Predictive ANN models for the optimization of extra virgin olive oil clarification by means of vertical centrifugation

Abstract

Eight artificial neural networks were developed as predictive models for regulated and nutritional parameters of oils obtained from vertical centrifugation. The networks were designed considering the NIR spectra of oily must at its exit from the horizontal decanter centrifuge, and the flows and temperatures of the oil and of the addition water. The results obtained in all the networks designed indicated the good creative capacity of the neural models through their quality indicators (RER and RPD). The correlations between the real data and those predicted by the networks were within the range of 0.76–0.99, the UV absorption networks (K232 and K270) being those which gave lower correlations. For polyphenols, tocopherols and carotene and chlorophyll pigments, the te‐Student external validation test makes a reliable prediction of these parameters. With predictive models, the vertical centrifuge can be modeled and regulated as well as monitored on‐line to control and optimize the clarification phase within the extraction process of extra virgin olive oil (EVOO). Practical applications The predictive ANN can use to regulate and model the vertical centrifuge to obtain an EVOO of a maximum regulated and nutritional quality possible, reducing water and energy consumption. Besides, with on‐line prediction of oil characterization, it will be possible to perform a correct classification prior to its storage in the cellar.

Grey Box Modelling of Decanter Centrifuges by Coupling a Numerical Process Model with a Neural Network

“…Funes et al [8] introduce an ANN predicting quality parameters of olive oil during its processing in a disc stack separator. Process variables, like the feed temperature and the feed flow of oil must and water and the temperature and the flow of oil at the outlet, serve as input parameters for the ANN.…”

Parameter optimization of double‐blade normal milk processing and mixing performance based on RSM and BP‐GA

“…The prediction ability of models can be evaluated by predicting root‐mean‐square error RMSEP, determination coefficient R 2 , and relative percent difference (RPD; Dai, Cheng, Sun, Zhu, & Pu, ; Dong, Zhao, et al, ; Funes, Allouche, Beltran, Aguliera, & Jimenez, ; Wu et al, ). The lower the predicted root‐mean‐square error is, the higher the determination coefficient and relative percent difference would be, and higher prediction precision and generalization of the model would be (Guo et al, ; Magwaza, Naidoo, Laurie, & Shimelis, ).…”

The Oil:Water Ratio in the Vertical Centrifuge Separator and Its Influence in Phenolic Compounds in the Virgin Olive Oil and the Olive Mill Wastewater (Alpechín)

“…Even so, there still remain important points for improvement, some within the remit of the mill master, but others at the level of manufacturers when the sector demands solutions for the correction of easily identifiable quality loss points (oil heating, flow control, etc.). Recent research has sought to improve control by means of different devices monitoring the correct operation of the centrifuge and the quality of the centrifuged oil [31,34], and hence improve competitiveness.…”