Subspace identification of balanced deterministic bilinear systems subject to white inputs

In this paper, we generalize a class of existing linear sub-space identification techniques to subspace identification algorithms for bilinear, discrete time, time invariant systems. It is shown that in the case that the input is white, most of the properties of linear subspace identification can be extended to equivalent properties for bilinear systems. Further it is shown that the model obtained will be nearly balanced. An example of the identification of a distillation column is given.