Analysis of Diesel engine combustion using imaging and independent component analysis

This paper reports on the first attempt of application of independent component analysis (ICA) to 2D images of combustion-related luminosity. The images are acquired from an optically accessible Diesel engine equipped with the common rail injection system and cylinder head of a most recent generation Euro 5 engine. The original data, from a sequence of crank-angle resolved images, are treated by ICA in order to identify leading independent structures. Two main independent components (IC) are extracted from sets of luminosity images, and the coefficients of the ICs are then used for further analysis, to study the transient during a single cycle, and for the assessment of cycle variability, along with data of dynamic in-cylinder pressure, rate of heat release and integral luminosity. In the analysis of a single cycle, the two independent components appear to be clearly separated and related to combustion events near the fuel jets and near the bowl walls, respectively. The analysis over the cycles separates the mean combustion luminosity field at each crank angle from the random, erratic flame structures related to cycle variability. Quantitative analysis of the statistics of the two independent components confirms the lower variability of the jet flames and the high variability of combustion near the chamber walls. This is in agreement with the idea that the extensive impingement of the fuel sprays on to the piston bowl walls in modern, high-speed, direct injection Diesel engines is responsible for increase of unburned hydrocarbons and smoke emissions. The developed procedure, including the ICA, is fast and reliable and can be prospectively applied to many different optical engine configurations.