Moisture content prediction of banana during drying process using artificial neural network

摘要

In this study an artificial neural network was developed to predict the moisture content of banana during drying process. Thus, the experiments were performed at three levels of air temperature (70, 80, 90°C), two levels of air velocity (0.5 and 1 m/s), and two levels of thickness (3 and 5 mm).In the purposed artificial neural network, air temperature, air velocity, drying time and slices thickness were considered as inputs and moisture content was considered as output.In this study, feed forward network with the tansig and logsig transfer function and Levenbery-Marqwardt learning rule and multi-layer perceptron network with the tansig transfer function and momentum learning rule were used for modelling of experimental data. These networks were compared with nine various mathematical equations. The results showed that the feed forward network is better than multi-layer perceptron network and mathematical equations to predict the experimental data.