|
|
|
|
|
| International Agrophysics |
| wydawca: | Instytut Agrofizyki
im. B. Dobrzańskiego
PAN
w Lublinie |
| ISSN: |
0236-8722 |
vol. 25, nr. 1 (2011)
|
|
|
poprzedni artykuł wróć do listy artykułów następny artykuł
|
|
|
Designing and optimizing a back propagation neural network to model a thin-layer drying process
|
|
| (pobierz wersję PDF )
|
|
|
S. Gorjian, T. Tavakkoli Hashjin, M.H. Khoshtaghaza
|
|
|
Department of Agricultural Machinery Mechanics, College of Agricultural Engineering, Tarbiat Modares University, Tehran, Iran |
|
|
vol. 25 (2011), nr. 1,
pp. 13-19
|
|
|
streszczenie
In the present study, the application of a back propagation) neural network for the prediction of moisture content of barberry fruit (berberis vulgaris) during drying was investigated. The important parameters, namely, pretreatment (no pretreatment, heat shocking, olive oil + K2CO3), air drying temperature (60, 70 and 80C), air drying velocity (0.3, 0.5 and 1 m s-1) and time (s) were considered as the input parameters, and moisture content as the output of the artificial neural network. Experimental data obtained from a thin-layer drying process were used for training and testing the network. Several criteria such as training algorithm, learning rate, momentum coefficient, number of hidden layers, number of neurons in each hidden layer, and activation function were given to improve the performance of the artificial neural net-work. The best training algorithm was Levenberg-Marquard with the least mean square error value. Optimum values of learning rate and momentum for the artificial neural network with gradient descent momentum training algorithm were set at 0.5 and 0.7, respectively. The optimal topologies were 4-20-1 and 4-25-5-1 with mean square error values of 0.00318 and 0.001 with logsig activation functions. Also, with tansig activation function, the optimal topologies were 4-20-1 and 4-15-15-1 with the mean square error values of 0.00293 and 0.00130. There was no significant difference between the two activation functions in optimal topologies. There was good correlation between the predicted and experimental values in optimal models.
|
|
słowa kluczowe
thin-layer, artificial neural network, moisture content, training algorithm
|
|
|
|
|
|
|
