Model-Reference-Based Intelligent Active Disturbance Rejection Attitude Control for Morphing Aircraft

Attitude stabilization of hypersonic morphing vehicles is challenging because geometric reconfiguration continuously alters aerodynamic characteristics, intensifies cross-axis interactions, and changes available stability margins. Parameterizing these effects across the morphing envelope is impractical, limiting conventional model-based adaptive control. This paper presents a learning-augmented model reference adaptive control scheme, TD3-MRAC, that compensates for reconfiguration-induced uncertainties without explicit aerodynamic representation. A fixed-geometry reference model defines the desired closed-loop response, while a backstepping controller provides nominal stabilization. To use only low-dimensional output measurements, a linear extended state observer reconstructs the system states and estimates the aggregate influence of unmodeled dynamics and external disturbances. Instead of a prescribed analytical adaptation law, a Twin Delayed Deep Deterministic Policy Gradient agent generates bounded incremental control commands to reduce mismatch between the morphing vehicle and reference dynamics. Stability is established independently of reinforcement-learning convergence: the reference closed loop is globally exponentially stable, the observer error dynamics are input-to-state stable with an adjustable estimation bound, and the complete closed-loop system is uniformly ultimately bounded with an ultimate bound explicitly characterized by the design parameters. Numerical evaluations under combined aerodynamic, parametric, and external perturbations show improved tracking over a conventional robust adaptive controller, especially in roll and yaw, where integral absolute error is reduced by up to 46\%, while pitch performance remains comparable. The trained policy remains effective on unseen perturbation combinations without additional training. These results indicate that TD3-MRAC offers a theoretically supported and practically implementable approach for morphing-flight control while substantially reducing dependence on labor-intensive aerodynamic uncertainty modeling.

Authors

Institutions

Publication Details

Journal
Research Square
Published
2026-11-18
DOI
https://doi.org/10.21203/rs.3.rs-11549312/v1
Primary Topic
Adaptive Control of Nonlinear Systems
Type
preprint

Funders

Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Model-Reference-Based Intelligent Active Disturbance Rejection Attitude Control for Morphing Aircraft

Fanyi Meng, Haipeng Chen, Gang Chen, Yanru Huang
4 links
Research Square
Adaptive Control of Nonlinear Systems
preprint

Model-Reference-Based Intelligent Active Disturbance Rejection Attitude Control for Morphing Aircraft

Fanyi Meng, Haipeng Chen, Gang Chen, Yanru Huang
preprint en

Abstract

Attitude stabilization of hypersonic morphing vehicles is challenging because geometric reconfiguration continuously alters aerodynamic characteristics, intensifies cross-axis interactions, and changes available stability margins. Parameterizing these effects across the morphing envelope is impractical, limiting conventional model-based adaptive control. This paper presents a learning-augmented model reference adaptive control scheme, TD3-MRAC, that compensates for reconfiguration-induced uncertainties without explicit aerodynamic representation. A fixed-geometry reference model defines the desired closed-loop response, while a backstepping controller provides nominal stabilization. To use only low-dimensional output measurements, a linear extended state observer reconstructs the system states and estimates the aggregate influence of unmodeled dynamics and external disturbances. Instead of a prescribed analytical adaptation law, a Twin Delayed Deep Deterministic Policy Gradient agent generates bounded incremental control commands to reduce mismatch between the morphing vehicle and reference dynamics. Stability is established independently of reinforcement-learning convergence: the reference closed loop is globally exponentially stable, the observer error dynamics are input-to-state stable with an adjustable estimation bound, and the complete closed-loop system is uniformly ultimately bounded with an ultimate bound explicitly characterized by the design parameters. Numerical evaluations under combined aerodynamic, parametric, and external perturbations show improved tracking over a conventional robust adaptive controller, especially in roll and yaw, where integral absolute error is reduced by up to 46\%, while pitch performance remains comparable. The trained policy remains effective on unseen perturbation combinations without additional training. These results indicate that TD3-MRAC offers a theoretically supported and practically implementable approach for morphing-flight control while substantially reducing dependence on labor-intensive aerodynamic uncertainty modeling.

Research Square
China Academy of Launch Vehicle Technology (CN), Xi'an Jiaotong University (CN)
National Natural Science Foundation of China
Adaptive Control of Nonlinear Systems
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon