State space theory and NR-LMS algorithm based method for structural dynamics parameter identification

A parameter identification method for structural dynamics system based on state space (SS) theory and normalized robust least mean square (NR-LMS) algorithm was proposed. By using this method, the identified dynamic system's input and output data were used to build its Hankel-Toeplitz model based on the state space theory. Iterative NR-LMS algorithm was applied to achieve parameters' estimates and Hankel matrix for this model. Singular value decomposition (SVD) method to Hankel matrix was employed for quantifying the order of this dynamic system. Modal parameters and the state space model's parameters also could be achieved from the Hankel matrix by certain calculation. A simulation of 3-DOF (degree of freedom) spring-mass system was employed to validate this method and experiment of identifying cantilever's parameters was studied. The results demonstrate this method can identify structural parameters accurately and quickly.