1.Key Laboratory of Advanced Transducers and Intelligent Control System, Ministry of Education and Shanxi Province, Taiyuan University of Technology, Taiyuan 030024, China 2.College of Physics and Optoelectronics, Taiyuan University of Technology, Taiyuan 030024, China
Fund Project:Project supported by the National Natural Science Foundation of China (Grant Nos. 61527819, 61875146), the Key R&D Program (High-Tech Field) of Shanxi Province, China (Grant Nos. 201903D121177, 201803D121064), and the Program for Sanjin Scholar of Shanxi Province, China (Grant No. 201617)
Received Date:11 November 2020
Accepted Date:05 December 2020
Available Online:11 May 2021
Published Online:20 May 2021
Abstract:Nowadays, distributed dynamic sensing technology based on stimulated Brillouin scattering has been widely employed in civil structure health monitoring, disaster warning, national defense, etc. In this paper, we propose and experimentally demonstrate a novel Brillouin optical correlation-domain analysis based-on gain-switch modulation and dual-slope assisted method for achieving high-accuracy large-range dynamic strain measurement. In single-slope assisted chaotic Brillouin sensing, the measurement accuracy of dynamic strain is deteriorated by the inherent characteristics of time delay signature and power stochastic fluctuations. First, the mechanism behind the acoustic field deterioration and principle of background noise suppression are analyzed theoretically. Then, the chaotic continuous pump light is modulated into pulsed light with a higher extinct ratio of 48.6 dB, where the electro-optical modulator is replaced by a gain switch. And thus, the noise peaks, induced by the secondary peaks and irregular basal oscillations of chaotic auto-correlation curve, are greatly restrained. Comparing with the electro-optical modulator-based system, the signal-to-noise ratio of stimulated Brillouin acoustic field is increased by 3.31 dB in simulation and the signal-to-background ratio of Brillouin gain spectrum is doubled in experiment. Consequently, the measurement accuracy of dynamic strain is improved from ± 40.2 με to ± 23.1 με and the relative error decreases from ± 5.0% to ± 2.9% in single-slope assisted system based-on gain switch modulation. In addition, a dual-slope assisted method is inspired to eliminate the detrimental effect caused by the intrinsic power fluctuations of chaotic laser. A verification experiment is pre-conducted that the dynamic strain could be correctly interrogated although a wide range pump power variation has been manually applied. The measurement accuracy is ultimately enhanced to ± 8.1 με and the relative error is ± 1% correspondingly, implying a higher stability. The dynamic range of this proposed system is retained at 800 με, which is approximately 5 times as large as the dynamic range of the traditional dual-slope assisted configurations. The 4-cm spatial resolution along 30-m FUT is also investigated and consistent with the result obtained previously. A larger measurement range and a higher vibration frequency would be further explored by using the multi-slope assisted method and piezoelectric ceramic oscillator respectively. The superior slope-assisted chaotic Brillouin optical correlation-domain analysis will provide a new solution for the accurate positioning and real-time monitoring of dynamic parameters in modern industry. Keywords:chaotic laser/ dynamic strain measurement/ slope assisted/ Brillouin optical correlation-domain analysis
原始时域门控CBOCDA系统中, 连续混沌泵浦光被EOM调制为脉冲光, 通过调节射频脉冲电压控制信号消光比; 然而, EOM自身特性只能提供小于40 dB的消光比, 且存在插入损耗大、消光比不稳定、偏压控制复杂等缺点[24], 导致传统时域门控系统的脉冲消光比差、系统复杂度高、背景噪声抑制不完全. 因此, 提出基于增益开关(gain switch, GS)的调制方案, 实现高且稳定的消光比而进一步提升信噪比(signal-to-noise ratio, SNR)、降低系统复杂度. 图2(a)所示为实验测得原始状态、EOM调制与GS调制混沌激光时序, 并对不同调制方式下的消光比$ER = 20 \cdot \log ({{{V_{{\rm{peak}}}}} / {{V_{{\rm{base}}}}}})$进行了对比分析. 由图2(a)可以看出, EOM调制提供仅25.1 dB的ER值, GS调制消光比可达48.6 dB, 比EOM调制提升了约23.5 dB. 根据(4)式可知, GS的高消光比调制可大幅抑制${z^{n, \;n = \pm 1, 2, \cdot \cdot \cdot }}$和${z^{{\rm{base}}}}$噪声峰, 为了验证这一效果, 本文首先仿真模拟了不同调制情况下的SBS声波场强度. 如图2(b)所示, 黑色曲线为连续混沌激光入射30 m光纤中时的声波场分布情况, 已知混沌激光时延腔长为5 m, 因此除光纤中心15 m处${z^0}$外, 10与20 m处存在强度较大的${z^{ \pm 1}}$, 5与25 m处甚至存在${z^{ \pm 2}}$; 此外, 混沌激光自相关基底激发的SBS噪声强度也被清晰观测到. 混沌SBS声波场信噪比定义为中心峰强度${A_{{\rm{cp}}}}$与噪声峰峰值强度${A_{{\rm{np}}}}$的比值$SNR = 10 \cdot \log $$ ({{{A_{{\rm{cp}}}}} / {{A_{{\rm{np}}}}}})$, 则基于连续混沌激光的SBS声波场SNR约为5.69 dB. EOM调制系统中, ${z^{ \pm 1}}$和${z^{{\rm{base}}}}$被抑制效果明显, ${z^{ \pm 2}}$几乎完全消失, SNR值也提高至8.33 dB, 然而未完全消除的噪声峰仍会随光纤长度持续累积并恶化探测光的SBS增益. GS的高消光比调制使得${z^{ \pm 1, 2}}$和${z^{{\rm{base}}}}$几乎均被完全抑制, SNR提升至11.64 dB, 约为基于连续混沌激光系统的2倍, 且与EOM调制系统相比也有3.31 dB的提升. 图 2 不同调制方式及作用效果 (a)不同调制方式下脉冲混沌光时序及消光比; (b)不同调制方式下仿真的混沌受激布里渊声波场强度及信噪比 Figure2. Different modulation methods and the corresponding effects: (a) The time series of pulse chaos and ERs of these under different modulation modes; (b) simulation results of chaotic SBS acoustic field amplitude and the corresponding SNR.
