Identification of transfer functions and statistical inverse problems

From viewpoint of statistical inverse problems we propose a new approach to identification of transfer functions between output variables in a power reactor with feedback loops. Using scalar times transformations of time series data we could find minimum phase innovation models via the singular value decomposition of Hankel matrices, and transfer functions between output variables were obtained from the identified innovation models. An example of Ringhals-1 data has been presented for identification of transfer functions. (C) 2003 Elsevier Science Ltd. All rights reserved.