Avionic Main Fuel Pump Simulation and Fault-Diagnosis Benchmark

TL;DR

This paper introduces a high-fidelity simulation benchmark for aircraft main fuel pump systems, including annotated data and baseline anomaly detection methods, to address data scarcity in critical cyber-physical systems.

Contribution

It presents a physics-informed co-simulation model with annotated fault data and demonstrates baseline anomaly detection techniques for aircraft fuel pumps.

Findings

Generated time-series data with health and fault annotations.

Unsupervised RNN-VAE effectively detects anomalies.

SOM-VAE successfully classifies operating modes.

Abstract

In many cyber-physical systems, especially in critical applications such as aeroplanes, data to train anomaly detection and diagnosis algorithms is lacking due to data protection issues and partial observability. To combat this inherent lack of data, we introduce a high-fidelity, physics-informed co-simulation of a common aircraft main-fuel-pump system modelled in \textsc{MATLAB/Simulink Simscape Fluids}. We also describe its generated time-series data with health and fault mode annotations. To show feasibility of our benchmark, we apply an unsupervised Recurrent Variational Autoencoder (RNN-VAE) for anomaly detection and a SOM-VAE for operating mode discretization, trained to separate healthy and faulty conditions.