Naslov
Demulsification analysis by neural networks

Autori
Sinčić, Dinko ; Kurtanjek, Želimir

Izvornik
Nafta (0027-755X) 51 (2000), 3; 303-308

Vrsta, podvrsta i kategorija rada
Radovi u časopisima, članak, znanstveni

Ključne riječi
demulsification; oil; naphtha; neural networks

Sažetak
The laboratory demulsification data have been analyzed by the neural net-work approach. The goal was to predict the demulsification process with the demulsifier component combination not tested experimentally in order to save time and money by speeding up demulsfier formulation. The paper shows how the analysis is performed and presents the promising result: when 6 randomly selected experimental points were left out of the training set, standard deviation of the prediction was 6, 5% while leaving out complete training set resulted in standard deviation of the prediction of 16, 5%.

Izvorni jezik
Engleski

POVEZANOST RADA