Vibration Based Virtual Sensing of Nitrogen Oxide Emission in CI Engines

Abstract

In addition to solid legal basis and strong regulatory authority, comprehensive and robust test techniques that reflect real-driving and in-use vehicle emissions are also key elements for achieving ever strict environmental requirements. However, most of the existing portable emissions measurement systems (PEMS) is expensive and bulky, making it impractical to roll out to the vehicle market. The combustion behavior is one of the most important excitation sources of engine vibration and noise emissions. By analyzing structural vibrations, it is possible to assess the in-cylinder combustion conditions, which in turn makes it possible to evaluate the NOx emissions (closely related to combustion temperature) and to achieve cost-effective non-intrusive on-line monitoring of engine in real-driving emissions (RDE). Firstly, a time-frequency reconstruction method is used to estimate second derivative of in-cylinder pressure curve based on measured vibration signal. Good consistency between reconstructed second derivative of in-cylinder pressure curve and measured curve con-firms the rationality of the proposed reconstruction method. Then, a virtual sensing method for NOx emissions using principal component regression (PCR) analysis with reconstructed in-cylinder pressure curve from measured structural vibrations has been presented. The results show that the PCR models have a good prediction of NOx emissions under various operating conditions. The coefficient of determination (R2) between the predicted value and the measured value is 0.971–0.995.