Misfire detection is a subject that has been deep studied during the last years and many methodologies have been developed for this purpose. Affordably detecting the misfire event and isolating the cylinder where the missing combustion took place can be considered a solved problem for engines with a limited number of cylinders. Misfire detection and in particular cylinder isolation is still challenging for engine operating conditions at very low load and high engine speed, for engines with a high number of cylinders, or when more than one misfire event is present within the same engine cycle (multiple misfire). In particular this last malfunctioning condition is very challenging, and its detection is enforced by the international regulations without requiring cylinder isolation, but only the number of misfiring cylinders. Many methodologies have been developed in the past based on the analysis of the instantaneous engine speed. The missing combustion effect on this signal is anyway very low when the number of cylinders is high and for engine operating conditions at low engine speed, giving rise to misdetection or false alarms as already mentioned. In addition when a misfire event takes place a torsional vibration is excited and shows up in the instantaneous engine speed waveform. If a multiple misfire occurs this torsional vibration is excited more than once in a very short time interval. The interaction among these successive vibrations can further generate false alarms or misdetection, and an increased complexity when dealing with cylinder isolation is necessary. The approach here presented permits enhancing existing misfire detection methods through optimized algorithm that allows correctly isolating the multiple misfiring cylinders over the entire engine operating range. This has been obtained by proper identifying the effect of the torsional vibration over the instantaneous engine speed. The identified waveform has been then used to filter out the torsional vibration effects in order to enlighten the effects of the missing combustions. In addition a proper instantaneous engine speed windowing has been introduced in order to increase the detection signal to noise ratio over the whole engine operating range. The integration of these two signal processing techniques has proven to be very effective on the engine investigated in this study, and it is easily extendible to other engine architectures. Particular care has been devoted to satisfy on-board implementation requirements in terms of memory allocation and computational power. The tests have been conducted on an L4 1.2 liter spark ignition engine mounted in a test cell. In-cylinder pressure signals have been acquired in order to validate the methodology here developed.

This content is only available via PDF.
You do not currently have access to this content.