Online Automatic Adaptation for Model-based Control of Diesel Engine

In this paper, an online automatic adaptation method for model-based control of a diesel engine is developed. Control-oriented models based on physics has been proposed as substitutes for conventional control methods to improve the performance of engine under real driving situation. Even such physical-rich models have fitting parameters and it is preferable to adapt the parameters according to the real-time operating condition. Therefore, an automatic adaptation method for the model is developed, and the method is based on neural network. The prediction accuracy of the model is evaluated by simulation and it is confirmed that the method can be applied online to a real engine by experiment.