Analysis of Energy and Exergy Variations in Osmotic Dehydration of Apple Using GMDH Neural Network and Hybrid ANN-GA

In this research, energy and exergy apple were estimated using a Group Method of Data Handling (GMDH) neural network and a hybrid artificial neural network-genetic algorithm (ANN-GA). Osmotic and natural samples were tested in the form of apple cubes with a height of 5 mm and sides of 10 mm ×10 mm, 8 mm × 8 mm and 6 mm × 6 mm dried by fluidized bed method at three air velocities of 4, 6, and 8 m/s and temperatures of 40, 45, and 50℃. Results showed that energy consumption, energy use ratio, exergy efficiency, and exergy loss elevated by increasing the temperature and air velocity and by reducing the sample sizes in both natural and osmotic samples. Also, it was observed that the GMDH neural network for energy consumption, energy use ratio, exergy loss, and exergy efficiency have linear correlation coefficients (R) of 0.95097, 0.92202, 0.91464, and 0.91258, respectively, suggesting that it outperforms the ANN-GA. The weakest performance of GMDH network was associated with the exergy efficiency. The ANN-GA had the best performance in energy consumption and lowest performance in the exergy loss.

Analysis of Energy and Exergy Variations in Osmotic Dehydration of Apple Using GMDH Neural Network and Hybrid ANN-GA Keywords:

Apple , Artificial Neural Network , Genetic Algorithm , Osmotic Dehydration , Group Method of Data Handling Neural Network