Deep Learning-Based Detection of Electrical Faults and Power Quality Disturbances in Aerospace Power Systems

The transition toward More Electric Aircraft (MEA) has introduced highly complex electrical architectures that impose strict requirements on reliability, safety, and realtime operation. Yet most existing research on power quality disturbances (PQDs) and electrical fault diagnosis targets conventional utility-scale power grids and relies on low-frequency analysis, which limits accuracy and applicability in aircraft electrical systems that operate at higher frequencies. This paper addresses this gap by presenting a deep learning-based framework for automated multiclass detection and classification of electrical faults and PQDs in aircraft electrical systems, with an emphasis on classification metrics, robustness, and applicability under aerospace constraints. A high-fidelity aircraft power system model inspired by the Boeing 787 electrical architecture was developed to represent operation at a 400Hz fundamental frequency, producing high-resolution signals under various fault and PQD conditions. Two datasets were generated; the first includes one-dimensional time-domain signals, enhanced through signal processing techniques and generative adversarial networks (GANs) to boost data diversity and robustness, made publicly available on IEEE DataPort. The second dataset comprises two-dimensional time-frequency representations derived from the short-time Fourier transform. Several deep learning architectures were evaluated, identifying a compact ResNet architecture as particularly effective.