Near critical carbon dioxide extraction of Anise (Pimpinella Anisum L.) seed: Mathematical and artificial neural network modeling

Artificial neural network model for the kinetics of canola oil extraction for different seed samples and pretreatments

“…In general, a more accurate fit provided by the ANN model with respect to the corresponding DMs is observed. These results are similar to those reported by other authors who modeled the mass transfer during the dehydration of kaffir lime peel (Lertworasirikul & Saetan, ) and the near critical carbon dioxide extraction of anise (Shokri et al, ) by means of ANN and mathematical models, obtaining a better accuracy with ANN than with the mathematical models. It is important to note that the ANN allows to adjust the data for all the studied conditions, capturing the relationship between the variables and the behavior of the phenomena with a single model, even processing incomplete data, which would facilitate its potential industrial application.…”

“…The selection of appropriate network architecture with optimum number of neurons in the hidden layers is an important factor because it's effects upon the network convergence as well as on the accuracy of estimations. The hidden layers connect inputs x to outputs y through a series of weights w interconnected mathematically by Equation (Shokri et al, ): where w ij is the weight of the i th component of the input vector which is connected to the j th neuron; n is the number of inputs to the neuron; b i is the bias associated with the j th neuron, adding an extra variable, which can make it more powerful than a network without thresholds (Hagan et al, ); and f is the activation function that gives the nonlinear behavior of the neuron. The activation function may be linear or nonlinear, depending on the network topology.…”

“…In order to ensure a successful model, two important factors must be considered: the number of layers, and the number of neurons in each hidden layer. Since many practical problems in neural modeling could be solved with a hidden layer (Rafiq et al, ), an ANN with two layers (a hidden layer and an output layer) was used with a backpropagation algorithm (BP) as supervised training algorithm because it provides quick adjustments and is easily applicable (Shokri et al, ).The weights were adjusted using the Levenberg‐Marquardt algorithm, which converges faster than gradient‐descent methods (Hagan et al, ). In this technique, the data set is divided into three subsets: training, validation, and testing.…”

“…An ANN is able to learn from examples and generalize from simultaneous parallel calculations of its elements, thus allowing to address different issues, even if the data are incomplete or contain errors (Rafiq, Bugmann, & Easterbrook, 2001). The use of this tool to model mass transfer kinetics is recommended by several authors (Baruch, Genina-Soto, Nenkova, & Barrera-Cort es, 2004;Lertworasirikul & Saetan, 2010;Shokri, Hatami, & Khamforoush, 2011). For example, studies have been conducted using ANNs to predict the extraction kinetics of anise essential oil, with the results showing that the proposed model was more accurate than a mathematical model of diffusion (Shokri et al, 2011).…”

Method for obtaining three products with different properties from fennel (Foeniculum vulgare) seed

“…Other methods include extraction with organic solvents, supercritical CO 2 and superheated (subcritical) water as well as ultrasound and microwave-assisted extraction (Dapkevicius et al, 1998;Hawthorne et al, 1993;Jimenez-Carmona et al, 1999;Lucchesi et al, 2004;Luque de Castro et al, 1999). Several EOs are found in lipid-bearing seeds from the Apiaceae family, including anise (Pimpinella anisum), coriander (Coriandrum sativum), cumin (Cuminum cyminum), and fennel (F. vulgare) (Damjanovic et al, 2005;Eikani et al, 1999;Illes et al, 2000;Shokri et al, 2011). Fennel EO (FEO) is extracted either from seeds or biomass and each has a distinct chemical profile and uses.…”