Controlled Combusion Engine (CCE) Technology Using Real Time Neural Network Strategy

This proposal addresses the development of new control technology for refining the performance of diesel engines. the lack of real time control of the combustion event, combined with the locally stoichiometric nature of combustion in the conventional deisel engine, limits potential for this class of engines. As a consequence of these operating conditions, unnecessarily high temperature and flame speeds are acheived and the system suffers from a loss of available energy, at a considerable penalty in fuel economy, as well as generating unnecessarily high levels of exhaust pollutants. The objective is to develop real time control of the actual combustion event in order to optimize both fuel efficiency and the reduction of pollutants. This will require an investigation into new sensor and actuator technologies as well as control techniques based on neural network concepts. Although neural network control is applicable to an engine utilizing conventional diesel combustion technology, it is particularly beneficial when combined with a truly lean burning mixture and the concept of Direct Injection Stratified Charge Combustion (DISCC).