MODAL ESTIMATION OF LUMPED PARAMETER SYSTEMS USING VECTOR DATA DEPENDENT SYSTEM MODELS

摘要

This paper applies the theoretical framework of the vector Data Dependent System (DDS) methodology to the response data from a lumped parameter system. This methodology has been tested both experimentally and through simulation using ACSL. The vector DDS model determines the elements of the mass, stiffness and damping matrices within five decimal places for the simulated system, and within experimental error for the real system. The DDS methodology uses discrete time difference equations to formulate an autoregressive moving average vector (ARMAV) model from digitized acceleration data. The ARMAV models are formulated in a state space format by simultaneously measuring the acceleration of each degree of freedom which is of interest. It is shown that the modal parameters (natural frequencies and mode shapes) can be evaluated by analyzing the system's free vibration. The system parameters (mass, damping, and stiffness matrices) can be evaluated by analyzing the response to a single known sinusoidal forcing function.