【作者】 王鹏；

【导师】 王春灿；

【作者基本信息】 北京交通大学 ， 信息与通信工程， 2025， 博士

【摘要】 类噪声脉冲(Noise-like pulse,NLP)是锁模激光领域中一种独特的脉冲形态,主要特征是复杂的时域结构和低相干,被应用于光学传感、非线性波长转换、精密机械加工、超连续谱产生、生物医学成像等前沿领域。该脉冲包络内包含大量幅度和相位高度随机的飞秒/亚皮秒量级子脉冲,在单脉冲层面呈现类似“噪声”的强随机特性,而在脉冲序列层面却表现出恒定的重复频率和稳定的时、频域平均轮廓。这种特殊的时频结构赋予其承受高非线性的能力,因而从主控振荡器直接输出的光谱宽度可突破增益窗口限制,并且脉冲能量可达微焦量级。同时,NLP锁模技术具有普适性,无需复杂的色散管理即可灵活地在各种配置的光纤谐振腔中实现,显著降低激光系统的设计难度和成本,更引发了研究者对其产生机制和物理特性的持续深入研究。本论文聚焦被动锁模光纤激光器中NLP的生成及波动机制,以及NLP向相干脉冲转化等核心问题展开深入研究,具体研究内容及所取得的创新成果如下:(1)结合实际参数,对全正常色散掺镱光纤激光器中NLP的产生机制和波动特性开展了理论建模与数值仿真研究。通过构建多维参数空间,深入探究了可饱和吸收体、泵浦功率水平及光谱滤波等关键参数对NLP锁模动力学的影响。分析了NLP在腔内往返过程中的相对强度噪声(Relative intensity noise,RIN)的动态特性,进而提出一系列谐振腔的结构和参数优化策略以最大化抑制腔内RIN水平,为开发低波动特性的NLP锁模光纤激光器提供了理论基础和设计指导。(2)研究了基于非线性偏振旋转和Lyot滤波器的NLP锁模掺镱光纤激光器。通过调节光谱滤波器与泵浦功率,该激光器可在NLP状态、调Q NLP状态和光学怪波状态之间实现灵活切换。生成的NLP具有14.8 MHz的重复频率、56.6 ps的波包宽度、1029 nm的中心波长和8 nm的3-dB带宽。利用色散傅里叶变换技术实时跟踪光谱的动态演化,揭示了此怪波的生成机制:NLP内部偶发的自发拉曼噪声,经过滤波器的波长选择作用,在数次腔内往返过程中被掺镱光纤及拉曼效应迅速放大形成怪波,后又随着增益的耗尽而迅速耗散。揭示了光纤激光系统中一种非常规的脉冲动力学,为光学怪波研究开辟了新的生成机制和技术途径。(3)提出并实现了基于前馈强度调制的NLP锁模掺铒光纤激光器。调制信号来自前馈光激发的电信号,使调制器能够模拟可饱和吸收效应。通过调节前馈光强和调制器偏置电压,该激光器能够在NLP状态、“h”形脉冲状态、调Q状态以及调Q锁模状态之间实现灵活切换。所生成的NLP具有2.79%的RIN水平、54.3 dB的信噪比、12.57 MHz的脉冲重复频率以及190 ps的平均包络宽度。系统研究了偏置电压、泵浦功率和前馈光强等关键实验参数对输出脉冲特性的影响,验证了前馈调制锁模方案的可控性和灵活性。本研究为NLP光纤激光器的实现提供了一种创新解决方案。(4)提出并实现了基于腔外前馈强度调制的全带宽(覆盖重复频率)RIN抑制技术,可有效降低NLP光源的RIN水平。相较于常规抑制RIN的方法,本方案具有高带宽和零滞后的优势,具备在单脉冲层面对每个NLP进行实时连续监测和调制的能力。通过调整优化实验参数,成功实现将NLP光源的RIN从5.62%降至1.91%。另外,结合数值仿真阐明了参数调控规律,分析了实验装置的改进方向,为开发低噪声NLP激光器提供了新的技术路径。(5)提出了以高掺锗大正常色散光纤作为拉曼介质的、基于NLP同步泵浦的超快拉曼光纤激光器,可用于波长转换以及NLP向相干脉冲的转化。通过数值计算研究了基于NLP同步泵浦的拉曼脉冲动力学,揭示了强走离效应在维持线性啁啾以及相干性中的关键作用。进而提出使用高掺锗大正常色散光纤作为拉曼介质:一方面,强走离效应能够平均化NLP的混沌振荡结构;另一方面,高掺锗光纤具有更高的拉曼增益,可缩短拉曼作用光纤长度,从而显著抑制自发拉曼噪声的积累。所生成的相干拉曼脉冲具有陡峭光谱边缘、线性频率啁啾及亚皮秒量级脉宽等特征。本研究论证了NLP向相干脉冲转化的可行性,有望在未来进一步进行实验验证,为超快光纤激光领域开辟新的发展方向。更多还原

