Predicting of Moisture Ratio for Bitter Leaf (Vernonia amygdalina) Using Artificial Neural Network

Abstract

This was developed artificial neural network model for predicting the moisture ratio of bitter leaf (Vernonia amygdalina) utilizing drying time, drying rate and temperature as input parameters. Bitter leaf was dried in an oven dryer at four different temperatures (40 oC, 50 oC, 60 oC and 70oC) and the experimental data gotten was used to train the network using different configurations which consisted of different number of neurons and transfer functions. The best performing model was adjudged using the mean square Error value and was selected to predict the moisture ratio of bitter leaf. It consisted of four hidden neurons and used the tangent sigmond transfer function. The model gave a mean square error value of 0.00011205 and an R-value of 0.999. The regression coefficient (R2) value for the correlation between the predicted and experimental outputs for the model was 0.998. These results proved that the model developed showed good generalization. ANN helps in predicting fast, accurate, efficient and a reliable tool. It is recommended that other drying techniques be used for drying the product (bitter leaf).