Aircraft Fault Prediction Training Data Balancing With Synthetic Flight Series
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Flight sensor data for aircraft components is extremely unbalanced, with a high prevalence of nominal flights and rare fault events, making it challenging for machine learning models to accurately predict component faults.
Innovation Solution
A method is employed to generate a balanced training dataset by correlating flight sensor data with component fault data, using a remaining useful life (RUL) function to assign labels, and generating synthesized flight series through copying and modifying original flight series to balance the dataset, thereby improving the learning capability of machine learning models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If flight sensor data is used directly for training machine learning models, then the dataset reflects real operational conditions, but the dataset is extremely unbalanced with rare fault events making accurate fault prediction challenging
Solution Approach 1:
The patent creates synthesized flight series by copying and modifying existing flight data. Specifically, it generates synthetic fault instances by copying nominal flight data and introducing simulated fault conditions, thereby increasing the quantity of fault events in the training dataset without requiring additional real-world fault occurrences.
Solution Approach 2:
The patent transforms the distribution of fault events by applying parameter changes to the dataset. It uses techniques such as resampling, data augmentation, and synthetic data generation to alter the frequency and distribution of fault events, converting the extremely unbalanced dataset into a more balanced training dataset that enables accurate fault prediction.
2Reliability
If synthesized flight series are generated by copying and modifying original flight series, then the dataset balance is improved, but the complexity of data processing increases
Solution Approach 1:
The patent generates synthesized flight series by copying existing flight data records and applying modifications. This approach balances the dataset by creating additional fault instances from nominal data, improving reliability while using systematic copying procedures that manage processing complexity.
Solution Approach 2:
The patent applies parameter changes to transform nominal flight data into synthetic fault instances. By systematically modifying parameters such as sensor readings, fault indicators, and temporal characteristics, it achieves dataset balance through controlled transformations rather than complex manual data generation.
Data Source
AI summary
The present disclosure provides a method of generating a balanced training dataset for a machine learning model in one aspect, the method including: receiving flight sensor data corresponding to a plurality of flights, and applying one or more criteria to the flight sensor data to generate a training dataset including a plurality of first instances corresponding to flights of the plurality of flights. The method further includes assigning, using component fault data, respective labels to the plurality of first instances, and generating, for groups of one or more labels of the respective labels, a respective plurality of flight series. Each flight series includes a respective sequence of second instances that is based on some of the plurality of first instances, and that concludes with a second instance that is assigned a label included in the group.