EFFICIENT ITERATIVE ARMA APPROXIMATION OF MULTIVARIATE RANDOM-PROCESSES FOR STRUCTURAL DYNAMICS APPLICATIONS

An efficient Auto-Regressive Moving-Average (ARMA) approximation metho d is presented for simulating stationary random processes with specifi ed (target) power spectra in conjunction with structural dynamics appl ications. It involves an iterative algorithm developed for minimizing a physically motivated 'energy' measure, in the frequency domain, of t he ARMA approximation of an AR representation of the target spectrum. The iterative algorithm can be used to adjust, for better spectral mat ching the parameters of an arbitrary ARMA approximation of the random process determined by any other method; this is accomplished without i ncreasing the requisite order of the ARMA approximation. The efficienc y of the proposed method is demonstrated by considering spectra which are commonly used in earthquake engineering and ocean engineering.