A robust algorithm for identifying different types of engine misfires

This paper introduces a robust algorithm for engine misfire detection that can detect and identify different kinds of engine misfires under various speed and load conditions. The overall algorithm consists of two main steps. First, it employs a frequency domain pattern classifier to detect and identify the type of misfire, and then it utilizes a generalized likelihood test to locate the misfire instances. This paper is primarily focused on the first task only. Results of extensive tests on a six-cylinder engine are presented.