English static mirror for SEO/GEO · AI-assisted translation · Read Chinese original

Deep Learning-Based Detection of Electrical Faults and Power Quality Disturbances in Aerospace Power Systems (arXiv:2609.10479)

Forum topic · 小凯 · 2026-09-11

Summary

This paper presents a hardware-aware deep learning framework for multiclass detection of electrical faults and power quality disturbances in 400 Hz aerospace power systems, addressing the gap left by methods designed for conventional 50/60 Hz grids. Using a high-fidelity simulation model inspired by the Boeing 787 electrical architecture, the authors generate voltage and current waveforms covering 21 normal, disturbance, switching, open-circuit, and short-circuit conditions, forming two datasets of 73,500 samples each (1D time-series and STFT time-frequency representations). Signal augmentation, domain randomization, and class-specific GANs increase waveform diversity, and the time-series dataset is released via IEEE DataPort. Benchmarking 1D/2D CNNs, LSTMs, CNN-LSTM hybrids, ResNet, MobileNet, and VGG, a compact ResNet achieves the best accuracy-complexity tradeoff with 96.94% test accuracy using 175,685 parameters. After 8-bit quantization and deployment on a Xilinx Zynq UltraScale+ MPSoC ZCU102, it retains 95.87% accuracy with a mean accelerator latency of 6.90 ms, demonstrating edge-AI feasibility for aircraft electrical health monitoring.

Research area: Machine Learning Authors: Ian C. Guzmán, Radu Babiceanu, Berker Peköz Published: 2026-09-09 arXiv: 2609.10479

Abstract

More Electric Aircraft require fast and reliable monitoring of high-frequency electrical networks, yet most power quality disturbance and fault diagnosis methods are developed for conventional 50 or 60 Hz grids. This work presents a hardware-aware deep learning framework for multiclass detection of electrical faults and power quality disturbances in a 400 Hz aerospace power system. A high-fidelity simulation model inspired by the Boeing 787 electrical architecture generates voltage and current waveforms for 21 normal, disturbance, switching, open-circuit, and short-circuit conditions. Two datasets, each containing 73,500 samples, are formed from one-dimensional time-series signals and short-time Fourier transform time-frequency representations. Signal-processing augmentation, domain randomization, and class-specific generative adversarial networks increase waveform diversity, and the time-series dataset is released through IEEE DataPort. We compare 1D and 2D convolutional neural networks, long short-term memory networks, CNN-LSTM hybrids, ResNet, MobileNet, and VGG models under common training conditions. A compact ResNet provides the best accuracy-complexity tradeoff, achieving 96.94 percent software test accuracy with 175,685 parameters. After 8-bit quantization and deployment on a Xilinx Zynq UltraScale Plus MPSoC ZCU102, the model achieves 95.87 percent accuracy and a measured mean neural-network accelerator latency of 6.90 ms per input record. The results establish simulation-based, accelerator-level feasibility for embedded edge AI in aircraft electrical health monitoring and motivate future end-to-end data acquisition and experimental validation.

Key Points

  • Targets 400 Hz aerospace power systems rather than conventional 50/60 Hz grids
  • Simulation model inspired by the Boeing 787 electrical architecture covers 21 fault/disturbance classes
  • Two 73,500-sample datasets: 1D time-series and STFT time-frequency representations; time-series data released on IEEE DataPort
  • Augmentation via signal processing, domain randomization, and class-specific GANs
  • Compact ResNet wins the benchmark: 96.94% accuracy with 175,685 parameters
  • 8-bit quantized deployment on Xilinx ZCU102 MPSoC: 95.87% accuracy, 6.90 ms mean accelerator latency

Tags

#deep-learning#aerospace#power-quality#fault-detection#edge-ai#fpga#resnet#more-electric-aircraft

This page is an English static mirror generated for search and AI citation. It may be a full translation or structured summary of the Chinese original. Canonical interactive discussion lives on the Chinese page: https://zhichai.net/topic/178634731