Volume 52 Issue 3
Mar.  2023
Turn off MathJax
Article Contents

Xu Wenjing, Xian Jinhong, Sun Dongsong. Polarization lidar system for smoke and dust monitoring and experimental research[J]. Infrared and Laser Engineering, 2023, 52(3): 20220508. doi: 10.3788/IRLA20220508
Citation: Xu Wenjing, Xian Jinhong, Sun Dongsong. Polarization lidar system for smoke and dust monitoring and experimental research[J]. Infrared and Laser Engineering, 2023, 52(3): 20220508. doi: 10.3788/IRLA20220508

Polarization lidar system for smoke and dust monitoring and experimental research

doi: 10.3788/IRLA20220508
Funds:  National Natural Science Foundation of China (41774193)
  • Received Date: 2022-07-21
  • Rev Recd Date: 2022-10-31
  • Publish Date: 2023-03-25
  •   Objective   The fires in forests, wetlands, grasslands and other natural areas are characterised by their sudden and destructive nature, and it is important to reduce the damage caused by fires through early detection and fighting. Traditional fire monitoring methods such as manual inspections and cameras do not allow for 24/7, wide-area monitoring, and there is a lag in detecting fires. Therefore, the use of lidar with high precision, high resolution, long detection distance and sensitivity to changes in aerosol particle concentration, etc., can play an important role in the field of smoke and fire monitoring, to achieve early detection and early warning of fire. Researchers have made some explorations in this area. However, for lidar detection distance of 2 km or more, the single pulse energy of the laser was on the order of millijoule, and there is a human eye safety risk for outdoor use. Moreover, the researchers have not given an analysis of the measurement and application in a multi-obstacle environment. Therefore, a polarimetric lidar system with a day and night detection, and detection distance of more than 6 km and the single pulse energy of the laser on the order of microjoule is proposed.  Methods   Laser wavelengths adapted to outdoor long-range detection are obtained through simulations. The lidar scanning strategies are designed for different installation scenarios, for flat environments and for environments with many obstacles, respectively. As the lidar measurement areas are at a certain height from the ground, correction for fire point positioning errors is based on a Gaussian plume model. A portable lidar system with polarization channels was built to further validate the simulation results, scanning control strategies and inversion algorithms through field experiments.  Results and Discussions   By simulating the detection distance of lidar with different wavelengths, the results show that the detection distance of lidar with 1 064 nm wavelength is 1.3-1.4 times of 532 nm wavelength (Fig.2). By optimizing the scanning strategy and algorithm (Fig.4), the influence of fixed obstacles and temporary moving obstacles can be eliminated. In order to avoid obstacles of similar height around the installation site, a certain elevation angle is usually set for the lidar, and the horizontal distance deviation and vertical height measurement deviation resulting from the existence of the detection elevation angle are calculated. When the elevation angle of lidar detection is 2°, the measured height deviation at 6 km is 209.397 m. The Gaussian plume model is used to simulate the soot concentration distribution. When the atmospheric stability is B and the average wind speed is 1 m/s, the high value point of soot concentration distribution at 200 m height is ≥1 km from the ground fire point, it provides a correction basis for accurate location of fire point. Outfield measurements by using 1 064 nm polarization lidar in both mountainous and plain environments can quickly and accurately identify fire points, which demonstrate the feasibility of using lidar for smoke and fire monitoring.  Conclusions   A scanning polarization lidar can rapidly identify fire smoke and dust. The field experiments were conducted in Panshan County, Panjin City, Liaoning Province, around Yanghu Scenic Area, and Guanyin Mountain Forest Park, Dongguan City, Guangdong Province, respectively. The polarization lidar was able to identify the smoke and dust quickly under the open area and multi-obstacle mountainous area. Observational data will be accumulated in subsequent experimental tests to verify the optical properties of various types of soot particles and further improve the identification efficiency.
