Modeling and Optimization of Anethole Ultrasound-Assisted Extraction from Fennel Seeds using Artificial Neural Network

Abstract

Extraction of essential oils from medicinal plants has received researcher’s attention as it has a wide variety of applications in different industries. In this study, ultrasonic method has been used to facilitate the extraction of active ingredient anethole from fennel seeds. Effect of different parameters like extraction time (20, 40, and 60 min), power (80, 240, and 400 Watts) and solid particle size (0.3, 1, and 1.7 mm) on the anethole extraction yield have been studied. The box-Behnken design method has been used for the design of experiments to reduce the number of experiments. A second-degree polynomial was proposed to predict the relationship between independent variables and the dependent variable. An artificial neural network was trained with experimental data to provide another model for the system. Optimal results achieved when using the Levenberg-Marquardt back-propagation algorithm, Logsig, and Tansig transfer functions for hidden and output layers and the number of 10 neurons in the hidden layer. Coefficient of determination, sum of squared errors, root of mean square error, and absolute average deviation were found to be 0.9933, 0.0199, 0.0059, and 2.1944 for the ANN model and 0.9851, 0.0425, 0.0059 and 2.1944 for the design of experiment (DOE) model, respectively.

Keywords

• artificial neural network
• box-behnken design
• essential oils
• surface methodology
• ultrasound