Real-time misfire detection for automobile engines with medium data rate crankshaft sampling

TL;DR: In this article, a data signal containing a plurality of groups of data points generated from sampling crankshaft rotational displacement over three predetermined angular sampling windows is generated, each of the plurality of sampled data points represents a cylinder firing event.

Abstract: A method and system for detecting engine misfire in an internal combustion engine. A data signal containing a plurality of groups of data points generated from sampling crankshaft rotational displacement over three predetermined angular sampling windows is generated. Each of the plurality of groups of sampled data points represents a cylinder firing event. The data signal is conditioned, and the plurality of groups of sampled data points are mapped into single misfire detection data points. The mapped single misfire detection points are then compared to a misfire detection dynamic threshold, and a misfire detection signal is output if the misfire detection points fall outside the dynamic threshold. The misfire detection system achieves high degree of accuracy through a multi-stage signal conditioning, multi-rate signal processing and statistical decision technology and a mixed size of window sampling strategy, and is capable of being installed on a large scale in motor vehicles through use of existing onboard automotive microcontroller technology.