  • [1] Ma Zenan, Gao Zhongliang, Wei Jianheng, et al. Research progress inforest fire smoke monitoring equipment [J]. Forestry Machinery & Woodworking Equipment, 2022, 50(2): 4-9. (in Chinese) doi:  10.3969/j.issn.2095-2953.2022.02.001
    [2] Wang Dong, Jia Zhicheng, Xia Yuhang, et al. Research progress and trend in forestry and grassland fires monitoring technology [J]. World Forestry Research, 2021, 34(2): 26-32. (in Chinese) doi:  10.13348/j.cnki.sjlyyj.2021.0006.y
    [3] Huang Kehui, Zhu Jing, Huang Yiping, et al. Fire automatic detection system based on Doppler weather radar [J]. Meteorological Monthly, 2013, 39(2): 241-248. (in Chinese) doi:  10.7519/j.issn.1000-0526.2013.02.014
    [4] 黄克慧, 朱景, 周必高, 等. 多普勒天气雷达自动探测森林火灾系统介绍[C]//第27届中国气象学会年会雷达技术开发与应用分会场, 2010: 595-599.
    [5] Sassen K. Advances in polarization diversity lidar for cloud remote sensing [J]. Proceedings of the IEEE, 1994, 82(12): 1907-1914. doi:  10.1109/5.338078
    [6] Utkin A B, Fernandes A, Simões F, et al. Feasibility of forest-fire smoke detection using lidar [J]. International Journal of Wildland Fire, 2003, 12(2): 159-166. doi:  10.1071/WF02048
    [7] Leroy-cancellieri V, Augustin P, Filippi J B, et al. Evaluation of wildland fire smoke plume dynamics and aerosol load using UV scanning lidar and fire-atmosphere modelling during the Mediterranean Letia 2010 experiment [J]. Natural Hazards and Earth System Science, 2014, 14(3): 509-523. doi:  10.5194/nhess-14-509-2014
    [8] Heese B, Hofer J, Baars H, et al. Wild fire aerosol optical properties measured by lidar at Haifa, Israel [C]//The European Physical Journal Conferences, 2018, 176: 05049.
    [9] Qiao L. Study on polarized light scattering characteristics of fire smoke particles[D]. Hefei: University of Science and Tech-nology of China, 2008. (in Chinese)
    [10] Zhang Q. Study on the characteristics of light scattering for fire smoke and non-fire particles[D]. Hefei: Hefei University of Technology, 2010. (in Chinese)
    [11] Xu Chidong, Ji Yufeng. A case study of particulate pollution case measured by polarization micro-pulse lidar [J]. Journal of Atmospheric and Environmental Optics, 2011, 6(1): 27-32. (in Chinese) doi:  http://gk.hfcas.ac.cn/CN/
    [12] Uchino O, Tabata I, Kai K J, et al. Polarization properties of middle and high level clouds observed by lidar [J]. Journal of the Meteorological Society of Japan Ser II, 1988, 66(4): 607-616. doi:  10.2151/JMSJ1965.66.4_607
    [13] Megie G. Laser remote sensing: fundamentals and applications [J]. EOS Transactions American Geophysical Union, 1985, 66(40): 686. doi:  10.1029/EO066i040p00686-05
    [14] Song Yuehui, Hua Dengxin, Li Shichun, et al. Detection of bottom aerosols in urban area using micro-pulse polarization lidar [J]. Acta Photonica Sinica, 2012, 41(10): 1140-1144. (in Chinese) doi:  10.3788/gzxb20124110.1140
  • 加载中
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Figures(11)  / Tables(3)

Article Metrics

Article views(101) PDF downloads(42) Cited by()

Related
Proportional views

Polarization lidar system for smoke and dust monitoring and experimental research

doi: 10.3788/IRLA20220508
  • 1. School of Earth and Space Sciences, University of Science and Technology of China, Hefei 230026, China
  • 2. School of Atmospheric Sciences, Nanjing University, Nanjing 210023, China
Fund Project:  National Natural Science Foundation of China (41774193)

