Submitted to: International Journal of Food Engineering
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 2/16/2009
Publication Date: N/A
Citation: N/A Interpretive Summary: Changes in food processing practices and opportunities for innovative food products have spurred interest in processes that extract desirable compounds from food. Several published models of the extraction process have described extraction in the liquid phase or solid phase, but little information is available for using advanced analytical techniques to predict the rate of extraction. An analytical model known as the Radial Basis Function (RBF) artificial neural network was found to predict the rate and yield of extraction with less than 3% error. Additionally, the RBF network provides fast computation and articulates relationships between the extraction process and physical properties. Computations using the RBF network can provide accurate, consistent, and reliable predictions of the rate and yield of extracted food compounds.
Technical Abstract: An artificial Radial Basis Function (RBF) neural network model was developed for the prediction of mass transfer of the phospholipids from canola meal in supercritical CO2 fluid. The RBF kind of artificial neural networks (ANN) with orthogonal least squares (OLS) learning algorithm were used for modeling. The mass transfer rate was controlled by pressure, temperature, CO2 flow rate, and material matrix property. The proposed model assumes a three-layer structure with a fast back-propagation learning algorithm. Different temperatures, pressures, and solubility were used to train the proposed mass transfer model of supercritical CO2 fluid. The effectiveness of the proposed neural network approaches is demonstrated by simulation. The results showed that the RBF network is consistent in predicting mass transfer rate in supercritical CO2 fluid, and is appropriate for mass transfer modeling to monitor the separation process.