Abstract
Keywords - Turbocharged diesel engine, simulation, non-linear model, air management, real-time control.
Abstract - The present work shows developments made in air management control in a turbocharged diesel engine. The platform enables non-destructive testing on a virtual engine that closely resembles a real engine. A virtual engine is more than just an experimentally adjusted non-linear model it also handles sensor and actuator problems (dynamic non-linearities such as hysteresis, measurement noises, sampling aspects, etc.) using data acquisition cards.
A non-linear model using first principles (mass and energy conservation, fluid dynamics, etc.) has been developed for studying the air management problem in a turbocharged diesel engine. Experimental data has also been used to adjust the model. Results from this model show behaviour similar to a real engine. The model has been programmed on a dedicated computer using a real-time simulation platform built from Matlab/Simulink and xPC target (a dedicated computer with a real-time operating system). The xPC target has data acquisition cards to interface with real-time external controllers and simulate sensors and actuators. Together, these components form the virtual engine. Once a virtual engine is available, it can be used like a real engine. Advanced control and identification techniques can be evaluated in near real conditions but without malfunction risks.
The second part of the development platform consists in a data acquisition and control oriented device such as the PXI-RT from National Instruments. This type of devices facilitates flexible signals connections that enable easy monitoring of all significant engine variables. Sophisticated control algorithm implementation is also possible thanks to the programming facilities offered by Labview RT.
The work shows the results obtained for air management control in a turbocharged diesel engine, specifically, manifold pressure and air mass flow. These variables are often required to achieve better power performance and lower emissions. PI independent and multivariable predictive control have been tested for several situations. Interesting results have been obtained in each case and these form the initial steps to a successful advanced air management control.