In this project, a grey box model concerning the modeling and prediction of the bending stiffness of the boar. Nowadays, the model is able to predict bending stiffness and tensile stiffness to within the laboratory measurement accuracy. The disturbances are modeled with the help of nonlinear Kalman filters, and the the non-whiteness of their residual may serve as an indicator for the eventual need of adaptation.
The model was implemented in the Frövi mill wide information system as a bending stiffness predictor for simulator for the process engineers and operators. It was programmed in Visual Basic and has benn running for more than a year, giving predictions every 12 minutes with prediction errors having zero mean and standard deviation = 0.4nN, to be compared with the laboratory measurements every hour with the measurement standard deviation = 0.4mN. It is also possible to use the program as a steady state simulator, i.e. obtaining the resulting bending stiffness, layer thickness and layer tensile stiffness as the rsult of assumed constant values of all input and control variables.