Model-Based Adaptive Air/Fuel Control for an Automotive Gasoline Engine

Introduction: The accurate control of the air/fuel ratio in automotive gasoline engines is vital for satisfying the ever more stringent emission regulations. The most common pollution abatement system for SI port-injection engines is the three-way catalytic converter, TWC. It derives its name from its ability to simultaneously reduce NO_x and oxidize CO and HC. State-of-the-art systems are capable of removing more than 98% of the pollutants. However, this can only be achieved by operating the engine within very narrow air/fuel ratio limits. To obtain the desired air/fuel ratio, the engine management system has to compensate all the disturbances caused by the dynamic systems that are involved.

The fuel control system comprises both a feedforward and a feedback controller, where the former calculates the necessary amount of fuel to be injected based on the information about the actual operating point of the engine. The feedback controller adjusts the amount of fuel calculated on the basis of measurements by an air/fuel-ratio sensor and thus compensates for any disturbances which are not considered by the feedforward controller.

The fuel path dynamics of an engine are not only a function of the operating points but they also vary due to the ageing of its components. Hence the increasing demands on the performance of the air/fuel ratio controller no longer can be met by one globally robust controller. More sophisticated adaptive control algorithms will thus be required.

During its lifetime the air/fuel ratio sensor undergoes an ageing process which results in a substantial change of its dynamics. To some extent, a robust controller can mitigate this change. However, the trade-off between robustness and performance may be in conflict with the increasing demands on the effectiveness of the control system. Figure 1 shows a new control system with the ability to identify changes in the sensor dynamics and adapt its parameters. It supersedes the need for global robustness and circumvents the mentioned trade-off.

Achievements:

In a first step a (non-adaptive) robust control law has been developed. Figure 2 shows a Nyquist diagram of the open control loop.

The circles indicate a bound for the distribution of the Nyquist curves due to the uncertainty of the plant parameters. Due to the variation of the engine’s load and rotational speed, the dynamics of the fuel path experience considerable changes. Since load and rotational speed can be measured easily, the controller parameters are gain-scheduled. The resulting controller has been tested on an engine test bench. Figure 3 shows a comparison of measurement and simulation results of the response of the control system to input disturbances.

Goals:

In a further step the ageing processes of the sensor need to be analyzed. Based on this knowledge an adaptive controller is to be designed.

In order to determine the variations of the sensor dynamics during its lifecycle different scenarios need to be looked at. One scenario may be the “regular” ageing of the sensor as a slow and steady process, whereas another scenario may cover sudden changes due to abnormal events during operation. Based on the scenarios considered possible, a suitable model of the sensor which is capable of describing the most important implications of all possible sensor variations is to be developed.

The adaptation block of the controller basically consists of two components, one being responsible for the tracking of the changes in the sensor dynamics and the other being accountable for a suitable adaptation of the control law. However, it is of crucial importance that the resulting adaptive controller still offers robustness margins for disturbances other than changes of the sensor dynamics. Since the controller also contains gain-scheduled parameters for the various operating points of the engine, the adaptation mechanism must not be affected by these changes.