A preliminary study modelling NO emission by subset selection using a genetic algorithm and in-cylinder parameters

Introduced in this paper is the application of a genetic algorithm to perform subset selection to reduce the number of input parameters into a time history dependent model for the estimation of NO emission. For this work, a bespoke cycle, denoted as a sweep test, was utilised to provide the data for training the model. Input parameters into this model are in-cylinder parameters: indicated mean effective pressure, engine speed, peak pressure, peak pressure timing and the maximum rate of pressure rise, in addition to: intake air flowrate, instantaneous fuel consumption and boost pressure. Shown was that these input parameters allowed a high correlation between the estimated NO emission and the measured NO emission on the NRTC. A key advantage of subset selection is in being able to interpret the model itself to gain a physical understanding of what input parameters influence NO emission. A significant outcome from this work was in identifying that, for the engine under investigation, a time history of 8.5 s is needed to accurately estimate NO emission.