Hamid Bakhshabadi; Habibollah Mirzaee; Alireza Ghodsvali; Seyed Mahdi Jafari; Aman Mohammad Ziaiifar
Abstract
Introduction: Black Cumin seed (Nigella sativa L.) as one of the novel edible oil resources used commonly nowadays as seasoning in food product industries due to considerable medicinal properties and high nutritional impacts. Oil extraction by pressing method as an approach compared to other methods ...
Read More
Introduction: Black Cumin seed (Nigella sativa L.) as one of the novel edible oil resources used commonly nowadays as seasoning in food product industries due to considerable medicinal properties and high nutritional impacts. Oil extraction by pressing method as an approach compared to other methods including solvent extraction is faster, safer and cheaper. In the oil extraction process, the preparation of the seeds is a substantial stage for obtaining oil with high quality and efficiency. Microwaves are electromagnetic waves that have a frequency ranged from 300 MHz to 300 GHz with corresponding wave lengths ranged from 1 mm to 1 m. On the other hand the artificial neural network as a powerful predictive tool in a wide scale of process parameters has been studied on an industrial scale in this research in order to achieve a simple, rapid, precise as well as effective model in the oil extraction of Nigella sativa L seed.
Materials and Methods: In the present study Black Cumin seeds after preparation including cleaning and passing resistance of the samples in front of air and moisture were stored in a plastic bag until the day of experiments. Then, they have been pre-treated with microwave within different processing times (90, 180 and 270 S) and powers (180, 540, and 900 W). Afterwards, seeds’ oil was extracted by screw rotational speed levels approach (11, 34 and 57 rpm). Different selected parameters including extraction efficiency, oil acidity value, color and oxidative stability were determined. To predict the alterations trend, the artificial neural network (ANN) design in MATLAB R2013a software was used.
Results and Discussion: According to MSE and R2 values obtained in this study, feed forward neural network with transfer function sigmoid hyperbolic tangent and Levenberg- Marquardt learning algorithm with topology of 3-10-5 (input layer with 3 neurons– a hidden layer with 10 neurons – output layer with 5 neurons) were selected as the optimal neural network with R2 more than 0.995 and MSE equal to 0.0005. Also, the results of the optimized and selected models were evaluated and these models with high correlation coefficients (over 0.949), were able to predict the changes' trend. According to the complexity and multiplicity of the effective factors in food industry processes and the results of this research, the neural network can be introduced as an acceptable model for modeling these processes. By determining the activation function in neural networks which was a function of sigmoid hyperbolic tangent in this study and also, with having the amounts of weight and bias, the connections created by the neuro-fuzzy model can be extracted. By defining this simple created mathematical equation, in computer software such as Excel, we can have a useful, simple and accurate program for predicting the desired parameters in the process of oil extraction by using microwave pre-treatment. Due to high accuracy of neural model we can trust the prediction of these models with high confidence, and this model can be used to optimize and control the process, which can lead to saving in energy and time, and on the other hand, can create a better final product.
Alireza Ghodsvali; Mohsen Mokhtarian; Hamid Bakhshabadi; Fatemeh Arabamerian
Abstract
Malting is a complex biotechnological process that includes steeping; germination and drying of cereal grains
under controlled conditions of temperature and humidity. In this research malting process parameters were
predict by modular neural network with different activation function included, logsig-logsig, ...
Read More
Malting is a complex biotechnological process that includes steeping; germination and drying of cereal grains
under controlled conditions of temperature and humidity. In this research malting process parameters were
predict by modular neural network with different activation function included, logsig-logsig, tanh-tanh, logsigtanh,
logsig-identity and tanh-identity. Steeping time (x1) and germination time (x2) were used as input
parameters and hot water extract (y1), malting yield (y2) and enzyme activity (β-Gluconase) (y3) were selected as
output parameters. The results showed that using perceptron neural network with tanh-identity activation
function had the best result among all of activation functions to predict effective parameters of malting process.
As well, this network was able to predict hot water extract, malting yield and enzyme activity (β - Gluconase)
with R2 value of 1, 0.984 and 0.995, respectively.
