The NOx emission estimation by the artificial neural network: the results

The paper presents the preliminary investigations of nitric oxides (NOx) estimation from marine two-stroke engines. The Annex VI to Marpol Convention enforce to ship - owners necessity of periodical direct measurements of the NOx emission from the ship engines. It is very expensive procedure but with a low accuracy. Presented investigations show the possibility of estimation the NOx emission without direct measurements but using the artificial neural network (ANN). The paper presents chosen structures of ANN's usable to NOx emission estimation, the laboratory investigations and effects of estimation NOx emission. The paper reports the effects of investigations during different points of load the engine, with constant and changeable air/fuel equivalence ratio. The detailed results of measurement and calculation of NOx concentration in the exhaust gases of marine two-stroke diesel engine were presented. The results show that the multilayer perception neural network (MLP) is sufficient to NOx emission estimation during onboard exploitation. The MLP network with 15 neurons in the hidden layer has best accuracy for data sets collected during running the engine with speed equal 200 rpm and constant air/fuel equivalence ratio and for both considered speeds of the engine with changeable air/fuel equivalence ratio.