Predication Emission of an Marine Two Stroke Diesel Engine Based on Modeling of Radial Basis Function Neural Networks

As testing the marine large-scale low-speed two stroke engine to determine the engine performance map for different working conditions costs too much time and money. So the prediction of the marine engine exhaust emissions modelling is developed to define how the inputs affect the outputs. The marine engine exhaust emissions were measured for different engine loads conditions. Using Radial Basis Function Neural Networks (RBFNNs) model, the exhaust emissions of a marine diesel engine was predicted. According to the results, the network performance was sufficient for all emission outputs. In the network, engine speed (N), engine load (L), fuel flow rate (FFR), air mass flow rate (AMR), scavenge air pressure(SAR), maximum injection pressure (MIP), electronic parameters and environmental conditions were taken as the input parameters, and the values of emissions were used as the output parameters. The R2 values of the modeling were 0.984, and the mean % errors were smaller. However, filter smoke number (FSN) higher mean errors were obtained due to the complexity of the burning process and the measurement errors. The aim of this paper is to establish a new approach based on RBFNNs for prediction of the marine diesel engine emissios. The results showed that the values predicted by RBFNNs were parallel to the experiment.