Misfire Detection in Heavy Duty Diesel Engines Using Knock Sensors

Abstract: In this thesis the possibility of using knock sensors for misfire detection in heavy duty diesel engines is investigated. This is of great interest due to many emission legislations getting stricter, especially in the US where robust misfire detection is mandatory for these types of engines. In order to capture the relevant vibra- tions on the engine a pre-study is made where the resonance frequencies in the cylinders are calculated which can be used for bandpass filtering the knock sen- sor signal. These bandpass filters are used to run tests where the engine is run with normal combustion and misfire on individual cylinders. The tests are made using a straight six cylinder diesel engine with a displacement of 12.7 litres. The test data is used to create threshold maps that can be used for misfire detection and for sensitivity analysis. Thereafter a detection algorithm is developed in MATLAB which involves integrating the knock sensor signal between two prede- fined crank angle degrees and then comparing it with a threshold value that can be interpolated from the threshold map. The test results show that this type of detection algorithm with these types of sensors is possible. There are also some areas of improvements presented that can make the misfire detection even more robust.