Abstract:   Objective   The fires in forests, wetlands, grasslands and other natural areas are characterised by their sudden and destructive nature, and it is important to reduce the damage caused by fires through early detection and fighting. Traditional fire monitoring methods such as manual inspections and cameras do not allow for 24/7, wide-area monitoring, and there is a lag in detecting fires. Therefore, the use of lidar with high precision, high resolution, long detection distance and sensitivity to changes in aerosol particle concentration, etc., can play an important role in the field of smoke and fire monitoring, to achieve early detection and early warning of fire. Researchers have made some explorations in this area. However, for lidar detection distance of 2 km or more, the single pulse energy of the laser was on the order of millijoule, and there is a human eye safety risk for outdoor use. Moreover, the researchers have not given an analysis of the measurement and application in a multi-obstacle environment. Therefore, a polarimetric lidar system with a day and night detection, and detection distance of more than 6 km and the single pulse energy of the laser on the order of microjoule is proposed.  Methods   Laser wavelengths adapted to outdoor long-range detection are obtained through simulations. The lidar scanning strategies are designed for different installation scenarios, for flat environments and for environments with many obstacles, respectively. As the lidar measurement areas are at a certain height from the ground, correction for fire point positioning errors is based on a Gaussian plume model. A portable lidar system with polarization channels was built to further validate the simulation results, scanning control strategies and inversion algorithms through field experiments.  Results and Discussions   By simulating the detection distance of lidar with different wavelengths, the results show that the detection distance of lidar with 1 064 nm wavelength is 1.3-1.4 times of 532 nm wavelength (Fig.2). By optimizing the scanning strategy and algorithm (Fig.4), the influence of fixed obstacles and temporary moving obstacles can be eliminated. In order to avoid obstacles of similar height around the installation site, a certain elevation angle is usually set for the lidar, and the horizontal distance deviation and vertical height measurement deviation resulting from the existence of the detection elevation angle are calculated. When the elevation angle of lidar detection is 2°, the measured height deviation at 6 km is 209.397 m. The Gaussian plume model is used to simulate the soot concentration distribution. When the atmospheric stability is B and the average wind speed is 1 m/s, the high value point of soot concentration distribution at 200 m height is ≥1 km from the ground fire point, it provides a correction basis for accurate location of fire point. Outfield measurements by using 1 064 nm polarization lidar in both mountainous and plain environments can quickly and accurately identify fire points, which demonstrate the feasibility of using lidar for smoke and fire monitoring.  Conclusions   A scanning polarization lidar can rapidly identify fire smoke and dust. The field experiments were conducted in Panshan County, Panjin City, Liaoning Province, around Yanghu Scenic Area, and Guanyin Mountain Forest Park, Dongguan City, Guangdong Province, respectively. The polarization lidar was able to identify the smoke and dust quickly under the open area and multi-obstacle mountainous area. Observational data will be accumulated in subsequent experimental tests to verify the optical properties of various types of soot particles and further improve the identification efficiency.

    • 森林、湿地、草原等自然界火灾,具有突发性强、破坏性大等特点,特别是重特大火灾的发生,往往伴随着恶劣、极端的天气条件,扑救极为困难。因此,一旦发生火警,必须以极快的速度采取扑救措施。扑救是否及时、决策是否得当,大都取决于对林火行为的发现是否及时、分析是否准确、决策是否得当。扑救重特大火灾,对于世界各国都是十分棘手的难题,为最大限度减少火灾损失与危害,世界各国都把工作重点放在对初期火灾的扑救和预警上。

