Prediction models for the corrosion phenomena in Pulp & Paper plant ☆

Abstract In this paper two predictive methods have been developed in order to cope with the monitoring of the corrosion phenomena in a Pulp & Paper plant. The data of the process has been collected in Wisaforest pulp mill, located in Pietarsari (Finland), in three periods characterised by three different operating conditions. Two different strategies have been adopted, firstly using a classical multi-layer perceptron and then using the neuro-fuzzy expert systems. Moreover, both nonlinear moving average models and NARMAX ones have been designed. The results obtained allow to predict the corrosion potential accurately.