Grey-Box Modelling of Distributed Parameter Processes - Application to Chromatography

A systematic approach to grey-box identification of distributed parameter processes is proposed. Given powerful tools for grey-box modelling of lumped parameter systems, the principle idea is to separate the model reduction error from discrepancies between model structure and measurement data, through estimating the model reduction error. Hypothesis testing is also employed to ensure a reasonable structure of the tentative PDE model as well as order of the discretized model. The reduced model is augmented with a mesh control, which dynamically determines positions of discretization meshpoints in order to reduce the model reduction error.