Machine Learning Model for Spark-Assisted Gasoline Compression Ignition Engine

The study showcases the strength of machine learning (ML) models in imitating the operation of an advanced engine concept - the gasoline compression ignition (GCI) - at low loads. The GCI engine is prone to exceeding the limits of criteria emissions at such loads, especially at the cold start when the catalyst is not activated. One proposition to accelerate catalyst light-off is using spark-ignition. This, however, adds an extra level of complexity in identifying an optimum operation point. The ML models can be a useful tool in guiding the engine calibration process. In this study, the ML models are trained on GCI engine experiments, covering different intake conditions, injection strategies, and spark settings. The models can predict seven engine performance parameters: fuel consumption, four engine-out emissions, exhaust temperature, and coefficient of variation (COV) in indicated mean effective pressure (IMEP). The study considered four architectures to train the dataset, namely linear regression, support vector machine (SVM), random forest and CatBoost regressor. CatBoost, which is a gradient boosting tree-based regressor, outperformed all of the considered models. The models were evaluated using the leave-one-out-cross-validation method to obtain the most representative results of the model’s accuracy. This paper shows that the seven models have successfully captured the complex relationship between the input calibration parameters and the seven desired outputs. The developed models have the potential to be utilized in optimizing GCI engine performance - especially at low loads where the engine has issues lighting off the catalyst. Coupling ML models with suitable optimization algorithms can pave the way to pinpoint the global optimum operation point in less time and with less cost than traditional calibration approaches.