【Abstract】 Noise-like pulse（NLP）,a unique pulse regime in mode-locked lasers,is characterized by complex temporal structures and low coherence.It has been extensively applied in cutting-edge fields including optical sensing,nonlinear wavelength conversion,precision machining,supercontinuum generation,and biomedical imaging.The NLP envelope comprises numerous femtosecond/sub-picosecond sub-pulses with highly randomized amplitudes and phases,exhibiting strong noise-like stochastic characteristics at the single-pulse level while maintaining constant repetition rates and stable temporal/spectral average profiles at the pulse-train level.This distinctive time-frequency structure endows NLP with exceptional nonlinear tolerance,enabling direct generation of spectra exceeding the gain bandwidth limit from master oscillators and achievingμJ-level pulse energy.Moreover,NLP mode-locking demonstrates universality,allowing implementation in diverse fiber resonator configurations without complex dispersion management,significantly reducing laser system design complexity and cost.These advantages have spurred continuous investigations into its generation mechanisms and physical properties.This dissertation focuses on NLP generation,fluctuation mechanisms in passively mode-locked fiber lasers,and NLP-to-coherent-pulse conversion,with key research outcomes outlined below:（1）Theoretical modeling and numerical simulation studies were conducted on NLP generation mechanisms in all-normal-dispersion ytterbium-doped fiber lasers using practical parameters.By constructing a multidimensional parameter space,the effects of critical parameters—including saturable absorber characteristics,pump power levels,and spectral filtering—on NLP mode-locking dynamics were systematically investigated.The dynamic behavior of relative intensity noise（RIN）during intracavity pulse circulation was analyzed,leading to optimized cavity structures and parameter strategies for maximal RIN suppression.These findings establish theoretical foundations and design guidelines for developing low-fluctuation NLP mode-locked fiber lasers.（2）A mode-locked ytterbium-doped fiber laser incorporating nonlinear polarization rotation and a Lyot filter was investigated.Through spectral filtering and pump power adjustments,the laser achieved flexible switching among NLP,Q-switched NLP,and optical rogue wave states.The generated NLP exhibited a 14.8 MHz repetition rate,56.6ps envelope width,1029 nm central wavelength,and 8 nm 3-dB bandwidth.Real-time spectral evolution tracking via dispersive Fourier transform revealed unconventional rogue wave dynamics:Spontaneous Raman noise sporadically emerging within NLP underwent wavelength-selective amplification through ytterbium-doped fiber and Raman effects during multiple cavity roundtrips,forming rogue waves that fully dissipated upon gain depletion.（3）A feedforward intensity modulation-based mode-locked erbium-doped fiber laser was proposed and implemented.The modulation signal derived from feedforward light-generated electrical signals enabled the modulator to emulate saturable absorption effects.By tuning feedforward light intensity and modulator bias voltage,the laser demonstrated flexible transitions among NLP,"h-shaped"pulse,Q-switched,and Q-switched mode-locked states.The NLP output achieved 2.79%RIN,54.3 dB signal-to-noise ratio,12.57MHz repetition rate,and 190 ps average envelope width.Systematic studies on bias voltage,pump power,and feedforward intensity validated the controllability and adaptability of this feedforward modulation mode-locking scheme.（4）An extracavity feedforward intensity modulation technique for full-bandwidth（covering repetition frequency）RIN suppression was developed.Compared to conventional methods,this approach offers high bandwidth and zero latency,enabling real-time continuous monitoring and modulation of individual NLPs at the single-pulse level.Through parameter optimization,RIN of the NLP source was reduced from 5.62%to 1.91%.Numerical simulations further elucidated parameter adjustment principles and proposed experimental setup improvements,providing a novel technical pathway for low-noise NLP laser development.（5）An ultrafast Raman fiber laser based on NLP-synchronized pumping was proposed,utilizing highly Germanium-doped large-normal-dispersion fiber as the Raman medium for wavelength conversion and NLP-to-coherent-pulse transformation.Numerical studies on Raman pulse dynamics under NLP-synchronized pumping revealed the critical role of strong walk-off effects in preserving linear chirp and coherence.The proposed high-GeO₂ fiber design achieves dual benefits:Strong walk-off averages chaotic NLP oscillations,while enhanced Raman gain shortens the required fiber length,effectively suppressing spontaneous Raman noise accumulation.The resulting coherent Raman pulses exhibited steep spectral edges,linear frequency chirp,and sub-picosecond durations.更多还原