【作者】 陈伟；

【导师】 孟洲；

【作者基本信息】 国防科学技术大学 ， 光学工程， 2013， 博士

【摘要】 随着光电子技术和光纤传感技术的日趋成熟，以光纤水听器为典型代表的干涉型光纤传感系统在潜艇探测、石油天然气储层勘探和地震监测等领域取得了广泛应用。近年来，随着掺铒光纤放大技术和光纤拉曼放大技术的进步，干涉型光纤传感系统朝着远程化方向发展，在增大传感距离的同时不可避免的加剧了光纤中的非线性效应。前人对非线性效应的研究主要针对光纤通信系统或者效应本身，而鲜有基于干涉型光纤传感系统的非线性效应研究。与光纤通信系统以误码率为主要研究对象不同，干涉型光纤传感系统最为关注的是相位噪声特性，因为它决定了系统的探测灵敏度，而包括受激布里渊散射（SBS）、四波混频（FWM）、调制不稳定性（MI）、自相位调制（SPM）和交叉相位调制（XPM）在内的各种非线性效应都会引入相位噪声，导致系统性能的严重下降，因此各种非线性效应的影响及抑制是发展远程干涉型光纤传感系统的关键技术。论文在介绍远程干涉型光纤传感系统基本结构和基本原理的基础上，首次对考虑各种非线性效应的远程干涉型光纤传感系统的相位噪声结构进行了详细分析，得出系统相位噪声有两个来源：强度噪声转化的相位噪声和激光线宽引入的相位噪声，分别源于光强的起伏和光频的抖动。前者又称为GM噪声，其线性部分可由SBS和FWM引入，而非线性部分源于光克尔效应，即由SPM和XPM引入。SBS、FWM和MI还可引起线宽展宽，导致相位噪声的增加。系统相位噪声结构的提出为远程干涉型光纤传感系统的设计和应用提供了重要指导。利用SBS的定域起伏模型研究前向输出光和后向散射光的强度噪声，对于前向输出光，强度噪声开始时很小，达到SBS阈值后迅速增大，然后逐渐趋于稳定；对于后向散射光，超过SBS阈值后，强度噪声呈减小趋势，且减小速度由快变慢，最终趋于稳定，这与实测的前后向强度噪声变化趋势是一致的。实测了前向输出光的相位噪声，显示与相应的强度噪声变化趋势一致，验证了相位噪声可由强度噪声转化而来的结论；同时通过测量SBS发生前后的线宽变化，证实了SBS可引起线宽展宽从而引入相位噪声的结论。采用光频调制和相位调制抑制SBS，前者抑制效果有限，而后者可将SBS阈值提高7dB。但相位调制基于激光线宽展宽，在抑制SBS及其引入的相位噪声的同时，会引入与线宽相关的相位噪声，故实际应用中要综合考虑SBS抑制作用和激光线宽展宽作用，以实测相位噪声最小为标准寻找二者的最佳平衡，该结论是前人研究中未曾涉及的。此外，研究了SBS对分布式光纤拉曼放大器的影响，并用SBS慢光技术测得布里渊增益带宽为50MHz。数值模拟了FWM效率及强度噪声与信道间隔和光纤长度的关系，并利用FWM准相位匹配条件进行解释。以双波传输为例实测了FWM引入的强度噪声，发现由FWM新产生光与原信道光之间能量交换不稳导致的强度噪声可以忽略，对强度噪声影响最大的是SBS而非FWM。以三波传输和四波传输为例实测了FWM引入的相位噪声，发现FWM对相位噪声的影响可以忽略，这是由于信道不完全等间隔导致新产生光与原信道光之间存在拍频，且大于干涉型光纤传感系统中所用窄带光探测器的带宽，从而滤除了拍频噪声，并基于此提出利用窄带光探测器抑制FWM引入的相位噪声，这也是前人没有讨论过的。此外，利用泵浦回波技术研究了FWM和SBS结合的特殊情形并从FWM相位匹配的角度进行解释。阐述了产生MI的三种物理机制，并从光纤中脉冲传输的非线性薛定谔方程出发对MI进行了数值模拟。在掺铒光纤放大器产生的宽带放大自发辐射光背景上观察了自发MI和感应MI。研究了MI阈值与输入光谱和光纤长度的关系，发现宽带光谱比单频光谱的MI阈值要高，且MI阈值与有效光纤长度成反比。分析了MI与SBS的不同，即MI对脉冲峰值功率响应而SBS对平均功率响应，这是由于MI的响应时间仅为10fs而SBS的响应时间达10ns。考虑到MI发生后会引入大量相位噪声，利用脉冲峰值功率控制和窄带光纤光栅滤波的方法抑制MI。研究了MI与FWM结合的调制不稳定性共振现象并从FWM相位匹配的角度进行解释，首次提出了一种FWM边带的选择方法，该方法基于连续光信道和脉冲光信道的同时使用。从光放大链光纤段数、输入功率、光纤长度和信道数目四个方面，详细阐述了各种非线性效应对系统性能的影响。对于超过50km的远程传输，当输入平均功率大于4mW时，要考虑SBS对系统的影响；当输入峰值功率大于110mW时，要考虑MI对系统的影响；当信道数目多于20时，要考虑XPM对系统的影响；由于干涉型光纤传感系统的窄带探测特性，无需考虑FWM对系统的影响。利用SBS阈值和MI阈值，综合考虑系统时分复用效率和传输距离等因素，对所用脉冲占空比进行优化。当系统中采用相位调制器抑制SBS时，对于50km传输光纤和25MHz调制频率，输入功率为3mW至5mW、6mW至10mW、12mW至14mW时对应的最佳调制度分别为0.57π、0.71π和0.86π。上述结论为远程干涉型光纤传感系统实际应用时的参数选择提供了重要依据。更多还原

【Abstract】 With the developments of optoelectronics and fiber sensing technologies,interferometric fiber sensing systems such as fiber hydrophone have been widely usedin the fields including submarine detection, oil and natural gas prospecting, andearthquake inspection. Recently, with the developments of erbium-doped fiber amplifier(EDFA) and fiber Raman amplifier (FRA), interferometric fiber sensing systemsdevelop towards the long-haul direction. Although the sensing distance is increased,fiber nonlinear effects also become significant. The foregone researches focus on fibercommunication system or nonlinear effects themselves, while pay little attention tononlinear effects in interferometric fiber sensing systems. Compared with fibercommunication systems which treats bit error rate (BER) as its research focus,interferometric fiber sensing systems pay great attention to phase noise, which decidesthe detection sensitivity of the system. Considering that a variety of nonlinear effectssuch as stimulated Brillouin scattering (SBS), four-wave