum mechanics , motion (physics) , physics , botany , control (management) , biology , database
Summary In this paper, a novel identification method for nonlinear systems is proposed. This method utilizes linear interpolation models to describe the nonlinear forces of the physical models, and the unscented Kalman filter (UKF) method is adopted for the task of nonlinear identification. With the help of a linear interpolation algorithm, the proposed method requires little prior knowledge of the form of the nonlinear stiffness. Therefore, this method takes advantage of both the independence of the linear interpolation points and the inherent mathematical properties of the UKF. The UKF method is also modified to better fit the needs of parameter identification. To further emphasize parameter identification, backward integration and observations of the previous states are used. Two numerical simulations of the nonlinear elastic stiffness and Bouc–Wen hysteresis are conducted to show the flexibility and efficiency of this method. In these 2 examples, the observation signals are generated by analytic models, and the identifications are conducted with a linear interpolation model.