南京工业大学学报（自然科学版）2025，Vol.47Issue(6)：633-642,10.DOI:10.3969/j.issn.1671-7627.2025.06.002
基于光谱与机器视觉的氢气火焰探测技术研究进展
Research progress on hydrogen flame detection technology based on spectroscopy and machine vision
摘要
Abstract
With the development of hydrogen energy technologies and their applications,hydrogen flame detection has become increasingly important in ensuring the safety of the hydrogen energy industry.Traditional photosensitive flame detectors have limited sensitivity to the low-radiation flames of hydrogen,which often leads to high false alarm rates and delayed responses in complex industrial environments.This paper reviews recent studies on the spectral characteristics of hydrogen flames,focusing on the origins of characteristic peaks in the ultraviolet,visible,and infrared bands.It also examines the classification and technical limitations of photosensitive flame detectors and highlights the recent progress in the application and optimization of machine vision-based detectors.The accuracy and robustness of machine vision-based detectors in hydrogen flame detection are improved by integrating infrared and visible images and introducing deep learning models.In the future,developing specialized detectors with multi-band composite sensing technologies that leverage the unique spectral characteristics of hydrogen flames is expected to greatly enhance early fire warning capabilities and provide reliable technical support for the safe development of the hydrogen energy industry.关键词
氢气火焰/火焰光谱/图像处理/火焰探测/计算机视觉/深度学习Key words
hydrogen flame/flame spectrum/image processing/flame detection/computer vision/deep learning分类
资源环境引用本文复制引用
王纪元,李怡,潘旭海,汪志雷,华敏,蒋军成..基于光谱与机器视觉的氢气火焰探测技术研究进展[J].南京工业大学学报（自然科学版）,2025,47(6):633-642,10.基金项目
国家重点研发计划政府间国际科技创新合作项目(2023YFE0199100) （2023YFE0199100）
