Search Results

In order to fulfil emission legislation and achieve good drivability of combustion-engine-powered vehicles, information about the air charge and feedback about the engine condition is necessary. In current systems, different sensors are used, e.g. the MAP (manifold air pressure) sensor and a lambda sensor. Aiming at reducing costs, efforts are being made to reduce the number of sensors while still retrieving the necessary information. Various engine speed based functions are state-of-the-art for automotive engines, e.g. for fuel-calibration, misfire-detection etc. Those functions evaluate the engine speed fluctuations during a working cycle induced by combustion. For multiple-cylinder engines, those influences are overlapping, therefore evaluation possibilities are limited. The work presented is based on the effect that at a single-cylinder engine, there is no overlap of combustion influences of various cylinders on the crankshaft.

Contemporary diesel engines are high-tech power plants that provide high torques at very good levels of efficiency. By means of modern injecting-systems such as Common-Rail Injection, combustion noise and emissions could be influenced positively as well. Diesel engine are therefore used increasingly in top-range and sports cars. Today's production ECUs have no or only very low feedback regarding the process in the combustion chamber. As long as this data is missing, the design of the maps in the ECU can only be a compromise, since production tolerances and aging processes have to be considered in advance. Disturbances in the combustion process may not be detected at all. If more knowledge about the course of combustion is provided, especially the start of combustion (SOC), various operating parameters, such as the pilot injection quantity or the beginning of current feed to the injector, could be adjusted more precisely and individually for every cylinder.

In order to optimize the performance and emission of engines, advanced control and diagnostic systems require detailed feedback information about the combustion process. In this context, cost-effective solutions are of interest. The contribution describes a method for reconstructing cylinder-individual features of each combustion cycle by processing the instantaneous fluctuations of the engine speed and the in-cylinder pressure of one cylinder. Model-based torque estimation, analyzing both of the signals simultaneously, provides an accurate estimation of the mean indicated pressure. Using this method, a new algorithm for advanced misfire detection is presented. Furthermore, a new pressure model with a feasible number of parameters is proposed. It is combined with the torque estimation in order to reconstruct the unknown pressure traces of the cylinders not equipped with sensors.