Evaluation of Nonlinear Estimation Methods for Calibration of a Heat-Release Model

Model-based analysis of in-cylinder pressure sensor signals has been a key component for internal combustion engine research, diagnostics and controller development during the past decades. This analysis is often based on simple thermodynamic models of the in-cylinder processes. In order for the analysis to give accurate results, the models need to be sufficiently calibrated. This paper investigates the use of the extended Kalman filter and the particle filter for the purpose of online estimation of top-dead-center offset, a convective heat-transfer coefficient and cylinder-wall temperature in a Gatowski heat-release model. Simulation results show that the filters are consistent in estimating the true parameters, that the assumed model uncertainty and heat-release noise density works as filter tuning parameters. The filters were found to be sensitive to errors on pressure-sensor offset and the cylinder compression ratio. The filters were also evaluated against experimental data and the result showed converge times of 200 engine cycles with acceptable steady-state variance for both filters.