Phase-Sensitive Optical Time-domain Reflectometry Based On Adaptive Modulation And Demodulation

Scattering in media has unique inherent randomness.Fiber Rayleigh scattering is one of them.The exploration and understanding of the characteristics of fiber Rayleigh scattering holds significant research value for optical fiber sensing,random fiber laser,etc.Based on optical fiber Rayleigh backscattering for sensing,phase-sensitive optical time-domain reflectometry(Φ-OTDR)is one of the most promising systems in the field of distributed optical fiber sensing.It stands at the forefront of both domestic and international research,playing a crucial role in national strategic development.This is not only due to its high sensitivity and fast response,but also because China has already laid close to sixty million kilometers of communication optical cables,accounting for approximately half of the total worldwide.With the iterative updates of integrated sensing and communication technologies,using existing communication optical cables for sensing will have significant applications in areas such as national defense and security(e.g.,underwater object positioning,border intrusion detection),geophysics(e.g.,seismic monitoring,oil and gas resource exploration),and urban life(e.g.,traffic and pipeline monitoring).This dissertation focuses on Φ-OTDR systems,with an emphasis on exploring new modulation and demodulation schemes.This includes the adaptive decoding method,adaptive waveform modulation,and adaptive signal recovery to form an adaptive ΦOTDR system with intelligent sensing capabilities.The main research contents are as follows:1.In order to overcome the limitation of requiring longer encoding lengths to achieve a high peak-to-side lobe ratio(PSLR)decoding in the conventional optical-pulse-codingΦ-OTDR system,a novel optical-pulse-coding Φ-OTDR based on mismatched filtering is proposed,achieving a scan-rate of 95%of the upper limit determined by the optical fiber length.At the receiving end,a mismatched filter is adaptively designed,enabling the decoding of even randomly phase coding,thereby offering new insights for integrated communication and sensing systems.Additionally,for improved signal processing efficiency in optical-pulse-coding schemes,we introduce a simultaneous single-bit coding modulation and single-bit reception decoding scheme based on pseudo-random codes.This scheme has three significant advantages:it reduces the hardware requirements for modulation and demodulation simultaneously,saves active optical amplification devices through pulse coding,and significantly enhances the computational efficiency of signal processing through one-bit sampling.These advantages further enhance the potential applications of the optical-pulse-coding Φ-OTDR system.2.A deep-learning-based adaptive waveform modulation Φ-OTDR is proposed to achieve probing pulse optimization based on channel response inversion and fading suppression in any given region.Initially,we adopt a strategy based on deep image priors to avoid the need for constructing a large-scale training dataset for pre-training deep learning models.Through iterative calculations,an effective network can be constructed,achieving the preliminary implementation of an adaptive waveform modulation system.Subsequently,a lightweight,generalizable,and intelligently computed SenseNet network is constructed based on a self-developed numerical model.This network extracts common features of optical fiber Rayleigh scattering responses,and perceives the current state of the sensing fiber.Fading suppression can be realized with adaptive waveform modulation in any given region.The SenseNet network,along with an adaptive waveform modulation algorithm,is deployed in real-time signal processing validation on a field-programmable gate array(FPGA).3.A deep-learning-based adaptive signal recovery Φ-OTDR is proposed,achieving the integration of demodulation and noise reduction signal processing.Firstly,to enhance the generality of the SenseNet network,a dual-pulse probing scheme for extracting Rayleigh scattering responses is proposed.This leads to the establishment of the adaptive denoising SenseNet(D-SenseNet),which overcomes cumulative phase noise caused by laser frequency drift.Demodulation based on the extracted Rayleigh scattering responses results in improved spatial resolution.Furthermore,demodulation and denoising functionalities are integrated into one deep learning network,resulting in a 2D-SenseNet(Demodulation and Denoising SenseNet)network,which is built based on the replica with randomized perturbations(RRP)and transition optimization of phase(TOP).The strain noise level eventually decreased from 241pε/(?)to 98pε/(?)at 10km sensing distance and 0.9m spatial resolution.This approach not only adds denoising capabilities compared to traditional demodulation methods,but also exhibits greater robustness compared to existing artificial intelligence denoising solutions.In general,this dissertation is dedicated to addressing the challenges in the intelligence enhancement of Φ-OTDR systems.It establishes a novel adaptive modulation and demodulation Φ-OTDR system.We propose a deep-learning-based approach for efficiently extracting fiber Rayleigh scattering responses,enabling seamless integration of sensing systems with artificial intelligence.Building upon this foundation,we further introduce adaptive waveform modulation based on SenseNet,as well as adaptive signal recovery based on D-SenseNet and 2D-SenseNet.These investigations not only establish the groundwork for developing an integrated demodulation,denoising,and decision-making SenseNet(3D-SenseNet),but also pave the way for new research directions.