调谐激光吸收光谱波长偏移修正算法研究

引用本文

唐七星, 张玉钧, 陈东, 张恺, 何莹, 尤坤, 刘国华, 鲁一冰, 范博强, 余冬琪. 调谐激光吸收光谱波长偏移修正算法研究[J]. 光谱学与光谱分析, 2018,38(11): 3328-3333.
TANG Qi-xing, ZHANG Yu-jun, CHEN Dong, ZHANG Kai, HE Ying, YOU Kun, LIU Guo-hua, LU Yi-bing, FAN Bo-qiang, YU Dong-qi. Research on Wavelength Shift Correction Algorithm for Tunable Laser Absorption Spectrum[J]. Spectroscopy and Spectral Analysis, 2018,38(11): 3328-3333.
Doi:10.3964/j.issn.1000-0593(2018)11-3328-06 复制到剪切板

Permissions

《光谱学与光谱分析》期刊社所有

调谐激光吸收光谱波长偏移修正算法研究

唐七星^1,², 张玉钧^1,^*, 陈东^1,³, 张恺^1,², 何莹¹, 尤坤¹, 刘国华^1,², 鲁一冰^1,², 范博强^1,², 余冬琪^1,²

1. 中国科学院环境光学与技术重点实验室, 安徽光学精密机械研究所, 安徽合肥 230031

2. 中国科学技术大学科学岛分院, 安徽合肥 230026

3. 合肥工业大学仪器科学与光电工程学院, 安徽合肥 230009

*通讯联系人 e-mail: yjzhang@aiofm.ac.cn

作者简介: 唐七星, 女, 1989年生, 中国科学技术大学博士研究生 e-mail: qxtang@aiofm.ac.cn

收稿日期: 2017-10-31

基金: 国家重点研发计划(2016YFC0201003), 省科技重大专项(15czz04124)资助

摘要

可调谐半导体激光器具有线宽窄、波长扫描快、室温工作等特点, 基于可调谐半导体激光器构成的激光吸收光谱气体测量系统在大气环境检测、工业生产过程在线检测中得到了广泛的应用。在实际测量系统中, 由于可调谐半导体激光器中心波长受温度等因素的影响发生偏移, 如不进行中心波长校正, 将造成序列光谱数据重叠, 处理后的光谱线型发生展宽, 进而影响后续的光谱线型拟合, 对气体浓度的反演精度产生影响。一般采用参考光谱吸收谱线寻峰方法进行序列光谱数据偏移的对齐, 但光谱数据中的随机噪声、背景噪声、漂移噪声等因素影响峰线波长的精度。为了降低上述因素的影响, 提出一种改进的时域相关光谱修正算法, 首先对光谱信号进行自相关, 在一定程度上提高光谱信号的信噪比, 然后再进行时域互相关处理, 能够准确的计算出激光器波长偏移量, 减少由此造成的光谱线型展宽的影响, 提高了浓度反演精度和测量稳定性。在激光吸收光谱气体浓度检测实验系统中进行了实验验证, 评估结果中, 原始数据标准差为1.482 8, 谱线寻峰方法与时域相关方法修正后数据标准差分别为0.433 9和0.293 6, 改进的时域相关修正方法修正后数据标准差为0.132 5, 改进的时域相关修正方法相关系数均优于0.992, 欧式距离的标准差为1.726 4。系统稳定性评估中改进方法波长漂移修正后标准偏差为0.144 3。

关键词: 激光吸收光谱; 激光器波长偏移; 时域相关; 谱线寻峰

中图分类号:O433.1 文献标志码:A

Research on Wavelength Shift Correction Algorithm for Tunable Laser Absorption Spectrum

TANG Qi-xing^1,², ZHANG Yu-jun^1,^*, CHEN Dong^1,³, ZHANG Kai^1,², HE Ying¹, YOU Kun¹, LIU Guo-hua^1,², LU Yi-bing^1,², FAN Bo-qiang^1,², YU Dong-qi^1,²

1. Key Laboratory of Environmental Optics & Technology, Anhui Institute of Optics and Fine Mechanics, Chinese Academy of Sciences, Hefei 230031, China;

