What question did this study set out to answer?

The goal is to enhance UAV flight stability by integrating machine learning with real-time weather data.

March 26, 2026

IA2UCS: An Intelligent Atmospheric‐Adaptive UAV Control System Using Machine Learning and Real‐Time Weather Integration for Enhanced Flight Stability

Key Points

The goal is to enhance UAV flight stability by integrating machine learning with real-time weather data.
Developed an Intelligent Atmospheric‐Adaptive UAV Control System (IA2UCS) using three modules: Robust Adaptive Control Module, Predictive Atmospheric Intelligence System, and Flight Safety Assessment Protocol.
Applied sliding mode control with deep reinforcement learning to address uncertainties in flight dynamics.
Employed ensemble learning algorithms like Random Forest and LSTM networks for weather prediction.
Utilized fuzzy logic for risk evaluation on flight conditions.
Demonstrated high resilience to actuator faults with rapid detection in less than 0.5 seconds.
Achieved stable recovery within 2 seconds post-fault, indicating strong control under adverse conditions.

Abstract

ABSTRACT Conventionally, Unmanned Aerial Vehicle (UAV) control systems are not able to handle atmospheric uncertainty such as wind turbulence, precipitation‐induced aerodynamic changes, and temperature changes, which degrade flight stability and mission reliability. Intending to introduce an Intelligence Atmospheric‐Adaptive UAV Control System (IA2UCS), this research uses Machine Learning (ML) algorithms with atmospheric data to improve the UAV performance in dynamic environmental conditions. These limitations are overcome in the proposed IA2UCS framework by the following three interconnected modules: (1) Robust Adaptive Control Module (RACM) that uses sliding mode control with Deep Reinforcement Learning (DRL) adaptation to deal with lumped uncertainties, (2) Predictive Atmospheric Intelligence System (PAIS) that uses ensemble learning algorithms, such as Random Forest, Long Short‐Term Memory (LSTM) networks, and Physics‐Informed Neural Networks (PINNs) to predict the weather, and (3) Flight Safety Assessment Protocol (FSAP) that uses fuzzy logic‐based risk evaluation to determine the flight conditions. The approach is to include real‐time sensor fusion, historical meteorological data analysis and adaptive parameter tuning to maximize flight performance. The proposed system demonstrates high resilience to actuator faults, rapid fault detection under 0.5 s, and stable recovery within 2 s, making it suitable for real‐world deployment in complex atmospheric environments.

Mark Helpful

Bookmark

Relay