Neural network as tool for virgin olive oil elaboration process optimization

Abstract A neural network has been designed in order to optimize the virgin olive oil elaboration process. The qualitative parameters of the fruit: fat and moisture, and the technological variables of process: olive paste temperature, olive paste injection flow to the Decanter, addition of micronized talc as coadyuvant, olive paste dilution degree at the input of the Decanter and oil off-carrier point in the Decanter were used as input. This optimization has been based on the optimal operation of the centrifugal separator, using the data of the loss of fat and moisture in the by-product olive pomace and the loss oil moisture at the exit of the Decanter, as indicators of its operation. The obtained network has been able to predict the fat content on dried matter of the olive pomace and the oil moisture with errors of RMSEP = 0.75% and 0.04%, and a lineal correlation of r  = 0.949 and 0.981, respectively.