On 9th December 2020, Kazuki Hamada(M1), Department of Mechanical Engineering, received the 63rd Japan Automatic Control Conference Best Presentation Award.
<About awarded research>
Presentation title: “Influence of Feature Selection Method on Prediction Performance of Machine Learning Model for Engine Combustion”
One of the unexplained phenomena of engine combustion is the cycle-by-cycle variations. It is considered that various factors interact, but the mechanism has not been clarified. In order to elucidate this mechanism, we used machine learning which is attracting attention in engineering field. In this study, we focus on the feature selection part and selected features using three feature selection methods. Furthermore, we investigated the prediction performance of the neural network when the selected features were used as inputs.
<Your impression & future plan>
It is thought that electric vehicles will become widespread in the future. Hybrid vehicles, which are equipped with an engine, is one of them. Therefore, it is important to improve the efficiency of the engine in future. For that purpose, it is necessary to elucidate the mechanism of combustion variation. We will analyze the variation using the feature selection method and work on constructing a control system that suppresses the variation.