Researchers: Aki Vehtari and Jouko Lampinen

In this work modeling of pumping process was studied. Test case was MC-pump of Ahlstrom Pumps, which is used for mass transfer in pulp and paper industry. In case that all of the measurable values of the pumping process have not been instrumented, the model of the pumping process can be used to estimate the state of the pump and the values of missing measurements.

The modeling of the pumping process is difficult, due to multi-modal inverse problem. For inverse problems there exists a well-defined forward problem which is characterized by a single-valued mapping. Often this corresponds to causality in a physical system.

In the case of pumping process, the forward problem can be defined with characteristic curves. In this work the modeling of the characteristic curves and solving the inverse problem with the MLP neural network was studied. The MLP neural network is a flexible method for nonlinear modeling. This work includes a review describing theory, characteristics, training and selection of model complexity of the MLP network.

There are several methods to inverse MLP network models. This study includes a review describing most common methods and their pros and cons.

The modeling of the characteristic curves was tested with three
different data sets using advanced methods for MLP network
training and model complexity selection. Four different methods
were tested for solving the inverse problem.