Modeling Performance and Emissions of a Spark Ignition Engine with Machine Learning Approaches 2022-01-0380

In the foreseeable future, the growing energy crisis and environmental pollution problem pose severe challenges to the automobile powertrains and exhaust systems. However, conventional optimization methods, including multi-dimensional computational fluid dynamics model and bench experiments, are very time-consuming or expensive. Adding the application of data-driven models to engine research and development has the potential to reduce computational costs or the number of in-depth experiments. This purpose of this study was to compare the performance of widely used artificial neural network (ANN) and random forest (RF) model for predicting the fuel consumption and engine-out emissions of a calibrated spark ignition (SI) engine for any given condition. To evaluate the performance of machine models established in this work, engine performance of ~2000 steady-state conditions were collected by a validated one-dimensional (1D) computational fluid dynamics (CFD) model with various spark timings, engine speeds and loads. In detail, a randomly selected 80% of dataset were used to train and the remaining 20% were used to validate the proposed machine learning models. A subset from the model predictions formed the test dataset to assess the model performance from a combustion viewpoint in addition to merely statistical indexes. The results indicated that both algorithms can serve as the tool to assist engine combustion analysis. Moreover, the ANN model performed better than RF, as evidenced by the lower root-mean-square error (RMSE) and larger coefficient of determination (R²) regardless of the dataset. Further, the ANN algorithm better characterized the relationship between the engine control variables and the engine fuel consumption rate or engine-out emissions. This was probably due to the fact that engine combustion related responses such as efficiency and emissions were better described by multiple interacted mathematical functions. Accordingly, it is recommended to use ANN model to forecast engine combustion related responses at least for the spark ignition engines.