      目前,用于森林、湿地、草原等自然界火灾监测的方法主要有地面巡护、瞭望台巡护、航空巡护、视频监控、卫星遥感和雷达遥感等。其中地面巡护和瞭望台巡护主要依靠人工,存在巡护面积小、火点确认不及时,外出巡查受天气影响大等问题。航空巡护利用巡护飞机进行林火探测,其优点为巡护视野宽、机动性大、速度快;但在天气条件较差时难以起飞,同时巡视受航线、时间的限制,无法24 h连续监测。利用摄像头拍摄视频对林区进行监控,可直观获得观察区域的状态,技术较为成熟;存在的不足为监控范围有限,需大量布点,受太阳光、气象条件影响较大,且基于热成像原理的摄像系统易产生误报。卫星遥感技术探测范围广、搜集数据快、能得到连续性资料,反映火的动态变化;存在的不足为准确率低、需要地面二次核实,且其分辨率不高,当热点达到3个像素时,火已基本成灾 [1-2];根据研究,多普勒天气雷达也能探测到由于森林大火、意外爆炸等引起的漂浮在空中的反射物回波,从而实现对于森林大火的监测和跟踪;但误判率较高,发现时间存在滞后[3-4]

      近年来,随着激光雷达遥感技术的快速发展,其在各行业的应用越来越广泛[5],因其具有精度高、分辨率高,且对气溶胶粒子浓度变化敏感等特点,可在烟火监控领域发挥重要作用,实现火灾的早期发现及预警。2003年,Andrei B.Utkin等人对利用激光雷达实现林火探测的可行性进行了探索分析,使用单脉冲能量20 mJ的532 nm激光雷达,实现2.5 km半径范围内火点的成功测量[6] 。2014年,V.Leroy Cance-llieri等人将单脉冲能量16 mJ的355 nm激光雷达烟尘实测结果与仿真模型的数据进行了对比,具有较好的一致性 [7]。2018年,Birgit Heese等人对利用532 nm激光雷达探测获得的以色列一次火灾烟尘的光学特性作出分析[8] 。在以上研究中,能够实现2 km以上烟尘探测半径的激光雷达,单脉冲能量均为毫焦级,户外扫描使用存在人眼安全风险,且研究人员未对多障碍物环境下的测量及应用给出分析。因此,文中提出了一种昼夜探测距离6 km以上,单脉冲微焦级的偏振激光雷达系统。通过对探测波长的优化选择,扫描及算法策略的设定,火点定位误差的分析,使其更加适用于烟尘监测。经外场点火实验证明了激光雷达实现火点快速检测的可行性,为推进产品化火灾监测预警预报激光雷达系统奠定了基础。

    • 当激光雷达系统发射的激光在大气中传输时,大气中的各种组分会对激光产生吸收和散射效应。通常以$ a=2\pi r/\lambda $r为粒子半径,λ为入射波长,作为判别标准,将散射过程分为瑞利散射和米散射。当$ a\ll \lambda $时,为瑞利散射;当$ a\approx \lambda $时,为米散射。由于烟尘粒子的尺度一般$ \gg 0.01\; {\text{μ} }\mathrm{m} $,即激光与之作用主要为米散射过程。

      散射过程的另一个特点是偏振状态的变化,偏振的程度和状态取决于粒子的大小、形状,入射光的偏振态及观测散射光的角度。线偏振激光入射球形粒子,其后向散射信号仍为与入射激光偏振方向相同的线偏振光,如果入射的是非球形粒子,则后向散射信号除了有与入射光偏振方向相同的信号外,还有其正交分量,此为粒子的退偏振效应[9-10]

      大气中气溶胶粒子一般为球形粒子,后向散射光不产生退偏。而沙尘、冰云、灰霾、烟尘等多为非球形粒子,使用激光雷达进行大气参量测量时退偏比升高。因此,通过反演获得消光系数及退偏振比,并结合多条件算法判别,可完成准确的火情预警。

    • 当激光雷达系统发射的激光在大气中传输时,大气中的各种组分会对激光产生吸收和散射,从而改变后向散射光的能量、光谱特性、偏振状态等信息。激光雷达的测量过程可用激光雷达方程描述[11-13]

