This paper reports on the application of independent component analysis (ICA) to 2D cycle-resolved images of the luminous combustion, collected in a port fuel injection spark ignition optically accessible engine. The method of ICA is employed for the identification of the independent spatial components, which along with the corresponding time dependent coefficients, represent the spatiotemporal evolution of the luminous combustion during a single cycle and over a number of cycles. ICA is applied on the data preprocessed by proper orthogonal decomposition (POD). POD-filtered ICA permits to determine only few dominant and relatively easy interpretable independent components. Successively, and for comparison, ICA is applied to the non-truncated data. It is demonstrated that ICA applied to single cycle permits to extract independent structures, clearly separated in time. The time dependent coefficients correlate well with the integral flame luminosity, and characterize time evolution of the combustion pattern in the chamber. The analysis over several cycles shows that independent components carry information about the dominant morphology of the cyclic variations. The low correlation of the corresponding (in terms of time succession) components for successive cycles is in agreement with the high spatial variability of the combustion process in spark ignition engines, mainly due to the combustion of fuel pockets created in the combustion chamber by the injection process. A different approach is then proposed, enabling separation of sources corresponding to objects moving in the field of view. Each whole sequence of 2D images (video) is considered as a single observed mixture of independent signals. Both linear and nonlinear mixing models are considered, and artificial examples are used to illustrate features and limitations of linear ICA when applied to nonlinear mixing. The procedure is successfully applied to a video reporting the motion of a luminous flame over a portion of the cylinder head. ICA separates the moving source from the background, with a mechanism explained by a nonlinear mixing model.

Independent component analysis of cycle resolved combustion images from a spark ignition optical engine

Continillo G;
2016-01-01

Abstract

This paper reports on the application of independent component analysis (ICA) to 2D cycle-resolved images of the luminous combustion, collected in a port fuel injection spark ignition optically accessible engine. The method of ICA is employed for the identification of the independent spatial components, which along with the corresponding time dependent coefficients, represent the spatiotemporal evolution of the luminous combustion during a single cycle and over a number of cycles. ICA is applied on the data preprocessed by proper orthogonal decomposition (POD). POD-filtered ICA permits to determine only few dominant and relatively easy interpretable independent components. Successively, and for comparison, ICA is applied to the non-truncated data. It is demonstrated that ICA applied to single cycle permits to extract independent structures, clearly separated in time. The time dependent coefficients correlate well with the integral flame luminosity, and characterize time evolution of the combustion pattern in the chamber. The analysis over several cycles shows that independent components carry information about the dominant morphology of the cyclic variations. The low correlation of the corresponding (in terms of time succession) components for successive cycles is in agreement with the high spatial variability of the combustion process in spark ignition engines, mainly due to the combustion of fuel pockets created in the combustion chamber by the injection process. A different approach is then proposed, enabling separation of sources corresponding to objects moving in the field of view. Each whole sequence of 2D images (video) is considered as a single observed mixture of independent signals. Both linear and nonlinear mixing models are considered, and artificial examples are used to illustrate features and limitations of linear ICA when applied to nonlinear mixing. The procedure is successfully applied to a video reporting the motion of a luminous flame over a portion of the cylinder head. ICA separates the moving source from the background, with a mechanism explained by a nonlinear mixing model.
2016
SI engine; Cycle-to-cycle variations; Independent component analysis; Optical imaging
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/20.500.12070/3799
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 18
  • ???jsp.display-item.citation.isi??? 17
social impact