Search

After the energy crisis in 1970s, the impacts of increasing numbers of vehicles, including natural gas vehicles (NGVs), on global air quality has highlighted the demand to optimise existing vehicle aftertreatment systems. Increasingly stringent emission regulations in many countries lead to challenges for automobile manufacturers to meet carbon monoxide (CO), total hydrocarbon and nitrogen oxides emission standards, thus further optimisation of current aftertreatment technologies is essential. Optimisation of the three-way catalysts (TWCs) converters is a challenge due to the spatially-distributed and transient nature of the operation of vehicles. The conventional method of investigating catalysts properties by analysis of the chemical species at the reactor outlet (end-pipe analysis method) provides large amounts of valuable knowledge, but due to the lack of intra-catalyst data, limits the possibility of designing and optimising catalytic converters. Spatially-resolved experimental techniques provide a solution to this drawback, by accessing spatiotemporal intra-catalyst information to help in the construction in more accurate and predictive models. Using spatially and/or temporally mapping of aftertreatment reactions in structured catalysts provide a density of information which is crucial for the optimisation of kinetics modelling for engine operations, however, difficult to access using traditional end-pipe analysis methods. The main requirement of catalytic converters modelling is to accurately predict their performance through mathematical expressions, representing physiochemical processes of these aftertreatment systems. These processes include mass transfer and surface reactions associated with the chemistry of interest. As mass transfers from the gas phase to the catalyst surface have significant influences in determining the kinetics of the reactions, the accuracy in describing these processes is crucial. The aim of the present work is to develop a novel methodology to determine, experimentally, the mass transfer coefficients in monolith catalysts using a spatially-resolved approach.

Learning view-invariant representation is essential to improving feature extraction for action recognition. Existing approaches cannot effectively capture details for human actions due to fast-paced gameplay and implicit view- dependent representation. In this paper, it goes beyond recognizing human actions from a fixed view and focusing on action recognition from an arbitrary view. This paper purposes a method to build an efficient data generating pipeline due to lack of original input data. This paper also provides a pipeline combining capturing modified I3D human actions features and use Multilayer Perception to achieve human action recognition and classification. The use of information captured from combination of virtual and real-life data, as well as different viewing angles, leads to high classification performance.