      式中:$ {P}_{p}\left(r\right) $$ {P}_{s}\left(r\right) $分别表示偏振激光雷达接收到的距离r处的大气后向散射信号功率的平行分量和垂直分量;P0为激光器发射功率;$ \; {\beta }_{p}\left(r\right) $$ \;{\beta }_{s}\left(r\right) $分别为后向散射系数平行分量和垂直分量;$ {\sigma }_{p}\left(r\right) $$ {\sigma }_{s}\left(r\right) $分别为消光系数平行分量和垂直分量;$ {k}_{p} $$ {k}_{s} $分别为平行分量探测通道和水平分量探测通道的系统常数。

      退偏比$ \delta \left(r\right) $定义为:

      经增益比及串扰标定后,取$K=\dfrac{{k}_{p}}{{k}_{s}}$,公式(3)可简化为:

      根据激光雷达方程,由Fernald反演算法求解出消光系数。消光系数的大小可反映气溶胶、烟尘及云雾粒子的浓度变化,但消光系数无法区分粒子形态,需通过退偏比进一步完成判别。沙尘、冰云等非球形粒子退偏比较大,海洋及大陆气溶胶粒子通常为球形,退偏比很小。其中沙尘的退偏比约为0.2~0.3,强烈沙尘暴可达0.4,冰云的退偏比约为0.4~0.5,灰霾的退偏比约为0.1~0.2[14]。自然界火灾引起的烟尘,由于树木燃烧过程中产生水汽,其混合物更接近球形粒子,其退偏比一般<0.1。

      同时,火灾烟尘早期过程的浓度分布具有区域集中,定点出现,随风向延展等特点,可据此与其他造成气溶胶浓度升高的现象进行区分,降低误判率。

      结合消光系数及退偏比,利用火点判别算法,扣除掉背景谱扫描中获得的固定障碍物,即可实现火灾烟尘的实时监测。

    • 偏振激光雷达具有两个偏振通道,在实现气溶胶后向散射系数、消光系数测量的同时,还可获得退偏比信息。系统结构示意图如图1所示。

      Figure 1.  Structural schematic of polarization lidar

      激光器发射线偏振光,经扩束镜扩束后入射大气,通过望远镜系统接收大气回波信号,经滤光片抑制背景后,通过偏振分束器件将回波信号中的平行分量、垂直分量分别耦合入光电探测器件,将光信号转化为电信号后由采集控制系统进行采集存储,并进行分析反演,获取大气粒子状态及分布信息。偏振激光雷达系统架设在扫描云台上,可实现对关注区域水平及垂直方向的快速扫描测量。

      表1为偏振激光雷达性能指标,系统选取了微脉冲、高重频激光源,保障人眼安全的同时,提升信噪比。光电探测器为单光子计数型APD,对弱信号具有较高响应,从而实现最大探测距离在10 km以上的远距离测量。

      ParametersValues
      LaserNd:YAG
      Wavelength/nm1064
      Single pulse energy/μJ≥100
      Pulse width/ns≤10
      Polarization stateHorizontal polarization
      Frequency/kHz1-10
      Time resolution/s5
      Spatial resolution/m15
      Detection range/km≥10 @Nighttime
      ≥7 @Daytime
      Scanning angle/(°)0-360@ Horizontal
      −5-90@ Vertical

      Table 1.  Main technical parameters of the polarization lidar

    • 在森林、湿地、草原等区域进行火灾监测时,由于需覆盖的面积较大,从激光器成本和稳定性,对野外环境和生物的影响,烟尘浓度监测灵敏度以及有效探测距离等几个方面考虑,选择激光雷达系统最优波长。

      激光雷达常用波长为355、532、1064、1550 nm。前三个波长由Nd:YAG 固体激光器产生,其中1064 nm为基频,器件价格和系统稳定性最优。出射1550 nm激光的为光纤激光器,虽然更易于雷达系统装调,但价格较高,出于性价比考虑,在系统中未选择该波长。

