Statistical Modeling of Engine Combustion System with Output Uncertainty Evaluation and Its Application to Control Input Design

Abstract

In this paper, we propose a modeling method for an engine combustion system by applying the approximated Gaussian process regression. We model not only the output behavior of the combustion system, but also the uncertainty evaluation for the output estimation. The experimental results show that the proposed method achieves almost the same level of modeling accuracy, while more efficiently reducing the computational cost than previous methods. Finally, we applied the constructed models and their uncertainty evaluations to control input design. For given desired outputs, we find the corresponding inputs of the engine system using the models and their uncertainty evaluations.