Development of a hydrogen car & emissions modelling using artificial intelligence tools

Hydrogen will dramatically change the way we live in the future. While some published literature details the practical uses and applications of hydrogen as a fuel, little evidence is available in the public domain on the conversion technologies and associated emission data of this emerging technology. In this thesis, the conversion of a petrol engine to run on hydrogen is proposed and executed from first principles. Research was directed on key areas such as fuel storage, delivery and safety systems as well as integration of an effective engine management system. The performance of the vehicle in TARGA Tasmania 2006 clearly establishes its reliability and demonstrates a functional prototype that attests to the viability of hydrogen as an alternative fuel. As part of this investigation, neural network models based on experimental data acquired from a comprehensive range of engine operating conditions are used for the prediction of emissions such as carbon dioxide, carbon monoxide, unburnt hydrocarbons and oxides of nitrogen. These predictions are based upon study of the qualitative and quantitative effects of engine process parameters such as mass air flow, RPM, air-to-fuel ratio, exhaust gas temperature and engine power on the output emissions. The predictive models show excellent accuracy in estimating emissions for various engine operating conditions on both hydrogen and petrol. This model when integrated into the existing engine management system could act as a \virtual sensor\" for emission prediction. This provides an avenue for further reduction of emissions based on greater understanding of the effects of engine process parameters on emissions."