Modelling pressure drop in water treatment

The nonlinear capabilities of a generally structured neural network and the extrapolation capabilities of a more rigid model used in predicting loss of head pressure profiles in a water treatment plants are compared. The input variables used were incoming water flowrate, ph of water entering the filters and solids build up on the filters. On-line plant sensors were used to obtained data at 15-min intervals. Results show the serial structure to be more robust allowing extrapolation outside the range of the training data set as compared to the parallel arrangement yielding a 20% reduction in RMS validation error.