Performance evaluation of artificial neural networks for a fish oil biodiesel fueled diesel engine: Paying a pathway to sustainable energy in environmental progress

The purpose of this study is to develop a neural networks model to predict the performance monitoring of the compression ignition engine at various injection timings (21°, 24°, 27°bTDC) using fish oil biodiesel thereby increasing the sustainability of the biodiesel and leading to greener environment. The influence of the injection timing on the engine performance, emission, and combustion are observed in a four‐stroke single cylinder, constant speed, direct injection with a rated output of 4.4 kW using fish oil biodiesel blended with diesel. Feed forward back propagation, Elman feed forward, and Cascade back propagation neural networks‐based models are created to predict the parameters like ignition delay, maximum rate of pressure rise, combustion duration, smoke, hydrocarbon, oxides of nitrogen, carbon monoxide, and carbon dioxide, brake thermal efficiency, brake specific fuel consumption, and exhaust gas temperature. Scaled conjugate gradient and Levenberg–Marquardt have been used as training functions. The predictive capability of the models is compared with each other. It is observed that Elman feed forward with trainlm is the best model with the correlation coefficient is 0.9–1 and low root mean square error. The developed artificial neural network (ANN) predicts the engine parameters with the acceptable correlation limits and proves the efficiency and precision of the model which significantly reduces the time consumption and cost incurred in real time application. Hence by using limited experimental investigations the more accurate engine performance and emissions can be determined for wider engine operating conditions using Elman feed forward ANN.