2. Science Island Branch, University of Science and Technology of China, Hefei 230026, China

3. School of Instrument Science and Optoelectronic Engineering, Hefei University of Technology, Hefei 230009, China

Abstract

Tunable semiconductor laser is featured by narrow bandwidth, fast wavelength scanning, and room temperature working temperature, etc. The laser absorption spectroscopy gas measurement system based on tunable diode laser is widely applied in atmospheric environmental monitoring and on-line industrial production process detection. In the actual measurement system, the center wavelength of the tunable semiconductor laser is affected by temperature and other factors. If the center wavelength is not corrected, the spectral data will be overlapped and the processed spectral line will be widened, which will affect the subsequent spectral line fitting and affect the accuracy of gas concentration inversion. Generally, the reference spectrum absorption line peak-finding method is used to align the offset of the spectral data. However, the accuracy of the peak wavelength in the spectral data is affected the random noise, the background and drift noise. In order to reduce the side effects mentioned above, an improved algorithm of the time domain correlation is proposed. Firstly, the autocorrelation of the spectral signal is carried out to improve the spectral signal-to-noise ratio to a certain extent, then the time-domain cross-correlation processing is applied, which can accurately calculate the laser wavelength offset, reduce the influence of spectral broadening, and improve the concentration inversion accuracy and measurement stability. Experiments have been carried out in the detection experimental system of gas concentration based on laser absorption. Experimental results show that the standard deviation of the original data is 1.482 8 while the standard deviation of data correction is 0.433 9 and 0.293 6, respectively. The corrected standard deviation of improved method of time domain correlation correction is 0.132 5 with the correlation coefficient higher than 0.992. The standard deviation of Euclidean distance is 1.726 4. The system stability evaluation indicates that the modified standard deviation of the wavelength drift correction is up to 0.144 3.

Keyword: Laser absorption spectroscopy; Laser wavelength shift; Time domain correlation; Spectrum peak-finding

文章图片

引言

可调谐半导体激光吸收光谱技术(TDLAS)具有无需预处理、选择性强、响应速度快、高灵敏度和高精度等优点^{[1, 2, 3, 4]}, 已经广泛应用在气体检测中。

在气体浓度测量过程中温度变化, 电流扫描信号直流电平漂移, 驱动电流与温度的漂移, 热敏电阻、激光器芯片的非完全耦合, 光路、电路老化等会造成激光器波长偏移^[5]。 Stewart等^[7]提出了基于频率调制技术的光谱吸收光纤气体检测方法, 通过对DFB激光器的中心波长进行调制, 使其对气体的吸收线进行缓慢扫描, 得到与气体浓度相关的高次谐波信号。王贵师等^[7]利用Labview软件设计数字比例一积分一微分(PID)算法反馈控制半导体激光器的电流来提高激光器输出波长稳定性。孙延光等^[8]在文献[7]的基础上采用了电流和温度双重反馈稳频技术的瓦斯浓度测量方案等等。上述研究从激光器波长控制的角度解决激光器波长的偏移, 对于激光器波长偏移修正算法的研究并不多。针对这种问题, 通常是利用谱线寻峰和相关^{[9, 10]}的方法, 实现对光谱偏移的复位和对准, 但在对该方法分析讨论后发现其精度受随机噪声、背景噪声、漂移噪声的影响。为了降低这些因素的影响, 提出改进的基于互相关的波长修正算法, 并对一些结论进行了仿真分析和激光吸收光谱气体浓度检测系统实验, 验证了算法的有效性和适用性。

1 实验系统的基本组成和测量原理

1.1 实验系统的基本组成

实验系统结构图如图1所示, 主要包括激光器及控制部分、光学结构部分、数据处理三部分。根据HITRAN数据库选择CO的近红外2 330 nm单根吸收线实现气体的检测。采用NEL公司中心波长为2 330 nm的DFB激光器作为光源, 通过激光器控制模块改变温度、电流, 从而改变激光器的输出波长, 控制其稳定在2 330.18 nm, 信号发生器产生锯齿波扫描信号叠加在激光控制器上, 使激光波长在扫描周期内通过吸收线。激光器输出的单模光束分成参考光路和探测光路, 10%光强经准直器1准直后通过20 cm的5%的标准气体参考池, 到达InGaAs光电探测器1; 90%光强经准直器2准直后通过20 m的气体多次反射池, 由InGaAs光电探测器2接收。探测信号经过放大滤波, 数据采集与信号处理, 反演得到CO的浓度。

