A sensor-software based on artificial neural network for the optimization of olive oil elaboration process

An artificial neural network (ANN) was built for real-time prediction of the moisture and fat content in olive pomace using two-phase olive oil processing. Technological variables were used as input, including olive paste flow, olive paste temperature, coadyuvants addition, water dilution level, position of the exit of the oil in the 'horizontal centrifuge decanter', and the Wavelet pretreated near infrared spectra from the on-line scanned oils at the exit of the decanter. The results obtained indicate a very good predictive capacity of the three-layer ANN model with values of r = 0.961 and RMSEP = 0.32% for fat content and r= 0.970 and RMSEP= 1.01% for moisture. (c) 2007 Elsevier B.V. All rights reserved.