根据本课题组的早期研究结果[23,25,26], CBOCDA系统噪声峰的抑制效果将直接体现为混沌BGS的SBR值水平, 因此, 本实验继续对比了不同调制方式下的混沌BGS. 为了更好地比较分析BGS包络及SBR, 将EOM调制和GS调制系统中测得的混沌BGS进行了归一化处理, 如图3所示. 可以看出, GS调制系统背景噪声水平明显更低, 其SBR值为15.22 dB, 约为EOM调制系统的2倍. 此外, 图3插图为测得的原始增益谱, 可以看出, GS高消光比调制显著提升了SBS增益强度, 通过PD探测到约0.013 mV的峰值增益增量. GS调制实现了与长腔反馈混沌激光[21]相同的噪声抑制效果, 且实用性更强、易于实现长距离高信噪比BGS测量. 此外, 已知SA技术是通过探测锁定频率下的布里渊增益实现动态应变解调的, 因此, 优化后的增益谱可促进理论上获得更精确的动态应变测量. 图 3 不同调制方式下的归一化混沌布里渊增益谱及其信号背景噪声比; 插图为测得的原始布里渊增益谱 Figure3. Normalized chaotic BGSs and SBR analysis of these, under different modulation modes. Inset view: the measured original BGSs, corresponding to the main view.
基于图1所示的实验装置, 泵浦路分别利用EOM和GS实现脉冲强度调制, 探测光频率锁定在混沌BGS下降沿10.865 GHz处, 通过电动拉伸位移平台的往复运动提供动态应变. 为保证探测光功率对动态应变的线性响应, 本实验施加的动态应变大小为0—800 με, 消除由功率-应变转换系数引起的测量误差[20]. 图4(a)和图4(b)分别给出了EOM调制与GS调制下SSA-CBOCDA系统测得的动态应变时序及其正弦拟合曲线, 由图可知, 两种调制方式下动态应变值均可被准确地解调. 为了表示测量精度, 本文首先比较了两条正弦拟合曲线的相关系数${R^2}$, GS调制系统的相关系数0.9793略大于EOM调制系统, 但均接近于1, 表明两者的拟合结果均较好, 但是GS调制系统的相关性更高. 此外, 本文详细分析了两种调制方式下的动态应变测量误差, 如图4(c)所示, 蓝色和红色点图分别表示EOM调制、GS调制系统中15次独立重复测量的动态应变误差均值及不确定度, 插图为单次测量误差的测算依据和典型结果. EOM调制系统中误差明显更大, 标准偏差${\sigma _{\rm{E}}}$为 ± 40.2 με, 相对误差达 ± 5.0%. 经过GS调制改进, SSA-CBOCDA系统动态应变测量误差${\sigma _{\rm{G}}}$降低至 ± 23.1 με, 相对误差缩小为 ± 2.9%. 此外, 不同调制系统中标准偏差的不确定度均较小, 进一步表明精度测量结果准确性高、系统稳定性好. 图4(d)所示为两种调制方式下测得的动态应变的振动频率, 结果表明基频1.3 Hz和二次谐振频率2.6 Hz均可被准确检测; 需要指出, 本实验采用的电动位移拉伸平台由一对步进电机控制, 电机的低速往复运动导致振动频率较低, 下一步将考虑使用压电陶瓷振荡器提供高频的动态应变. 图 4 不同调制方式下的动态应变测量结果 (a) EOM调制系统动态应变测量时序及正弦拟合曲线; (b) GS调制系统动态应变测量时序及正弦拟合曲线; (c)两种调制方式下的动态应变测量误差, 插图为单次测量典型结果; (d)两种调制方式下的动态应变频率 Figure4. The results of dynamic strain measurement under different modulation modes: (a) Time series and sine fitting curve of dynamic strain in EOM modulation system; (b) time series and sine fitting curve of dynamic strain in GS modulation system; (c) measurement error of dynamic strain under two modulation modes, inset view is the typical result of a single measurement; (d) vibration frequency under two modulation modes.