mixing (FWM), modulationinstability (MI), self-phase modulation (SPM) and cross phase modulation (XPM)introduce phase noise and lead to system performance degradation, influences andsuppression techniques of nonlinear effects are key technologies for the long-haulinterferometric fiber sensing systems.Based on the structure and principle of the long-haul interferometric fiber sensingsystem, the phase noise structure of the system considering a variety of nonlinear effectsis analyzed for the first time. The phase noise results from two sources i.e. phase noisetransferred from intensity noise and phase noise induced by laser linewidth, originatingfrom intensity fluctuation and frequency dithering, respectively. The former is alsocalled GM noise with its linear part introduced by SBS and FWM and its nonlinear partresulting from optical Kerr effect i.e. SPM and XPM. SBS, FWM and MI can causelinewidth broadening, leading to the increase of phase noise. The proposition of phasenoise structure provides guidance for the design and application of long-haulinterferometric fiber sensing systems.The SBS localized fluctuating model is used for investigating the intensity noise ofthe forward output light and the backscattered light. It is found that, for the forwardoutput light, the intensity noise is small at first, increases dramatically when reachingthe SBS threshold, and then tends to stabilize gradually. For the backscattered light,after the SBS threshold, the intensity noise decreases fast at first and then slowly, andbecomes stable at last. The above results agree with the measured variations of theforward and backward intensity noise. The forward output phase noise is also measuredand its variation is accordant with that of the corresponding intensity noise, whichverifies that phase noise can be transferred from intensity noise. The linewidth with and without SBS are measured, which confirms that SBS can cause linewidth broadeningand introduce phase noise. Frequency modulation and phase modulation are used tosuppress SBS. The suppression effect of the former method is very limited, while thelatter method can improve the SBS threshold by7dB. However, phase modulation isbased on laser linewidth broadening. Although SBS as well as its induced phase noise issuppressed, the phase noise related to linewidth is induced. Therefore, in practicalapplications, both SBS suppression and laser linewidth broadening should be considered,and the measured phase noise is used for finding the optimum balance of the twoopposite effects, which has been not referred in former researches. Furthermore, theinfluence of SBS on FRA is investigated, and the Brillouin bandwidth is measured to be50MHz using SBS slow light technique.The relations between FWM efficiency as well as its intensity noise and thechannel spacing and fiber length are numerically simulated, and the results areexplained using FWM quasi-phase-matching condition. Taking two-wave transmissionfor example, FWM induced intensity noise is measured and it is found that the intensitynoise caused by the energy exchange instability between FWM generated light andoriginal light can be ignored, and what