Addressing Gearbox Health Monitoring Challenges for Helicopters: A Machine Learning Approach
Guilherme Moreira1, Alexandre Pereira2, Airton Nabarrete3
1Instituto Tecnológico de Aeronáutica (ITA), Programa de Pós-graduação em Aplicações, Pça. Mal. Eduardo Gomes, 50, 12228-970 São José dos Campos, SP, Brazil.
Anais Da Academia Brasileira De Ciencias
|December 19, 2024
Summary
Military helicopter gearboxes can shed metal particles due to wear. This study uses machine learning on flight data to predict particle detachment, enhancing safety and maintenance for the H225M fleet.
Area of Science:
- Aviation Engineering
- Mechanical Engineering
- Data Science
Background:
- Military helicopter transmissions, like the H225M, face significant dynamic loads.
- Wear and fatigue in gearboxes cause ferromagnetic particle detachment, posing safety risks and increasing maintenance needs.
Purpose of the Study:
- To apply machine learning algorithms for predicting ferromagnetic particle detachment in helicopter transmissions.
- To enhance operational safety and maintenance efficiency for the Brazilian H225M fleet.
Main Methods:
- Utilizing data from the Flight Data Recorder (FDR) and Health and Usage Monitoring System (HUMS).
- Implementing pre-processing techniques for effective machine learning model application.
- Developing predictive models to forecast particle detachment events.
Main Results:
- Demonstrated the feasibility of using FDR and HUMS data for predicting particle detachment.
- Identified key indicators and patterns associated with gearbox wear and particle release.
- Provided a foundation for proactive maintenance strategies.
Conclusions:
- Machine learning offers a viable approach to predict gearbox particle detachment in military helicopters.
- Proactive prediction can significantly improve aviation safety and reduce operational costs for the H225M fleet.
- Data pre-processing is crucial for successful implementation of machine learning in aviation maintenance.
