基于物联网和频谱分析的林火早期探测技术  

Research on early forest fire detection based on the internet of things and spectrum analysis

作　　者：张朔 李雨桐 高德民[1] 管志浩 ZHANG Shuo;LI Yutong;GAO Demin;GUAN Zhihao(College of Information Science and Technology,Nanjing Forestry University,Nanjing 210037,China)

机构地区：[1]南京林业大学信息科学技术学院,南京210037

出　　处：《林业工程学报》2021年第3期149-153,共5页Journal of Forestry Engineering

基　　金：中国博士后特别资助基金(2018T110505);中国博士后基金(2017M611828)。

摘　　要：森林火灾是一种具有突发性和危险性的自然灾害,根据火情种类可将森林火灾分为:地下火、地表火和树冠火,其中树冠火极其危险,因为它的传播速度是地表火的十几倍,因此,尽早对火灾类型进行分类是非常重要的。笔者设计了一种基于物联网技术和声音频谱分析技术的林火探测系统,此系统适用于森林火灾连续监测和早期探测的无线声学测深,具有良好的空间分辨率和时间分辨率。为实时监测林区,将大量传感器节点部署在林区,实现林区全覆盖。根据无线传感器的有效通讯距离以及能耗问题,本研究选用具有低功耗、远距离传输特点的LoRa设备,进行数据传输。最后,对声音传感器采集到的音频数据进行声音频谱分析,结合分类算法和交叉谱区分树冠火和地表火的声音频率差异。但是,分类方法的准确性会受到一些因素的影响,如传感器的分布、声音传输中的能量损耗、数据传输的延迟等。通过实验测试表明树冠火的声音频率为0~400 Hz,地表火的声音频率为0~15000 Hz,并可以通过声音频谱分析技术实现从声音角度来探测森林火灾的发生。Forest fire is one of the most dangerous and unexpected natural disasters.Based on the differences between burning substances,forest fires can be roughly divided into three types,namely,underground fire,surface fire,and crown fire.Among them,crown fire is the extremely dangerous one as it can spread about more than ten times faster than surface fire,so it is vital to detect the forest fire as early as possible.Unfortunately there are a large number of obstacles in the forest areas commonly,which would reduce the efficiency of forest fire detection system based on spectral monitoring,resulting in the forest fire detection system cannot monitor the occurrence of forest fires in time.In this study,we design a wireless acoustic sounding system based on the internet of things technology and sound spectrum analysis technology for detecting the occurrence of the forest fires and classifying the types of forest fires,which has good spatial resolution and time resolution and is suitable for continuous monitoring and early detection of forest fires.In order to monitor the forest areas and forest fires in real-time,we have to deploy a large number of sensor nodes in the forest area to achieve full coverage of the forest area.According to the effective communication distance and energy consumption of the wireless sensors,in our research,we select the LoRa equipment with low power consumption and long-distance transmission characteristics to carry out data transmission.Finally,we analyze the frequency spectrum of the audio data collected by the sound sensors.Depending on the classification algorithm and the cross-spectrum method,we can distinguish the sound frequency difference between the crown fires and the surface fires.However,the accuracy of the classification method is affected by some factors,such as the distribution of wireless sensor nodes,the loss of energy in sound transmission,and the delay of data transmission.The experimental results show that the sound frequency of crown fire is about 0-400 Hz,while that of sur

关 键 词：森林火灾 地表火 树冠火 物联网 声谱分析 LoRa 

分 类 号：S762[农业科学—森林保护学]