Model Based Emissions and NVH Optimization

Throughout the years, the legislations which drive the vehicle development have experimented constant evolutions. Especially when it comes about pollutant emissions and NVH ( Noise, Vibration & Harshness). However, it is complex to understand which calibration strategy promotes the best balance about lowest levels of emissions, vibrations, and noise if considered the number of inputs to be explored, becoming the searching for the optimum calibration a huge challenge for the development engineering team. This work proposes a methodology development in which complex problems can be solved by model based solutions regarding the best balance finding of emissions reduction and noise attenuation. The methodology is based in machine learning approach which provides a virtual behavior of engine phenomena making possible a wider comprehension of the problem and hence the opportunity to explore enhanced solutions. The study case scenario used to apply the method was a 6.4 liters engine which presented a huge noise during the catalyst heating phase. The method allowed to find a calibration propose that could eliminate the noise without harming emissions levels. The methodology presented the capacity to solve the problem in a shorter time and higher quality than the conventional method. Based on its success, research is ongoing to refine the methodology to be applied in several other powertrain configuration.