Combustion Control of Diesel Engine using Feedback Error Learning with Kernel Online Learning Approach

13th International Conference on Motion and Vibration Control; 12th International Conference on Recent Advances in Structural Dynamics

Widayaka, Elfady Satya^1 ; Ohmori, Hiromitsu^2

Graduates School of Science and Technology, Keio University, Japan^1

Department of System Design Engineering, Keio University, 3-14-1, Hiyoshi, Kohoku-ku, Yokohama

223-8522, Japan^2

关键词: Adaptive controllers;    Combustion control;    Engine modeling;    Feedback error learning;    Multi-variable models;    Online learning;    Speed of convergence;    Tracking performance;   

Others  :  https://iopscience.iop.org/article/10.1088/1742-6596/744/1/012107/pdf
DOI  :  10.1088/1742-6596/744/1/012107

来源: IOP

【 摘 要 】

This paper shows how to design Multivariable Model Reference Adaptive Control System (MRACS) for "Tokyo University discrete-time engine model" proposed by Yasuda et al (2014). This controller configuration has the structure of "Feedback error learning (FEL)" and adaptive law is based on kernel method. Simulation results indicate that "kernelized" adaptive controllers can improve the tracking performance, the speed of convergence and the robustness to disturbances.