	Figure Option View Download New Window
	图1 实验系统结构图Fig.1 Experimental system construction

1.2 实验系统的测量原理

根据Beer-Lambert定律

$I = I_{0} \exp (- S^{*} ΦPcL) (1)$

式(1)中, I₀ 为入射光强, I为透射光强, S^*为吸收线强, Φ 为归一化线型函数, P为气体压强, c为吸收气体的组分浓度, L为检测系统实际光路长度。由此得到式(2)

$c = \frac{1}{S^{*} ΦPL} (\ln \frac{I_{01}}{I_{1}}) = \frac{A}{S^{*} PL} (2)$

式(2)中, A为积分吸光度。

由此可以利用实验系统的标准吸收信号和所测量的信号进行浓度反演, 则待测浓度表示为

$c = \frac{A}{A_{0}} \frac{L_{0}}{L} c_{0} (3)$

式(3)中c₀为标准气体样品池的浓度值, L₀为标准气体样品池的光路长度, A₀为标准气体样品池的积分吸光度, L为实验系统实际光路长度, A为拟合积分吸光度。

2 波长偏移修正方法理论分析

定义Y(m)为用于拟合的标准吸收谱信号

$Y (m) = S (m) + n (m) (4)$

式(4)中S(m)为“ 纯净信号” , n(m)为随机噪声。

参考光路光谱信号为C(m)

$C (m) = x (m) + f (m) + n' (m) (5)$

式(5)中, x(m)为参考光路光谱信号的“ 纯净信号” , n'(m)为随机噪声, f(m)为光学系统可能产生的漂移噪声, 如图2所示。

	Figure Option View Download New Window
	图2 干涉条纹图Fig.2 Interference fringe diagram

2.1 谱线寻峰方法

假设信号为单峰信号, 见式(6)

$\begin{array}{l} \max (Y (m)) = \max (S (m) + n (m)) (6) \\ \max (C (m)) = \max (x (m) + f (m) + n' (m)) \end{array}$

由于n(m)和n'(m)的影响, 即使f(m)为0时, Y(m)与C(m)的峰值吸收波长坐标也不一定相等。

2.2 时域相关的方法

假设Y(m)和C(m)的时域连续信号分别为Y(t)和C(t)。延迟时间为τ , 根据式(7)

$R_{YC} (τ) = \lim_{T \to \infty} \frac{1}{T} \overset{T}{\int_{0}} Y (t) C (t - τ) d t (7)$

对应的离散信号, 可表示为

R_YC(Δ )= $\frac{1}{m} \overset{m}{\sum_{i = 1}}$ Y(i)C(i-Δ )= $\frac{1}{m} \overset{m}{\sum_{i = 1}}$ [S(i)+n(i)][x(i-Δ )+f(i-Δ )+n'(i-Δ )]=R_Sx(Δ )+R_Sf(Δ )+R_Sn'(Δ )+R_nx(Δ )+R_nf(Δ )+R_nn'(Δ )=R_Sx(Δ )+R_nn'(Δ )(8)

式(8)中Δ 为偏移量。

由于信号与噪声并不相关^[10], 即相关值为0, 所以R_Sf(Δ )=R_Sn'(Δ )=R_nx(Δ )=R_nf(Δ )=0。假设随机噪声均值为0, 则随着m的增大, R_nn'(Δ )衰减至0。

2.3 两种方法比较分析

对比上述两种方法, 可以得出以下结论:

(1)在f(m)为0时, 即只受到随机噪声的影响时, 时域相关的方法比谱线寻峰方法更精确。这主要是因为随机噪声的影响, 使得Y(m)与C(m)的峰值吸收波长坐标不一定相等。但时域相关的方法中由于信号与噪声是不相关的, 即提高了信号的抗干扰能力, 能够更准确的计算出激光器的波长偏移量。

