Hybrid Model of the Gasoline Engine for Misfire Detection

This paper proposes a novel hybrid model for an internal combustion engine, with the power generated due to combustion as the input and the crankshaft speed fluctuations as the output. The individual cylinders of the engine are considered as subsystems for which a nonlinear model, based on the physical principles, is derived. The proposed model is linearized at an operating point, and a switched linear model is formed. The simulation results of the proposed model are validated by matching the results with the experimentally observed data. Using the properties of the validated model, it is shown that the crankshaft speed variations observed in the engine are a Markov process. A novel algorithm that is based on the Markov chain is proposed to detect the misfire in the spark ignition engines. In the ensuing engine rig experiments, an igniter misfire is introduced in the system and is successfully detected. The analysis of the data shows that the engine also has an air leakage in a cylinder (a developing misfire), which is experimentally confirmed later.