Rotorcrafts operate HUMS for fault diagnosis, but the capabilities of HUMS analysts are important due to the complex mechanical configuration of rotorcraft and the difficulty of analysis. Accordingly, an advanced algorithm is needed to identify faults, and flight data can be used to develop the algorithm. However, flight data contains abnormal or missing data due to problems such as sensor defects, which can be a major problem in algorithm development. Therefore, in this paper, we study a gap filling method using MLP to obtain normal data when there are problems such as abnormal or missing data in rotorcraft flight data. Existing methods typically fill the gap through interpolation or curve fitting using the before and after values of missing data. However, in this study, normal values were estimated using the values of other parameters at the moment of missing or abnormal data occurred. As a result of evaluating the performance of the trained MLP using normal data, it was confirmed that the output of the model well estimates the changing trend of the target. And when abnormal data was applied to the trained model, it was confirmed that point outliers were eliminated and appropriate values were found in the case of missing data.
Unexpected failures of avionics equipment critically affect flight safety, operational availability, and maintenance costs. To address these issues, Condition-Based Maintenance Plus (CBM+) has emerged as a strategy to optimize maintenance timing based on equipment condition rather than fixed schedules. However, while aviation research has largely focused on engines and structures, studies on avionics systems remain limited, often relying on simulations. This study proposes a novel data-driven approach to predict avionics equipment failures using actual aircraft operational data. Maneuver-related sequences were analyzed to investigate correlations between flight patterns and equipment faults, and a two-stage framework was developed. In the feature extraction stage, a CNN-LSTM encoder compresses 10 s maneuver sequences into compact yet informative representations. In the fault prediction stage, AI models classify failures of the Fire Control Radar based on these features. Experiments with real flight data validated the effectiveness of the method, showing that the CNN-LSTM encoder preserved essential maneuver information, while the combination of Standard Scaling and Multi-Layer Perceptron achieved the best performance, with a maximum Fault Recall of 98%. These findings demonstrate the feasibility of practical CBM+ implementation for avionics equipment using only flight data, providing a promising solution to improve maintenance efficiency and aviation safety.
31 August 2025 in The Journal of Korean Institute of Information Technology
상태기반정비는 시스템의 잔존유효수명을 예측하는 고장예지 기술을 바탕으로 최적의 유지보수 시점을 결정하는 비용 효율적인 전략으로, 항공 분야를 위시한 다양한 산업에 적용되고 있다. 이를 위해서는 베어링의 진동신호, 배터리의 전압 및 전류 신호 등 시스템의 건전성을 나타내는 모니터링 데이터가 필수적이다. 그러나 항전시스템에서는 모니터링되는 만여 개의 변수 중 무엇이 고장과 관련이 있는지조차 알려져 있지 않다. 본 논문에서는 대한민국 공군에서 제공받은 실제 비행 데이터를 활용하여 항전장비의 고장 관련 변수를 추출하는 방법을 제안한다. 데이터의 복잡도를 낮추기 위해 오토인코더를 이용해 차원을 축소한 후, 랜덤 포레스트를 통해 고장 관련 변수를 추출하였다. 결과로써 유의미한 후보 변수 30개를 추출하였으며, 이 중 다섯 개 변수는 다양한 조건에서도 일관되게 높은 중요도를 보여 신뢰성 있는 고장 관련 변수로서의 가능성을 확인하였다.