(2)在n(m)和n'(m)为0时, 即只受到漂移噪声的影响时。假设f(m)相比x(m)信号强时, 此时不再为单峰信号, 谱线寻峰方法失效。

(3)谱线寻峰方法准确性的影响因素比时域相关的方法复杂。进一步研究发现谱线寻峰方法的准确性受到较多因素的影响, 除了随机噪声的影响外, 还会受到背景噪声、漂移噪声f(m)的频率与幅值等影响。时域相关方法的准确性主要取决于f(m)对x(m)的影响。

2.4 改进的时域互相关光谱修正算法

从上述分析可知, 谱线寻峰方法受到多种因素的影响, 为了提高波长偏移量估计的准确性和适用性, 通过对标准吸收光谱信号和参考信号利用式(9)和式(10)先进行自相关, 提高两路信号的信噪比, 再对两路自相关信号R_YY(Δ ')和R_CC(Δ ″)进行互相关运算, z找出R_RC(Δ )中的最大值, 其所对应的横坐标Δ _j便是激光器的波长偏移量, 从而提高其准确性与灵敏度。利用式(1)再对两路自相关信号R_YY(Δ '), R_CC(Δ ″)进行互相关运算, 得到R_RC(Δ )

R_YY(Δ ')= $\frac{1}{m} \overset{m}{\sum_{i = 1}}$ Y(i)Y(i-Δ ')= $\frac{1}{m} \overset{m}{\sum_{i = 1}}$ [S(i)+n(i)][S(i-Δ ')+n(i-Δ ')]=R_SS(Δ ')+R_Sn(Δ ')+R_nS(Δ ')+R_nn(Δ ')=R_SS(Δ ')(9)

R_CC(Δ ″)= $\frac{1}{m} \overset{m}{\sum_{i = 1}}$ C(i)C(i-Δ ″)= $\frac{1}{m} \overset{m}{\sum_{i = 1}}$ [x(i)+f(i)+n(i)][x(i-Δ ″)+f(i-Δ ″)+n(i-Δ ″)]=R_xx(Δ ″)+R_xf(Δ ″)+R_xn(Δ ″)+R_fx(Δ ″)+R_ff(Δ ″)+R_fn(Δ ″)+R_nx(Δ ″)+R_nf(Δ ″)+R_nn(Δ ″)=R_xx(Δ ″)(10)

$R_{YC} (Δ) = \frac{1}{m} \overset{m}{\sum_{i = 1}} R_{YY} (Δ') R_{CC} (Δ ″) (11)$

此时改进的时域相关光谱修正算法如图3所示。

	Figure Option View Download New Window
	图3 算法改进示意图Fig.3 Schematic diagram of algorithm improvements

2.5 仿真验证

通过仿真来验证以上一些结论的正确。仿真中“ 纯净信号” S(m)按照正态分布, 方差为1得到高斯曲线(如图4), 其峰值坐标为(225, 0.398 9)。同样的方法改变幅值大小, 产生参考光路光谱信号的“ 纯净信号” x(m), 其峰值横坐标同样为225。

	Figure Option View Download New Window
	图4 S(m)的信号图Fig.4 The signal diagram of S(m)

由图4可知, 当激光器波长产生偏移, 则峰值坐标发生改变, 如果信号不进行波长偏移修正, 在进行多次采样数据处理时, 光谱线型会展宽。且由式(3)可知积分吸光度A与浓度c成正比, 可以通过积分吸光度的计算实现浓度的反演。光谱线型的展宽, 从而影响气体浓度的测量精度。

2.5.1 随机噪声的影响

随机噪声对三种方法准确性研究, 仿真中预设的光谱信号Y(m)是在S(m)的基础上加高斯白噪声, 参考光路光谱信号C(m)是在x(m)的基础上加高斯白噪声。其中Y(m)(如图5)为恒定不变的, Y(m)的峰值坐标为(219, 0.409 4)。通过改变C(m)中高斯白噪声的大小, 改变C(m)的信噪比, 比较两种方法在不同信噪比中求取激光器的波长偏移量的能力。其仿真结果如图6所示。

