Prediction of Cumulative Biomethane Yield Using Artificial Neural Network - Case Study of an Industrial Biogas Plant

Abstract

There are many facets to the applications of Artificial Intelligence (AI) in the energy sector however, this research focuses on the utilization of Artificial Neural Networks (ANN) as parts of AI technique to simulate and model the operating performance of an industrial biogas plant data set. In this study, eight (8) model network architectures were developed using the ANN tool of MATLAB 2016a version and it was found that the best result was obtained based on the model performance evaluation metrics used such as Root Mean Square Error (RMSE), Mean Absolute Percentage Error (MAPE), Mean Absolute Deviation (MAD) and Determination Coefficient (R2) was as a result of the combination of two activation functions namely: tansig and logsig. The model, that produced the best result was a result of the architecture that contains 2 hidden neurons and the training algorithm of Scaled Conjugate Gradient (SCG). It was also observed that the ANN-predicted network diagram is better than the observed.