An Edge AI System Based on FPGA Platform for Railway Fault Detection

TL;DR

The paper tackles the need for reliable, real-time railway fault detection by presenting an FPGA-based edge AI system that uses a lightweight CNN to inspect track images. It demonstrates a practical hardware-software co-design where CNN inference runs on the FPGA (with loop tiling and mixed-precision quantization) and results are wirelessly reported to a GUI. The system achieves $88.9\%$ accuracy and demonstrates superior energy efficiency ($3.41$ GOPS/W) compared with GPU ($2.44$ GOPS/W) and CPU ($0.73$ GOPS/W), corresponding to $1.39\times$ and $4.67\times$ improvements, respectively. This work highlights the viability of real-time, low-power railway inspection at the edge using FPGA platforms, enabling safer and more scalable rail infrastructure monitoring.

Abstract

As the demands for railway transportation safety increase, traditional methods of rail track inspection no longer meet the needs of modern railway systems. To address the issues of automation and efficiency in rail fault detection, this study introduces a railway inspection system based on Field Programmable Gate Array (FPGA). This edge AI system collects track images via cameras and uses Convolutional Neural Networks (CNN) to perform real-time detection of track defects and automatically reports fault information. The innovation of this system lies in its high level of automation and detection efficiency. The neural network approach employed by this system achieves a detection accuracy of 88.9%, significantly enhancing the reliability and efficiency of detection. Experimental results demonstrate that this FPGA-based system is 1.39* and 4.67* better in energy efficiency than peer implementation on the GPU and CPU platform, respectively.

Figures (3)

• Figure 1: The edge AI system for railway fault detection
• Figure 2: Sample input images to the model
• Figure 3: The lightweight neural network architecture