should be concerned most is SBS rather thanFWM. Taking three-wave and four-wave transmission for examples, FWM inducedphase noise is measured and it is found that the influence of FWM on phase noise canalso be ignored, which is because the channels are not perfectly equally spaced and thebeat frequency between the generated light and original light is larger than thebandwidth of the narrowband photodetector used in the interferometric fiber sensingsystem, leading to that the beat noise is filtered. Based on this, suppressing FWMinduced phase noise with narrowband photodetector is proposed, which has not beendiscussed in the former researches. Furthermore, the pump echo technique is used tostudy the case when FWM and SBS are combined, and the result is explained from theFWM phase-match point of view.Three kinds of physical mechanisms of MI are introduced. MI is numericallysimulated using nonlinear Schordinger equation which describes pulse transmission inthe fiber. Spontaneous MI and induced MI are observed on the background ofbroadband amplified spontaneous emission (ASE) caused by EDFA. The relationsbetween the MI threshold and the input spectrum and fiber length are investigated and itis found that broadband spectrum has a higher MI threshold than monochromaticspectrum, and MI threshold is inversely proportional to effective fiber length. Thedifference between MI and SBS is analyzed and it is found that MI responds to pulsepeak power while SBS responds to average power, which is due to that the respondingtime of MI and SBS are10fs and10ns, respectively. Considering that MI causes amass of phase noise, the methods of pulse peak power control and narrowband fibergrating filtering are used to suppress MI. The modulation instability resonance (MIR) introduced by the combination of MI and FWM is investigated and explained fromFWM phase-match point of view. A method of selecting FWM sidebands is proposedfor the first time, which is based on the simultaneous utilization of a CW channel and apulse channel.Influences of nonlinear effects on system performance are discussed from fouraspects including fiber segment number of optical amplifier link, input power, fiberlength and channel number. For long-haul transmission over50km, the influence ofSBS should be considered when input average power is above4mW, and the influenceof MI should be considered when input peak power is above110mW, and the influenceof XPM should be considered when channel number is more than20, while theinfluence of FWM need not to be considered due to the narrowband detectioncharacteristic of interferometric fiber sensing system. Using SBS and MI thresholds andconsidering both the time-domain multiplexing (TDM) efficiency and transmissiondistance, pulse duty cycle is optimized. When phase modulator is used for SBSsuppression, for50km fiber and25MHz modulation frequency, the optimummodulation indices are0.57π,0.71π and0.86π when the input power is between3mWand5mW, between6mW and10mW, between12mW and14mW, respectively. Theabove conclusions provide good guidance for the parameter selection in the practicalapplications of interferometric fiber sensing systems.更多还原