	Figure Option View Download New Window
	图5 Y(m)的信号图Fig.5 The signal diagram of Y(m)

	Figure Option View Download New Window
	图6 随机噪声对三种方法影响的仿真结果Fig.6 The simulation results of the affections of random noise on the three methods

已知激光器的波长偏移量为0。由图6可见, 谱线寻峰方法与时域相关方法受随机噪声的影响较大, 且随着信噪比的减小, 表现的更明显。而改进的时域相关修正方法受随机噪声的影响很小, 在信噪比大于3时, 能够准确的计算出激光器的波长偏移量为0。且在微弱信号时, 也能够准确的计算出激光器的波长偏移量0, 最大的偏移量为-1。谱线寻峰方法与时域相关方法求得波长偏移量的标准差分别为8.306 7和2.580 8, 改进的时域相关修正方法求得波长偏移量的标准差为0.242 7。

2.5.2 漂移噪声的影响

漂移噪声对三种方法准确性研究, 假设没有随机噪声。仿真中预设的光谱信号Y(m), 随机噪声f(m)的影响为0, 此时Y(m)等于S(m)。参考光路光谱信号C(m)在x(m)的基础上加上漂移噪声f(m)。其中Y(m)恒定不变, 通过改变C(m)中漂移噪声的大小和频率, 改变C(m)的信噪比。为了更明显的看到漂移噪声f(m)的影响, 以连续三个周期出现, 即三个不同的C(m), 举例说明仿真图如图7所示。此时三个不同的C(m)的信噪比分别为3.842 5, 7.041 1和3.078 4。

	Figure Option View Download New Window
	图7 C(m)的信号图Fig.7 The signal diagram of C(m)

比较三种方法在不同信噪比中求取激光器的波长偏移量的能力。其仿真结果如图8所示。

	Figure Option View Download New Window
	图8 漂移噪声对三种方法影响的仿真结果Fig.8 The simulation results of the affections of drift noise on the three methods

已知激光器的波长偏移量为0。由图8可见, 三种修正方法在信噪比大于8时, 都能够准确的计算出激光器的波长偏移量为0。但当信噪比小于8时, 三种方法都受影响。谱线寻峰方法与时域相关方法求得波长偏移量的标准差分别为0.554 7和0.615 7, 改进的时域相关修正方法求得波长偏移量的标准差为0.474 6。

3 实验验证

用激光吸收光谱CO浓度检测实验系统测量浓度为50× 10^-6的CO气体。设置锯齿波扫描频率170 Hz, 采样速度为200 kHz的采集卡, 采集不同时刻的12条光谱信号, 如图9所示, 可以看到信号产生了偏移。

	Figure Option View Download New Window
	图9 参考光路光谱Fig.9 The optical spectrum of inner light path

分别对不进行任何波长偏移修正的光谱信号反演CO的浓度与对这三种方法修正激光器波长偏移后的光谱信号反演CO的浓度, 如图10所示。由图10可知, 原始数据标准差为1.482 8, 谱线寻峰方法与时域相关方法修正后数据标准差分别为0.433 9和0.293 6, 改进的时域相关修正方法修正后数据标准差为0.132 5。可以看出利用改进的时域相关修正方法反演出的气体浓度更精确, 距离50× 10^-6偏移量更小, 证明了这种改进方法的有效性。

	Figure Option View Download New Window
	图10 修正后的光谱比较Fig.10 Recovery comparison of spectra

为了更好地区分三种方法的微小差异, 通过对这三种方法消除激光器波长偏移量后的光谱信号与预设的光谱信号进行相关系数和欧式距离^[11]法评估, 如图11所示。并定义正负仅代表与预设光谱信号相比向左或向右偏移。由图11(a)可见, 谱线寻峰方法与时域相关方法的相关系数的平均值分别为0.990 9和0.993 0, 改进的时域相关修正方法的相关系数的平均值为0.996 5。验证了改进的时域相关修正方法消除激光器的波长偏移效果更好, 相关系数均优于0.992。对比三种方法欧氏距离如图11(b)所示, 谱线寻峰方法与时域相关方法欧式距离的标准差分别为4.136 6和3.289 9, 改进的时域相关修正方法欧式距离的标准差为1.726 4。充分说明了改进的时域相关修正方法消除激光器波长偏移的有效性, 能够准确地实现光谱信号波长偏移的复位。保证信号的准确, 实现气体浓度的准确测量。

