A Graphical User Interface for Model Reduction of Complex Fuels Based on Principal Component Analysis and Artificial Neural Networks

摘要

A graphical user interface (GUI) for model reduction of complex fuels is developed. The GUI is based on principal component analysis (PCA) for model reduction and artificial neural network (ANN) for the reconstruction of thermo-chemical scalars. The target complex fuels' chemical mechanisms are characterized by hundreds or thousands of chemical species and thousands or tens of thousands of chemical reactions. An account for the detailed chemistry of such fuels is computationally expensive. The GUI is designed to generate a significantly-reduced set of scalars, principal components (PCs), that describe the complex composition space originally represented by the full set of thermo-chemical scalars. The approach is a viable alternative to chemistry reduction schemes and may be used in combination with these schemes. The GUI is built with many features that enable the selection of a subset of the thermo-chemical scalars for PCA, the determination of the number of PCs that reproduce the original scalars' variance, the evaluation of chemical source terms and diffusion coefficients of the PCs and the generation of output for a posteriori calculations in computational codes. Validation of the GUI is presented to illustrate its performance.