Optimization of the parameters of biocatalytic hydrolysis of vegetable oil using the methods of neural networks and genetic algorithms

Abstract

At present, ensuring the competitiveness of domestic oil and fat products on the domestic and foreign markets should be implemented at the expense of knowledge-intensive production based on the introduction of innovations that provide a qualitatively new level of technological development. In this regard, in the presented work, research was carried out on the optimization of the parameters of biocatalytic hydrolysis of sunflower oil. NovoCor AD L (Novozymes, Denmark) enzyme preparation was used as a biocatalyst. The mathematical apparatus of artificial neural networks was used for the process modelling. The three-layer direct signal transmission network developed as a result of construction, learning and verification was used to calculate the fitness function in the optimization of biocatalytic hydrolysis using genetic algorithms. The software implementation of the mathematical apparatus was performed in the MATLAB environment. The conducted research made it possible to establish the optimal values of the main parameters of the biocatalytic hydrolysis of vegetable oil: the molar ratio of water to oil is 14 : 1, the amount of enzyme is 3.75 % in relation to the mass of oil, the temperature is 60 °C, and the reaction time is 480 minutes. The set optimal parameters were tested in the conditions of experimental and industrial production. According to the test results, the degree of oil hydrolysis was 96 ± 1.8 %, which correlates well with the modelling data.