	Figure Option View Download New Window
	图11 光谱信号的评估图 (a): 相关系数图; (b): 欧式距离图Fig.11 The evaluation diagram of spectral signal (a): The diagram of correlation coefficient; (b): The diagram of euclidean distance

进一步评估改进的时域相关修正方法对浓度反演计算中的稳定性, 对固定浓度为40× 10^-6的CO进行1 h测量。测量过程中采用配分充入气体后, 封闭气体多次反射池的进出口, 可以认为测量过程中CO浓度不变。测量结果改进方法波长漂移修正后如图12所示, 标准偏差为0.144 3, 实验结果表明, 改进的时域相关修正方法提高了测量稳定性。

	Figure Option View Download New Window
	图12 连续测量Fig.12 Successive measurements

4 结论

针对激光吸收光谱检测系统中激光器的波长偏移问题, 搭建了激光吸收光谱气体浓度检测实验系统。研究了两种波长偏移修正方法的特性, 谱线寻峰方法容易受到随机噪声、背景噪声、漂移噪声的影响; 为了减少影响因素, 提出了改进的时域相关光谱修正算法, 能够准确计算出激光器波长偏移量, 提高了测量稳定性, 并通过仿真验证及实验验证。评估结果中, 改进的时域相关修正方法相关系数均优于0.992, 欧式距离的标准差为1.726 4。系统稳定性评估中改进方法波长漂移修正后标准偏差为0.144 3。

The authors have declared that no competing interests exist.

参考文献

文献列表

[1]	Sur R, Sun K, Jeffries J B, et al. Applied Physics B, 2014, 116(1): 33. [本文引用:1]
[2]	Pogány A, Ott O, Werhahn O, et a1. Journal of Quantitative Spectroscopy & Radiative Transfer, 2013, 130: 147. [本文引用:1]
[3]	GAO Yan-wei, ZHANG Yu-jun, CHEN Dong, et al(高彦伟, 张玉钧, 陈东, 等). Acta Optica Sinica(光学学报), 2016, 36(3): 267. [本文引用:1]
[4]	HE Ying, ZHANG Yu-jun, WANG Li-ming, et al(何莹, 张玉钧, 王立明, 等). Chinese Journal of Lasers(中国激光), 2014, 41(1): 227. [本文引用:1]
[5]	LIU Hui-fang, LI Bin, HE Qi-xin, et al(刘慧芳, 李彬, 何启欣, 等). Acta Photonica Sinica(光子学报), 2016, 45(4): 0423004-1. [本文引用:1]
[6]	Zhang Yan, Zhan Ming, Jin Wei. Optics Communications, 2003, 220: 361. [本文引用:1]
[7]	WANG Gui-shi, CAI Ting-dong, WANG Lei, et al(王贵师, 蔡廷栋, 汪磊, 等). Chinese Journal of Lasers(中国激光), 2011, 38(10): 1008002. [本文引用:3]
[8]	SUI Yan-guang, DONG Zuo-ren, CHEN Di-jun, et al(孙延光, 董作人, 陈迪俊, 等). Chinese Journal of Lasers(中国激光), 2013, 40(4): 0408002. [本文引用:1]
[9]	DING Wu-wen, SUN Li-qun, YI Lu-ying(丁武文, 孙利群, 衣路英). Acta Physica Sinica(物理学报), 2017, 66(10): 100702-1. [本文引用:1]
[10]	SHEN Ting-ao, TU Ya-qing, LIU Xiang-yu, et al(沈廷鳌, 涂亚庆, 刘翔宇, 等). Chinese Journal of Scientific Instrument(仪器仪表学报). 2014, 35: 2153. [本文引用:2]
[11]	Liberti L, Lavor C, Maculan N, et al. SIAM Review, 2014, 56(1): 3. [本文引用:1]

2014

0.0

... 引言可调谐半导体激光吸收光谱技术(TDLAS)具有无需预处理、选择性强、响应速度快、高灵敏度和高精度等优点^[1,2,3,4], 已经广泛应用在气体检测中 ...