      其次,激光雷达应用于野外环境,需考虑激光波长对环境及生物的影响,可见光会对昆虫、鸟类以及野生动物产生惊扰,并产生光污染,因此选用不可见光更为适宜。

      激光雷达在进行烟尘浓度监测时,利用的是弹性散射原理。弹性散射包括米散射和瑞利散射。当粒子尺度与激光波长可比拟时发生米散射,当粒子尺度远小于激光波长时发生瑞利散射。瑞利散射的强度与波长的四次方成反比,因此,当探测目标物为分子时,应尽量选择短波。而烟尘的粒子尺度较大,激光雷达探测主要为米散射过程,可考虑较长波长,减少瑞利散射信号的干扰。

      最后,因为森林、湿地、草原等环境较为辽阔,应在保障安全的前提下,尽量满足远距离探测的要求,从而实现用较少的激光雷达布设量完成较大范围的探测。以下对532 nm和1064 nm激光波长的雷达系统进行了仿真,仿真过程所使用的硬件参数见表2

      ParametersValue
      Wavelength/nm5321064
      Single pulse energy/μJ150
      Polarization stateHorizontal
      Frequency/kHz10
      Diameter of telescope/mm100
      Integral time/s10
      Spatial resolution/m15
      Detector quantum efficiency40%3%
      Received field angle/mrad0.4
      Filter bandwidth/nm1

      Table 2.  Simulation parameters of lidar detection performance

      相比于1064 nm波长,532 nm波长太阳背景光辐射更强,大气透过率却更低,图2分别模拟了白天和夜间、不同能见度条件下,采用两种波长的微焦级激光雷达水平通道的探测距离。1064 nm波长在5 km能见度条件下,可实现全天时水平6 km以上探测距离。

      Figure 2.  Simulation of 532 nm and 1064 nm polarization lidar detection range under different visibility conditions. (a) Daytime; (b) Night-time

      综上所述,采用1064 nm波长激光,更加适用于自然环境下火灾烟尘的探测。

    • 进行自然界火灾烟尘测量,其现场安装环境具有多样性,草原及湿地监测,视野较为开阔,但用于山林火灾监测的激光雷达,扫描视场易受山体、树木的遮挡。图3给出了不同环境的扫描视场示意图。

      Figure 3.  Diagram of lidar scanning in different environments. (a) Moun-tain, many obstacles; (b) Grassland, wetlands, fewer obstacles

      火灾的早期发现及预警,时效性非常重要,而雷达扫描测量需要一定的周期性,如何快速锁定火点位置非常重要。因此,需对雷达的扫描策略进行优化设置。

      图4所示,当雷达安装完毕并开启后,如进行障碍物较多的山林地区测量时,设置多仰角PPI扫描,如进行草原或湿地等视野较为开阔地点的测量时,设置单仰角PPI扫描。共扫描10组数据,作为背景谱保存,并结合GIS地理信息确定扫描所覆盖的具体区域。该步骤主要用于后续固定障碍物扣除,以及扫描区域自动选取。激光照射到火灾烟尘粒子后,会有强回波,需根据信号退偏比剔除云雾、气溶胶等引起的信号增大。如无异常信号,则偏振激光雷达进入日常扫描模式,单方向积分30 s,扫描角度间隔2°。如发现异常信号,为防止误判,需对两个周期以上的信号进行核验。如未同时出现异常,则进入日常扫描模式,如均出现异常,则进入火点追踪模式。在异常点周边30°角范围内进行快速扫描测量,单方向积分5 s,扫描角度间隔0.5°~1°,结合GIS地理信息给出火点位置并做告警处理。

      Figure 4.  Flowchart of scanning and data processing strategy

    • 受激光雷达安装位置影响,为最大程度实现大覆盖范围测量,需要对激光雷达探测仰角进行调整,图5为不同仰角下,高度及距离偏差随探测距离的变化情况。即有仰角时,距离激光雷达安装点位越远的距离处,实际探测高度与激光雷达安装水平面存在一定的高度偏差。当仰角为2°时,6000 m处的实际探测高度为209.397 m。当仰角为5°时,6000 m处的实际探测高度为522.934 m。同时,雷达获取的距离值为斜程距离,转化为水平方向距离时也存在距离偏差,在数据反演及火点定位时需予以校正。

