Using Artificial Neural Networks for Representing the Brake Specific-Fuel Consumption and Intake Manifold pressure of a Diesel Engine

It is very common that diesel engines are equipped with exhaust gas recirculation (EGR) and variable geometry turbo-charger (VGT). Due to more and more stringent emissions laws and high pressure on fuel economy, new technologies, such as, variable valve actuation, are introduced to diesel engines. The additional degree of freedom caused by the new technologies will cost ECU and increase the complexity of the mapping and calibration. Therefore, neural networks are needed to represent intake manifold pressure and BSFC. On the other hand, in the general air-path control, intake manifold pressure and the break specific fuel-consumption (BSFC) are important variables. It is essential that they can be represented by neural networks. In this paper non-linear autoregressive exogenous input (NLARX) neural networks are used to represented the intake manifold pressure and BSFC, respectively. It is shown that NLARX neural networks could represent intake manifold pressure and BSFC quite well.