如前所述, 为了免除光纤松弛对测量结果的影响, 本实验将光纤拉伸段施加约300 με的预拉伸, 并基于此状态下的增益谱选定频率${v^ + }$, ${v^-}$. 如图5(b)所示, 此时${G_{300 - }} \approx {G_{300 + }} \approx 0.5$, 即${R_{\rm{B}}} \approx {\rm{0}}$dB; 当施加应变分别为0和800 με时, 根据图中圆点标识, 计算得增益比值分别约为–0.42和0.62 dB, 均保持在图5(a)${R_{\rm{B}}}$曲线所示的线性区内. 具体实验时, 如图6(a)所示, 分别测量两个频率下施加不同大小准静态应变时的探测光时序, 计算得不同应变对应的${R_{\rm{B}}}$曲线; 由图6(a)可以看出, 0与800 με时${R_{\rm{B}}}$取值与理论分析基本一致, 当应变以200 με步进从0 με增加至800 με时, ${R_{\rm{B}}}$几乎呈现线性上升趋势, 且300 με时${R_{\rm{B}}}$值近似于0 dB. 不同应变对应${R_{\rm{B}}}$的均值被进一步提取并做线性拟合, 如图6(b)所示, 相关系数$R_{\rm{R}}^2$约等于0.9990表明了${R_{\rm{B}}}$与应变值的良好线性关系; 根据曲线斜率可得, 该DSA系统的最终增益比—应变转换系数${C_{\rm{s}}}$约为0.0012 dB/με. 图 6 双斜坡辅助系统的增益比-应变转换系数 (a)不同大小准静态应变对应的增益比曲线; (b)根据线性拟合曲线计算的转换系数 Figure6. Estimate of the conversion coefficient, ${C_{\rm{s}}}$, between the strain value and ${R_{\rm{B}}}$ in DSA system: (a) The ${R_{\rm{B}}}$ curves under different quasi-static strain; (b) final coefficient calculated from the linear fitted curve.
为了验证DSA技术可以消除泵浦功率波动影响, 本文测量了基于上升沿和下降沿中心频率锁定的SSA动态应变响应时序, 动态应变大小为0—600 με, 同时手动添加了泵浦光先减后增的大范围功率波动. 如图7(a)所示, 无论上升沿或下降沿频率锁定, SSA技术均可实时反映动态应变引起的布里渊增益强度变化, 然而其响应时序呈现与泵浦光功率一致的先下降后上升趋势, 峰值功率波动达0.7 dB, 无法利用功率-应变转换系数准确解调动态应变大小. 根据DSA技术原理, 将图7(a)中的两列时序进行比值计算并利用${C_{\rm{s}}}$解调便得图7(b)中蓝色曲线所示动态应变序列. 可以看出, 动态应变大小已被准确还原, 利用正弦拟合曲线进一步验证了较好的动态应变测量正确度(相关系数${R^2} \approx 0.9642$). 图 7 双斜坡辅助技术消除泵浦功率波动影响的验证实验 (a)手动施加泵浦光功率波动时利用单斜坡辅助技术测得的动态应变响应时序; (b)双斜坡辅助技术解调的动态应变大小 Figure7. The verification experiment of pump-power-independent measurement by using DSA method: (a) Measured Brillouin amplitude using SSA method when a dynamic strain is applied with a manual change of pump power; (b) measured dynamic strain using DSA method.
基于上述验证实验和转换系数, 本文在仅有混沌激光自身功率波动的情况下, 利用DSA技术进行了动态应变大小及精度的最终测定. 如图8所示, 图8(a)中蓝色曲线为实际测量的动态应变时序, 0—800 με动态应变被准确测量; 对测量值做如红色点图所示的正弦拟合, 拟合相关性高达0.9972; 同时如图8(b)中蓝色点图所示, 通过15次独立重复测量计算得应变测量误差${\sigma _{\rm{D}}}$为 ± 8.1 με, 插图即为单次测量误差的测算依据和典型结果; 相对误差下降至 ± 1%, 动态应变测量精度明显提升. 动态应变标准偏差的不确定度仅± 1.0 με, 表明了双斜坡辅助系统具有更好的测量准确度和系统稳定性. 此外, 动态应变振动基频1.3 Hz也被准确检测, 动态应变频率与SSA系统测量结果一致(电动位移平台的运动速度并未改变). 图 8 施加0?800 με时的双斜坡辅助系统动态应变测量结果 (a)动态应变测量时序及正弦拟合曲线; (b)动态应变测量误差, 插图为单次测量典型结果; (c)动态应变振动频率 Figure8. Measurement results of DSA system when a dynamic strain of 0?800 με is applied: (a) Measured time trace and sine-fitted curve; (b) measurement error, inset view is the typical result of a single measurement; (c) vibration frequency.