Bayesian Inference for Combustion Phasing Estimation in Diesel Engine

This paper presents an example of combustion phasing estimation on a Diesel engine using a dedicated signal processing method applied to the engine bloc vibration signal. The proposed algorithm, based on the Bayesian inference, combines the measurement from a knock sensor with the combustion timing provided by an auto-ignition delay model. This methods does not require cylinder pressure sensor and can be implemented in real-time. The details about the methodological approach, the implementation on real time platform, and experimental results are reported in this paper. For the chosen application, we compare the estimated (knock signal) and measured (cylinder pressure signal) location of the maximum pressure gradient (LMG). The results show that this approach is relevant and the estimated LMG can be used for combustion control or diagnosis.