2013

0.0

2016

0.0

2014

0.0

2016

0.0

LIU

Hui-fang

, LI

Bin

, HE

Qi-xin

, et al(刘慧芳, 李彬, 何启欣, 等). Acta Photonica Sinica(光子学报), 2016, 45(4): 0423004-1.

By using Digital Signal Processor (DSP), a digital lock-in amplifier was experimentally demonstrated for extracting 1 f and 2 f signals in gas detection based on Tunable Diode Laser Absorption Spectroscopy (TDLAS). The orthogonal lock-in amplifying theory was introduced, the algorithm for extracting harmonic signals was designed, and both hardware structure and DSP software were proposed. By using the prepared CH 4 samples within the concentration range of 1%~5% and the developed lock-in amplifier, detailed experiments were carried out to derive the device's performances. As shown by experimental results, the system's signal-to-noise ratio at the frequency of 2 f signal is 34 dB under the CH 4 concentration of 5%, which indicates good functions of the lock-in amplifier. The amplitude ratio between 2 f and 1 f signals is linear to gas concentration. Considering gas preparation and diffusion time, the response time of the detection system is about 96 ~ 98 s. The concentration variation is-92 ppm~+118 ppm for the measured CH 4 sample with a concentration of 20000 ppm. The limit of detection obtained from the Allan variance is 29.52 ppm. Compared with analog and commercial lock-in amplifiers, the self-developed device shows great applications in infrared gas detection because of its simple structure, small size, low cost and easy integration.

1. State Key Laboratory on Integrated Optoelectronics, College of Electronic Science and Engineering, Jilin University, Changchun 130012, China; 2. Institute of Semiconductors, Chinese Academy of Sciences, Beijing 100083, China

利用可调谐二极管激光吸收光谱技术检测气体浓度时,为了从红外传感信号中提取一次及二次谐波信号来表征气体浓度,研发了一种基于数字信号处理器的数字正交锁相放大器.介绍正交锁相放大原理,设计谐波提取算法,给出数字正交锁相放大器的软硬件实现方案.利用配备的浓度为1%~5%的甲烷样品以及研制的锁相放大器,开展气体实验.实验结果显示,当甲烷浓度为5%时,在二次谐波对应的频率点处,测得的系统信噪比为34 dB,表明设计的锁相放大器具有较好的性能;测得的二次与一次谐波信号峰峰值的比值与气体浓度成线性关系;考虑动态配气以及气体沿管道传输的时间,检测系统的响应时间约为96~98 s;气体浓度为20 000 ppm时,测试浓度波动范围为-92 ppm ~ +118 ppm;根据Allan方差预测的系统检测下限为29.52 ppm.与模拟锁相放大器以及商用锁相放大器相比,本文研制的数字正交锁相放大器硬件结构简单、体积小、成本低、易于集成,在红外气体检测领域具有很好的应用前景.

... 在气体浓度测量过程中温度变化, 电流扫描信号直流电平漂移, 驱动电流与温度的漂移, 热敏电阻、激光器芯片的非完全耦合, 光路、电路老化等会造成激光器波长偏移^[5] ...

2003

0.0

2011

0.0

... Stewart等^[7]提出了基于频率调制技术的光谱吸收光纤气体检测方法, 通过对DFB激光器的中心波长进行调制, 使其对气体的吸收线进行缓慢扫描, 得到与气体浓度相关的高次谐波信号 ...

... 王贵师等^[7]利用Labview软件设计数字比例一积分一微分(PID)算法反馈控制半导体激光器的电流来提高激光器输出波长稳定性 ...

... 孙延光等^[8]在文献[7]的基础上采用了电流和温度双重反馈稳频技术的瓦斯浓度测量方案等等 ...

