Modal identification based on ARMA model fitting. 1st report. Scalar approach using recursive method.

Abstract

This paper presents a dynamic identification method which is based on ARMA (Autoregressive/ Moving-Average) model fitting. The method has the advantage that only observed random time history is required for the dynamic characteristics identification without calculating transfer function. The method has the following two procedures. First, in order to estimate ARMA parameters from observed time history, an order-update recursive (LWR) algorithm is introduced. Second, modal parameters are calculated by use of ARMA parameters. As a fundamental example, a five degrees of freedom system is identified with the aid of numerical simulation. From the results, it is shown that estimated modal parameters agree well with the exact value, even without as much data as was required in the former method.