      Figure 5.  Variation of height deviation and range deviation with detection range at different elevation angles

      虽然火点发生在地面,但烟尘会随气流扩散,尤其在风速较大的情况下,其上升到空中的浓度高值点与地面火点存在位置偏差,为进一步分析误差,采用高斯烟羽模型对不同风速、不同高度层烟羽状态进行仿真。

      高斯烟羽模型公式如下:

      式中:C为烟尘浓度,单位kg/m3Q为源强,单位kg/s;u为火点高度的平均风速,单位m/s;$\sigma_y 、 \sigma_{{z}}$分别为浓度标准偏差表示的y轴和z轴上的扩散参数;H为火点的有效高度,单位m。

      国家标准扩散参数见表3。选择点源扩散模型,分别对大气稳定度为B,平均风速为1 m/s,大气稳定度C,平均风速5 m/s的情况进行了仿真,如图6所示,(a)~(e)为大气稳定度B,平均风速1 m/s时,0、50、100、150、200 m高度层烟尘浓度的分布结果。可见随着高度的升高,烟尘浓度的高值点与实际火点的偏移距离逐渐增大,200 m高度时,偏移距离≥1 km。当大气稳定度为C,平均风速5 m/s时,(f)~(j)对应给出了0、50、100、150、200 m高度层的烟尘浓度分布,其浓度高值点距实际火点的偏移距离更大,200 m高度时,偏移距离≥1.5 km。同时,随着高度的增加,烟尘浓度高值逐渐下降,且平均风速为1 m/s时,各高度层的烟尘浓度值高于平均风速为5 m/s时的浓度值。即激光雷达工作在高仰角,高平均风速天气条件下时,对于远距离火点烟尘的探测,定位难度升高,需结合风速风向、大气稳定度,测量获得的回波信号高值点距离,对火点实际位置进行修正。由于实测过程中风速风向在动态变化,无法将测量地风场实时数据导入软件,因此,该部分的偏差校正需通过后期数据处理来完成。即将激光雷达扫描结果进行网格化处理,确定烟尘浓度高值点距雷达安装点距离后,根据激光雷达探测仰角、安装高度,获得该点位实际测量高度。根据高斯烟羽模型中该高度层烟尘浓度高值点距实际火点距离值,按风向的反方向画入网格图,获得校正后的火点位置。

      Atmospheric
      stability
      $ {\sigma _y} $${\sigma _{{z} } }$
      A0.22x/(1+0.0001x)0.50.2x
      B0.16x/(1+0.0001x)0.50.12x
      C0.11x/(1+0.0001x)0.50.08x/(1+0.0002x)0.5
      D0.08x/(1+0.0001x)0.50.06x/(1+0.0002x)0.5
      E0.06x/(1+0.0001x)0.50.03x/(1+0.0003x)
      F0.04x/(1+0.0001x)0.50.016x/(1+0.0003x)

      Table 3.  Table of national standard diffusion para-meters

      Figure 6.  Gaussian plume simulation results. (a)-(e) Atmospheric stability B, average wind speed of 1 m/s, concentration distribution map at 0 m, 50 m, 100 m, 150 m, and 200 m; (f)-(j) Atmospheric stability C, average wind speed of 5 m/s, smoke concentration distribution diagram at 0 m, 50 m, 100 m, 150 m, and 200 m

    • 为进一步验证激光雷达完成烟尘监测的可行性,通过人工点火方式在野外进行了实验。实验场地1为辽宁省盘锦市盘山县绕阳湖景区,位置坐标为(41.320253°N,122.053803°E)。激光雷达安装位置及现场照片如图7所示。该景区为湿地芦苇荡公园,视野开阔障碍物少,基本无遮挡,激光雷达安装高度距地面约为2 m。点火所用燃料为周边购置的稻草。提前清理周边场地易燃物,防止因人工点火引发起火区域扩大。点火实验结束后,彻底灭火,打扫残留物,避免余火复燃。

