Related Experiment Video
Updated: Jun 27, 2026
06:45
Design and Application of a Fault Detection Method Based on Adaptive Filters and Rotational Speed Estimation for an Electro-Hydrostatic Actuator
Published on: October 28, 2022
A High-Fidelity Multi-Model Benchmark Dataset for General Aviation Anomaly Detection Generated via Physics-Based
Lu Jing1,2, Yali Fang3, ZiYi Huang1
1Civil Aviation Flight University of China, Guanghan, 618307, China.
Scientific Data
|June 25, 2026
Summary
A new dataset addresses the lack of realistic flight data for general aviation safety monitoring. It includes synthetic anomalies in real flight data from two aircraft types, aiding robust anomaly detection algorithm development.
Area of Science:
- Aviation Safety
- Data Science
- Machine Learning
Background:
- General aviation (GA) safety monitoring lacks high-quality, anomaly-rich flight data.
- Existing datasets often lack physical realism or are limited to specific aircraft models.
Purpose of the Study:
- To create a comprehensive benchmark dataset for data-driven safety monitoring in GA.
- To enable the development of robust, model-agnostic anomaly detection algorithms.
Main Methods:
- Collected data from 120 real flight sorties (approx. 1 million data points) from Cessna 172 and Cirrus SR20 aircraft.
- Developed a physics-based synthetic injection framework to generate four anomaly types: throttle surge, flight path deviation, cylinder misfire, and pitch excursion.
- Structured the dataset hierarchically by anomaly type and aircraft model, providing paired normal and abnormal samples.
Main Results:
- The dataset comprises real flight data augmented with realistic synthetic anomalies.
- Statistical analysis and baseline anomaly detection benchmarks validate the dataset's utility.
- The dataset supports direct counterfactual analysis for anomaly detection model training and evaluation.
Conclusions:
- This benchmark dataset bridges the gap between simulated and real-world data for GA safety.
- It facilitates the development of advanced, reliable safety monitoring systems for heterogeneous aircraft fleets.
- The resource is crucial for advancing data-driven approaches to aviation safety.
