| Titre : | Real-time monitoring and fault diagnosis of quadcoter using multi-domain vibration analysis and machine learning |
| Auteurs : | Mohamed Seif El Islam Lalem, Auteur ; M'hamed Ouadah, Directeur de thèse ; Omar Touhami, Directeur de thèse |
| Type de document : | document électronique |
| Editeur : | [S.l.] : [s.n.], 2026 |
| Format : | 1 fichier PDF (6.4 Mo) / ill. |
| Note générale : |
Mode d'accès : accès au texte intégral par intranet.
Thèse de Doctorat : Électrotechnique : Alger, Ecole Nationale Polytechnique : 2026 Bibliogr. p. 133 - 144 |
| Langues : | Anglais |
| Index. décimale : | D001726 |
| Tags : | Quadcopter UAV Autonomous flight control Integral saturation backstepping control GWO Propeller fault detection Vibration analysis Machine learning IoT Predictive maintenance DNN Real-time health monitorin |
| Résumé : | The growth of autonomous unmanned aerial vehicles (UAVs) in civilian sectors has intensified, propelled by technology advancements and favorable legislation. This transformation presents significant issues in flight control and predictive maintenance due to highly nonlinear six degrees of freedom dynamics, under-actuated properties, and the danger of propeller failures. This thesis improves autonomous quadcopter systems through the development of integrated control optimization and intelligent health monitoring frameworks. The study harmonizes operational safety with autonomous mission demands, suggesting resilient options for improved UAV reliability. The study presents a thorough technique utilizing rigorous 6-DOF dynamic modeling, Integral Saturation Backstepping Control (ISBSC) design with Lyapunov- based stability proofs, and the Grey Wolf Optimizer (GWO), which demonstrates enhanced convergence compared to Genetic Algorithms (GA) and Particle Swarm Optimization (PSO) (50-100 vs. 135-500 fitness values). Notable progress in predictive maintenance employs ESP32-ADXL335 IoT-enabled hardware for real-time vibration collection, multi-domain feature extraction encompassing time, frequency, and time-frequency attributes, alongside refined machine learning classifiers. The SVM-GA-RMS combination attains 98.8% accuracy with a 40% reduction in false alarms, whilst the proprietary DNN achieves flawless 100% accuracy in quantifying five-class fault severity (0%-40% blade damage). This integrated system merges optimal nonlinear control with data-driven diagnostics, laying the groundwork for next-generation autonomous aerial aircraft in safety-critical civilian applications. |
Exemplaires (1)
| Code-barres | Cote | Support | Localisation | Section | Disponibilité | Spécialité | Etat_Exemplaire | ||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| T000499 | D001726 | Ressources électroniques | Bibliothèque centrale | Thèse de Doctorat | Disponible | Electrotechnique | En Traitement |
