Application of Artificial Neural Network in Determining Performance Profile of Compression Ignition Engine Operated with Orange Peel Oil-Based Biodiesel
Samson Kolawole Fasogbon
Department of Mechanical Engineering, University of Ibadan, Ibadan, Nigeria and Centre for Petroleum, Energy Economics and Law, University of Ibadan, Ibadan, Nigeria.
Chukwuemeka Uguba Owora *
Department of Mechanical Engineering, University of Ibadan, Ibadan, Nigeria.
*Author to whom correspondence should be addressed.
Abstract
Literature including one of our previous studies have confirmed the environmental friendliness of orange peeled oil biodiesel (OPOB) when applied to run compression ignition (CI) heat engines. There is also high degree of compatibility of physicochemical properties of OPOB with fossil diesel. However, there is limited knowledge on its performance indices in the same heat engines. This perhaps may have been due to few interests shown by researchers in the area or obviously due to difficult time and other quantum resources required in conducting the rigorous engine tests. To this end, this work conducted experimental study of performance profile of OPOB in direct injection CI engine; and afterwards applied artificial neural networks (ANNs) to ascertain the engine brake thermal efficiencies (BTE) and brake specific energy consumptions (BSEC). The ANN utilized the Levenberg Marquardt (LM), Scaled Conjugate Gradient (SCG) and Gradient Descent with Momentum and Adaptive Learning (GDX) training algorithms for the performance prediction. The choice of the three algorithms was to effect better comparative assessment. The input variables of the neural network were brake load, orange oil-diesel mixture percentages and engine speed. Statistical parameters such as correlation coefficient (R), mean absolute percentage error (MAPE) and root mean squared error (RMSE) were employed to investigate the performance of the neural networks. Among the three training algorithms, the Levenberg Marquardt trained algorithm estimated the BTE and BSEC with highest precision and accuracy; and lowest error rates. From the study, it is concluded that the performance profile of compression ignition heat engines operated with orange peel biodiesel compares favourably with fossil diesel. It also affirmed that Artificial Neural Network is a reliable tool in the prediction of performance indices of compression ignition engines when run with orange-peel oil based biodiesel.
Keywords: Artificial neural network, levenberg marquardt training algorithm, parametric performance indicators, orange peel oil biodiesel, engine performance, direct injection compression ignition engine