      Figure 7.  Lidar installation position and site photos

      此次实验在11:40~12:30期间,共进行了两次点火,激光雷达扫描角度为331°~337°,扫描角度间隔1°。图8为实验期间的激光雷达扫描伪彩图,显示半径4 km。11:43,距离激光雷达3 km,334°方位角处出现消光及退偏比高值点,即该位置处检测到了高浓度烟尘。在11:43~11:55期间,该高值点逐渐向东北方向移动。11:59的扫描图中未探测到消光及退偏比高值点,此时第一次点火的火点已熄灭,烟尘逐渐消失。重新添加稻草后,在同一位置进行第二次点火。12:06~12:21再次在雷达扫描图中识别到消光及退偏比高值点。

      Figure 8.  Lidar scanning pseudo-color images

      该日天气晴好,风向为西南风,风速2 m/s,与测量获得的烟尘扩散方向相一致。激光雷达扫描仰角0°,激光出射口距离地平面垂直高度2 m,根据高斯烟羽模型的仿真结果,烟尘浓度集中点距离地面实际火点约10 m,由于激光雷达的距离分辨率为15 m,处于同一距离门内,3 km处的火点定位未产生较大误差。但后续采集卡采样率提高,将距离分辨率优化至3.75 m后,应结合模型的偏移数据,获得火点的准确定位。

      实验场地2为广东省东莞市观音山森林公园,位置坐标为(22.8929945°N, 114.0724724°E)。激光雷达安装位置及现场照片如图9所示。由于扫描范围内地势起伏较大,扫描过程中激光束会照射到山体,体现在信号上为障碍物强回波。因此,根据扫描策略,首先进行背景谱扫描获得山体障碍信息,并在后续观测中进行扣除。

      Figure 9.  Lidar installation position and site photos

      图10为雷达PPI扫描距离校准信号分布伪彩图,1~1.2 km处因山体造成障碍物遮挡,产生强回波信号,在图中标注为蓝色。为防止误判,完成10次扫描测量。记录障碍物强回波信号的角度及距离信息,在后续火点反演中予以剔除。图11(a)左侧所示为雷达退偏比分布伪彩图,1.2 km附近有退偏比高值点,图11(b)为该测量径向上的退偏比廓线图。图11(a)左下角照片为现场实验测试的烟筒。由于烟筒产生的烟尘颗粒相比于林草燃烧产生的烟尘颗粒,水汽含量小,退偏振比值相对较大,约为0.12。单廓线积分时间5 s,实现了火点的快速识别。

      Figure 10.  Pseudo-color map of range calibration signal obtained from background spectrum scanning measurement

      Figure 11.  Fire point recognition. (a) Pseudo-color images of depolarization ratio obtained by scanning measurements; (b) Profile of depolarization ratio obtained by single radial measurement

    • 文中对不同能见度下532 nm和1064 nm激光雷达的有效探测距离进行仿真,确定了用于烟尘监测的偏振激光雷达波长优选1064 nm,可在能见度大于5 km的晴好天气下,实现全天候6 km以上的远距离测量。根据应用场景的不同,对雷达扫描及算法策略进行了设置,通过初始背景谱测量,剔除掉固定障碍物的影响,实现激光雷达自动扫描及数据反演。对激光雷达探测仰角引起的距离偏差进行计算,并用高斯烟羽模型仿真了不同高度层烟尘浓度高值点与实际火点的位置偏差,为火点的准确定位提供修正依据。

      通过在湿地公园及山地公园的点火实验,两种场景下,激光雷达均可快速测量到火点烟尘,并有效剔除障碍物引起的误判,验证了利用偏振激光雷达测量火灾烟尘的可行性。在后续的实验测试中,会积累观测数据,验证各种类型烟尘粒子的光学特性,进一步提高识别效率。

Reference (14)

Catalog

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return