2013

0.0

... 孙延光等^[8]在文献[7]的基础上采用了电流和温度双重反馈稳频技术的瓦斯浓度测量方案等等 ...

2017

0.0

DING

Wu-wen

, SUN

Li-qun

, YI

Lu-ying

(丁武文, 孙利群, 衣路英). Acta Physica Sinica(物理学报), 2017, 66(10): 100702-1.

Methane is an important raw material for various petrochemicals in industrial fields and as also a clean fuel in daily life. However, as an inflammable and explosive material, methane leak can lead to disastrous consequences such as fire and explosion. Furthermore, as a kind of greenhouse gas, methane has stronger influence on global warming than carbon dioxide. In this paper, we present a new high sensitive scheme for methane remote sensing, which can facilitate detection and location of methane leakage. And the 2 v 3 band (near 1653.7 nm) of methane is chosen as the target transition which is free from the absorption of the other molecule in atmosphere. A tunable distributed-feedback diode laser is adapted to scan across the target transition. A Fresnel lens with a diameter of 150 mm is employed to collect the ambient backscattering light from natural features such as the buildings. The first harmonic signal is used to normalize the second harmonic signal to remove the influence introduced by the unknown reflectance factor of the actual target, therefore no retro-reflector is needed. Traditional tunable diode laser absorption spectroscopy (TDLAS) method has difficulty in locating the second harmonic signal peak position in low concentration conditions because of low signal-noise-ratio (SNR). To improve the SNR especially in low concentration environment, a scheme named “baseline-offset” TDLAS is presented in the paper, in which a reference cell filled with standard methane sample is inserted into the measuring optical path. The reference cell can also be used to calibrate the sensor. Furthermore, the reference cell can be used to lock the central frequency of the diode laser to the absorption peak position to monitor concentration fluctuation continuously. In the peak-locking mode, the sensor demodulates the third harmonic signal as error signal to control the injection current of the laser source with PID control. Moreover, one advantage of peak-locking mode is that the measurement frequency is about two orders of magnitude higher than the traditional TDLAS method. With “baseline-offset” TDLAS, the remote sensor described in this paper obtains SNRs as high as 19 and 16 at a stand-off distance of 10 m and 20 m, respectively. With such a high SNR, there is no necessity for complex algorithm in absorption peak position location. By defining the standard deviation of the measuring concentration as the detection limit, experimental results show that the proposed methane remote sensor has detection limits of 5 ppm · m at a distance of 10 m and 16 ppm·m for 20 m, respectively, while measuring the ambient methane. In peak-locked mode, the experimental system has a detection limit of 22 ppm·m at a distance up to 37 m and can monitor rapid concentration fluctuation in.

State Key Laboratory of Precision Measurement Technology and Instruments, Department of Precision Instruments, Tsinghua University, Beijing 100084, China

讨论了一种新的高灵敏度甲烷遥测方法,利用可调谐激光二极管的调制光谱技术扫描甲烷的吸收峰,通过在测量光路中插入参考气池,增强低浓度情况下的吸收峰辨识能力,以提高甲烷浓度遥测信号的信噪比.此外,可以将激光器的中心波长锁定至气体吸收峰的峰值位置从而使仪器工作于吸收峰锁定模式,进行甲烷浓度的连续监测.实验结果表明,在测量距离分别为10 m和20 m时,周围环境中的甲烷积分浓度探测极限可以分别达到5 ppm·m和16 ppm·m.在吸收峰锁定工作模式下,系统在37 m距离处具有22 ppm·m的检出限,并可以监测甲烷浓度的快速变化.

... 针对这种问题, 通常是利用谱线寻峰和相关^[9,10]的方法, 实现对光谱偏移的复位和对准, 但在对该方法分析讨论后发现其精度受随机噪声、背景噪声、漂移噪声的影响 ...

2014

0.0

... 由于信号与噪声并不相关^[10], 即相关值为0, 所以R_Sf(#cod#x00394 ...

2014

0.0

... 为了更好地区分三种方法的微小差异, 通过对这三种方法消除激光器波长偏移量后的光谱信号与预设的光谱信号进行相关系数和欧式距离^[11]法评估, 如图11所示 ...