Estimation of Impulse Responses: a novel Method and its Use in Experimental Modal Analysis

A novel method for the estimation of Impulse Response matrices from measured input-output data, called IRE (Impulse Response Estimation) is presented. It is shown that the method is strongly consistent under colored output noise. Its performance is compared to two other methods achieving the same goal. Then, the IRE algorithm is combined with classical impulse response driven algorithms for experimental modal analysis (pLSCE, ERA and ITD) and with a new algorithm, IREAR (Impulse Response Estimation And Realization). A simulation example shows that IREAR outperforms the three classical algorithms. Nomenclature • Moore-Penrose pseudo-inverse • deterministic, noiseless part • stochastic, noisy part •12 columns c1 to c2 •r1:r2 rows r1 to r2 μ̂(•) sample mean σ̂(•) sample standard deviation σ(•) standard deviation E[•] expectation operator