|
| 基于无线感知的非视距火灾探测方法研究 |
| The Research on Non-Line-of-Sight Fire Detection Method Based on Wireless Sensing |
| 投稿时间：2025-03-04 修订日期：2025-11-05 |
| 查看/发表评论 下载PDF阅读器 |
| DOI： |
| 基金项目:台州市科学技术局（No. 24nya19）；上海市“科技创新行动计划”社会发展科技攻关项目（22dz1200902） |
|
| 中文关键词: 无线感知 火灾检测 接收信号强度 |
| 英文关键词:wireless sensing fire detection received signal strength |
| 摘要点击次数: 90 |
| 全文下载次数: 0 |
| 中文摘要: |
| 室内火灾对生命财产安全构成严重威胁，而烟雾传感器易受灰尘影响、视频检测方法受光照和视场限制，均难以在非视距场景下及时发现火情。本文提出一种基于 Wi-Fi 信道状态信息（Channel State Information，CSI）的非视距火灾检测方法：基于分析火焰及其燃烧颗粒物对无线信号的扰动特性，提出面向CSI数据的卡尔曼滤波与子载波选择算法，构建以支持向量机（Support Vector Machine，SVM）为核心的轻量级火灾检测系统。实验表明，该方法在由20厘米砖混墙体分割的非视距火灾检测场景中，检测准确率可达 96.83 % ，性能优于现有同类方案。相关数据集已公开在（https://github.com/T-bjq/WCFD-CSI-dataset），以支持后续研究。 |
| 英文摘要: |
| Indoor fires pose serious threats to life and property; however, smoke detectors often suffer from dust accumulation and vision-based methods remain constrained by lighting conditions and limited fields of view, rendering timely detection in non-line-of-sight (NLoS) scenarios challenging. An NLoS fire detection method based on Wi-Fi channel state information (CSI) is presented in this paper. By analyzing perturbations induced by flames and combustion particulates on wireless signals, a Kalman filter coupled with subcarrier selection is devised to preprocess CSI data, and a support vector machine (SVM) classifier is employed to realize lightweight fire detection. Experiments conducted in NLoS environments separated by 20 cm brick-concrete walls demonstrate a detection accuracy of 96.83%, surpassing that of existing schemes. Relevant data sets have been published in (https://github.com/T-bjq/WCFD-CSI-dataset) to support further research. |
| 关闭 |
