Title | ||
---|---|---|
Online Learning Of Automotive Gasoline Engine Model Using Robust Recursive Gaussian Process |
Abstract | ||
---|---|---|
The dynamical model structure based on the Nonlinear Auto-Regressive with eXogenous inputs (NARX) model is effective as a control model of automotive engine systems. In this paper, we propose an identification method of the NARX model using the Gaussian processes (GP). The first main result is the proposal of a robust online model updating method against outliers. In this approach, we apply the concept of robust Kalman filter to the GP-NARX model. The second main result is to propose a new dynamical Design of Experiment (DoE) algorithm with GP. We propose an input design method for acquiring training data of GP-NARX model. The effectiveness of the modeling strategy is demonstrated by an engine benchmark problem provided by the joint research committee of Society of Automotive Engineers and Society of Instrument (JSAE) and Control Engineers (SICE). |
Year | Venue | Field |
---|---|---|
2018 | 2018 ANNUAL AMERICAN CONTROL CONFERENCE (ACC) | Data modeling,Nonlinear autoregressive exogenous model,Computer science,Automotive engine,Kalman filter,Robustness (computer science),Control engineering,Gaussian process,Online model,Automotive industry |
DocType | ISSN | Citations |
Conference | 0743-1619 | 0 |
PageRank | References | Authors |
0.34 | 0 | 2 |
Name | Order | Citations | PageRank |
---|---|---|---|
Hiroyuki Oyama | 1 | 0 | 2.03 |
Masaki Yamakita | 2 | 266 | 57.24 |