MODELING OF BIODIESEL YIELD FROM AFRICAN PEAR SEED (DYACRODES EDULIS) OIL USING ARTIFICIAL NEURAL NETWORKS
The present study involves the application of artificial neural network (ANN) model to predict the yield of methyl ester derived from the seed oil of African pear (Dyacrodes edulis) using Levenberg-Marquardt algorithm. The propagation algorithm used for network training where 52 percent of the data was taken for training set, 13 percent for validation and the rest of the data for the test set. The regression coefficients of training, testing, validation and overall model developed using ANN had very good values of correlation coefficient ‘R’ (1.000, 0.9951, 0.8456 and 0.97171). It consists of three layers: input layer with four input variables, hidden layer with ten hidden neurons and an output layer with single output variable. The statistical analysis of the ANN model performance gave a mean squared error (MSE) of 1.3110, mean absolute error (MAE) of 1.6588, mean absolute deviation (MAD) of 1.3112 and a high value of correlation coefficient (R2 = 0.9447. This shows that the model performance is statistically validated. The overall result shows that ANN is an efficient method for empirical modeling and optimization of biodiesel yield. The results recommend that ANN provides an excellent means of identifying patterns in data and effectively predicting biodiesel yield based on investigating inputs.