Prediction of Quality in Reject Refining using Neural Networks

T. Ahvenlampi, M. Tervaskanto, and U. Kortela (Finland)


Neural networks, optimisation, identification, reject refining


The research idea is focused on developing neural network models for the reject refining subprocess. Results are compared with the LS method. The purpose of the models is to predict the average fiber length after the reject refining. The models will be used for the optimisation and the fault diagnosis purposes of the pulp plant. The methodology is applied to the reject refining of the thermo mechanical pulp (TMP).

Important Links:

Go Back