Modelling a non-linear pH process via the use of B-splines neural network

This paper presents a new modelling approach for a pH-process in the wet end approaching systems in papermaking, which play a very important role in the paper industry as the quality of finished paper depends on the different types of added chemicals whose reaction are very sensitive to pH values. pH control can be characterized by its severe non-linearity as reflected in the titration curve. By taking the strong acid equivalent as the state variable in the reduced model, a bilinear model of the system is established, which is connected by the severe non-linearity. The estimation of the equivalent titration curve is performed via a B-spline neural network and algorithms for parameter identification are developed.