A Study on the Control of Cycle-to-cycle Combustion Variations in a Gasoline Engine Using Machine Learning

Combustion variation is widely known as a factor that prevents engines from achieving high efficiency. In this study, a model to predict IMEP per cycle is constructed by machine learning. Furthermore, we propose a control method for cycle-to-cycle combustion variation using the model. The effectiveness and performance of the proposed method are experimentally validated on a spark-ignited gasoline engine test bench. From the experimental results, IMEP per cycle was not successfully controlled. This may be due to the low prediction accuracy of the model and the use of what is considered to be the highest efficiency for comparison.