Online Gaussian Process-Based Model Predictive Attitude Control for Underwater Gliders

In this paper, an online Gaussian process(GP)-based model predictive control(MPC) approach is proposed to solve the attitude control of underwater gliders(UGs) in the presence of model uncertainties. A GP model is trained online using measurement data to compensate for uncertainties of UGs including...

Full description

Saved in:
Bibliographic Details
Published in2023 42nd Chinese Control Conference (CCC) pp. 2771 - 2775
Main Authors Guo, Linyu, Min, Boxu, Gao, Jian, Song, Yunxuan, Chen, Yimin, Pan, Guang
Format Conference Proceeding
LanguageEnglish
Published Technical Committee on Control Theory, Chinese Association of Automation 24.07.2023
Subjects
Attitude control
Computational modeling
Data models
Model Predictive Control
Online Gaussian Process
Predictive models
Simulation
Training
Uncertainty
Underwater Glider
Online AccessGet full text

Cover

More Information
Summary:In this paper, an online Gaussian process(GP)-based model predictive control(MPC) approach is proposed to solve the attitude control of underwater gliders(UGs) in the presence of model uncertainties. A GP model is trained online using measurement data to compensate for uncertainties of UGs including external disturbances and inner model errors. In the process of training the GP model, a genetic algorithm is used to optimize hyperparameters to minimize the difference between the model and real system. Meanwhile, a small dictionary of 500 data is designed to reduce computational burden. Simulation results show that compared with standard MPC, the proposed GP-MPC controller has better transient and steady-state performances for a UG's attitude control.
ISSN:2161-2927
DOI:10.23919/CCC58697.2023.10240134