Application of genetic algorithm for optimization of vegetable oil hydrogenation process

Finding optimal reaction conditions leading to minimal total trans isomer and maximal cis-oleic acid formation during vegetable oil hydrogenation is very crucial. An Artificial Neural Network was developed and used to predict the amount of total trans isomer and cis-oleic acid during the hydrogenation process. Using a large number of experimental data from a pilot plant reactor, the Neural Network was trained and then validated with a validation subset. Having a reasonably accurate Neural Network model of the hydrogenation process, Genetic Algorithm was then used to search for a combination of process variables resulting in minimal total trans isomer and maximal cis-oleic acid formation during the hydrogenation process. The outputs of Genetic Algorithm (i.e. predicted process variables) were used in actual settings of the hydrogenation process to evaluate the effectiveness of the scheme. The results indicated that the scheme could be effectively used to identify the optimal hydrogenation conditions resulting in minimal trans isomer and maximal cis-oleic acid formation during vegetable oil hydrogenation process.