Dp 594 en
Exhaust Gas Recirculation and Air Flow virtual sensors for internal combustion turbocharged engines
Author: Jaroslav Tabaček
The emission standards are regularly made more stringent these days which put pressure on automotive industry to create advanced engine control that require the accurate
measurements. Also, replacing hardware sensors with virtual sensors without loss of
accuracy is highly desirable due to their lower production cost. The exhaust gas recirculation (EGR) mass flow value is very important information for controlling combustion
emissions and its physical measuring is difficult, which makes it an ideal candidate for
virtual sensing.
This thesis describes a process of a creation of EGR mass flow virtual sensor for
heavy-duty turbocharged diesel engine with EGR, which could be an alternative for
Venturi sensor. The air mass flow is closely related to the EGR mass flow therefore the
air mass flow virtual sensor is created as a by-product.
A basis of good virtual sensor is reliable model, so few modelling approaches for design
of engine's air path model are reviewed. Each of them requires a model creation for each
air path component. The particular components are introduced, a polynomial model is
created for each one and real steady-state data from a diesel engine are used for their
identification. The complete air path model structures which merge the component
models are developed and discussed from robustness, accuracy and complexity point
of view. Eligible models are simulated and their applicability for further use is also
discussed.
Based on these findings, the best model is used to create virtual sensor. The virtual
sensor design respects all eligible measurements available in a common production engine, which results in simplifying the engine model structure. The final observer design
is a combination of a reduced engine model, lambda sensor measurements and a linear
Kalman filter with scheduled gain. The possibility of the implementation on Engine
Control Unit is discussed.
- Tabaček Jaroslav, mailto:tabacjar@fel.cvut.cz