مدلسازی سینتیک خشک شدن لایه نازک میوه زرشک (Berberis Vulgaris) به کمک شبکه عصبی مصنوعی شیوا گرجیان 1، تیمور توکلی هشجین 2، محمد هادی خوش تقاضا 3

In order to investigate the effect of different levels of temperature, velocity and pretreatment on the drying time of thin layer of Barberries (berberis vulgaris), an experiment using a factorial experiment conducted carried out based on completely randomized block design with three levels of temperature and three levels of velocity for untreated berries and treated berries with heat shocking and "6% k₂co₃ + 3% olive oil" emulsion with three replicates. Drying times were affected by temperature and pretreatment at the probability level of 99%. The maximum drying time was recorded 2920 minutes for untreated berries at the temperature and velocity of 60 ºC and 0.3 m/s respectively and the minimum drying time was recorded 70 minutes for treated berries with k₂co₃ and olive oil emulsion. In our study multi layer perceptron (mlp) Neural Network with an adjustment learning algorithm of Levenberg-Marquardt with sigmoid logarithm and sigmoid tangent functions were used. The best topology of MLP with LM learning algorithm and Tansig threshold function can predict the moisture content was 4-30-16-1 with correlation coefficient of 0.9992 and actual error of 0.00025. Besides the best topology of this neural structure with Logsig threshold function can predict the moisture content was 4-25-5-1 with correlation coefficient of 0.9991 and actual error of 0.00032. These results indicate ability of artificial neural network to model and predict drying process.