ABSTRACT
This work presented the influence of metal oxides as the support for silver-supported catalysts on the catalytic oxidation of diesel particulate matter (DPM). The supports selected to be used in this work were CeO2 (reducible), ZnO (semiconductor), TiO2 (reducible and semiconductor), and Al2O3 (acidic). The properties of the synthesized catalysts were investigated using XRD, TEM, H2-TPR, and XPS techniques. The DPM oxidation activity was performed using the TGA method. Different states of silver (e.g., Ag° and Ag+) were formed with different concentrations and affected the performance of the DPM oxidation. Ag2O and lattice oxygen, which were mainly generated by Ag/ZnO and Ag/CeO2, were responsible for combusting the VOCs. The metallic silver (Ag°) formed primarily on Ag/Al2O3 and Ag/TiO2 was the main component promoting soot combustion. Contact between the catalyst and DPM had a minor effect on VOC oxidation but significantly affected the soot oxidation activity.
ABSTRACT
Soot formation models become increasingly important in advanced renewable fuels formulation for soot reduction benefit. This work evaluates performance of machine learning (ML) and deep learning (DL) to predict yield sooting index (YSI) from chemical structure and proposes a tailor-made convolution neural network (CNN)-SDSeries38 for regression problem. In ML, a novel quantitative structure-property relationship (QSPR) is developed for feature extraction and the relationship between molecular structure and YSI is built by ML algorithm. In DL, SDSeries38 contains 9 feature learning modules, 1 regression module for automated feature learning and regression. It adopts standard series network architecture and modular structure, each feature learning module is a stack of convolution, batch normalization, activation, pooling layers. ML-QSPR model outperforms SDSeries38 in accuracy (RMSE = 7.563 vs 19.58), computational speed and the former applies to fuel mixtures. In DL, SDSeries38 network exceeds 10 classical CNN and provides a generic architecture enabling transfer application to other regression problem. DL application to regression is still in its infancy and there is no complete guide on how to develop specific CNN architectures for regression. Some gaps need to be filled: (1) Specially developed CNN architectures for regression are required; (2) The performances of direct transfer learning the classical CNN architectures from classification to regression are modest. A modular structure with typical function modules may provide an ideal solution; (3) Going deeper into the sequence of convolution layers improves predictive accuracy, but bears in mind to keep the number of layers below the threshold to avoid vanishing gradient.
Subject(s)
Deep Learning , Machine Learning , Molecular Structure , Neural Networks, Computer , SootABSTRACT
The influence of a platinum:palladium (Pt:Pd)-based diesel oxidation catalyst (DOC) on the engine-out particulate matter (PM) emissions morphology and structure from the combustion of alternative fuels (including alcohol-diesel blends and rapeseed oil methyl ester (RME) biodiesel) was studied. PM size distribution was measured using a scanning mobility particulate spectrometer (SMPS), and the PM morphology and microstructure (including size distribution, fractal geometry, and number of primary particles) was obtained using high-resolution transmission electron microscopy (TEM). It is concluded that the DOC does not modify the size or the microstructural parameters of the primary particulates that make up the soot agglomerates. The PM reduction seen in the DOC is due to the trapping effect, and oxidation of the PM's volatile components. The DOC performance in reducing gaseous (e.g., carbon monoxide (CO) and unburnt hydrocarbons (HCs)) and PM emissions at low exhaust temperatures was improved from the combustion of alternative fuels due to the reduced level of engine-out pollutants.
