Direct multivariate subspace time identification

The paper presents a time domain method to identify structural modal parameters by fitting a discrete multivariate space-time model into noise corrupted, input–output measurement data. The subspace identification scheme proposed is an important characteristic of the method, leading to a deterministic and statistically bias free estimation of parameters. The identification model in this scheme, is kept to the minimum description order, in spite of the presence of noise in the system's input and output The method is used to estimate natural frequencies, damping factors and mode shapes in an experimental modal analysis test.