Misfire detection in ic engines using finite state automata

This paper presents a novel approach for the detection of misfire in an automotive engine using a finite state machine methodology. The algorithm is simpler and can easily be implemented on a microcontroller or more efficiently; using both microcontroller and FPGA. The other novel thing is that this method does not require system knowledge in the form of a mathematical model; instead knowledge of system behavior in general is sufficient for fault detection using this approach. The method uses crankshaft speed fluctuations as input and compares them with an ideal engine behavior to generate the residuals to identify the cylinder showing the maximum non-uniformity. The index of cylinder showing maximum nonuniformity is fed to a Mealy machine. The output of Mealy machine is processed for fault identification in the presence of noise. The method is capable to detect a single as well as multiple misfire events. The method is applied to a four stroke, four cylinder IC engine and simulation results are presented.