| Phase-sensitive optical time-domain reflectometric system(?-OTDR)can quantitatively measure the amplitude and frequency of disturbance signal and qualitatively detect the type of disturbance signal.It has many advantages such as multi-point disturbance monitoring at the same time,long monitoring distance,large dynamic range,high sensitivity,fast response,combination of transmission with sensing in a single fiber,less resources occupation,intrinsically passive sensing modality,immune to electromagnetic interference,long-term stable work performance in high temperature,high humidity,high corrosion and other harsh environments.As such,it is increasingly playing an important role in intrusion detection application at military bases,petrochemical plants,nuclear power plants,airports,banks,subways,prisons and other important security sites;it has also been widely used in damage detection of infrastructure such as oil and gas pipeline leakage,railway fracture and structural damage.Based on its great development potential and considerable practical value,it has been widely studied by researchers and widely anticipated by potential users.However,the system is not perfect in terms of stability of operation,real-time performance of signal processing,accuracy of disturbance signal identification,and systematicity of engineering applications.As a result,part of the potential user community is still on the sideline.To overcome these technical issues and move the system beyond its current stage as a laboratory novelty,the present work is undertaken to improve the overall performance of ?-OTDR system and promote the practical development of its engineering application,while carrying out related research work.The main research contents of this dissertation are as follows:(1)Established selection guidelines and design requirements for key modules in the ?-OTDR system,and provided necessary conditions for the functional realization of high-performance ?-OTDR system.Based on a general ?-OTDR system mathematical model,sensing principle and typical structure,combined with the actual needs of building the system from practical experience,the key modules of a ?-OTDR system,such as narrow-linewidth laser light source,pulse modulator,optical amplifier,optoelectronic balanced amplification detector,etc.,are analyzed from the standpoint of the working principle,performance analysis and contrast,to provide a reliable basis for researchers who can make a reasonable choice when faced with the task of module selection.To achieve the requirements of higher precision acquisition and more rapid transmission of the sensor data which have characters of high data volume and high real-time performance,the selection of ADC chip and the designation of anti-aliasing filter and DMA were fully taken into account.A Large cache PCI high-speed data acquisition card with a 12 Bit acquisition precision,125Ms/s acquisition velocity and 133M/s transportation velocity was designed successfully.To guarantee the achievement of the system functionality and decrease the digital signal processing burden at the same time,a mixer with low frequency conversion and an active low-pass filter and a band-pass filter with better amplitude-frequency and phase-frequency characteristics were designed.To supply power in a centralized way in while decreasing the influence of ripple noise on each function module,a structure with multi-path step-down voltage stabilizer was adopted and a multi-stage ?-shaped filter is used in the voltage conversion process to reduce ripple noise.A multiple power supply module with overall voltage ripple level less than 3 m V is achieved ultimately.(2)Eliminated the influence of initial phase random fluctuations and random frequency drift in the system on the demodulation result of system,improved the structure of ?-OTDR system,through a novel single-source dual heterodyne detection scheme.At the same time,the phase demodulation algorithm was optimized to provide sufficient conditions for the pattern recognition of disturbance signal.In the traditional ?-OTDR system,the pulse chopping signal and high frequency modulation signal have different clock sources,leading to random fluctuation of the initial phase of the probe light and have a serious effect on the optical fiber sensing system which detects the disturbance signal based on phase demodulation.Hence,we redesigned the acousto-optic modulator driver and direct digital frequency synthesizer(DDS),used the clock homology signal generated by the DDS for clock synchronization control of the chopper module,frequency modulation module and TTL external trigger module of the data acquisition card;A new type of of ?-OTDR system based on a novel single-source dual heterodyne detection scheme is proposed to solve the frequency drift of the narrow-linewidth laser and frequency drift caused by crystal diffraction rate changes of the acousto-optic frequency shifter(AOFS)due to long-term running temperature drifts.The heterodyne subsystems are composed of mixing the local oscillator with the probe light modulated by AOFS containing the FM information.The frequency drift in the system is followed,while taking the output of the interference light after photoelectric conversion as the local oscillator reference signal in the process of electric-mixing,effectively suppressing the emergence of spurious frequency in the demodulation results caused by the frequency drift.Through the improvement of the above two aspects,the accuracy of phase demodulation and frequency demodulation of the system is improved effectively.In addition,time-domain quadratic phase unwrapping is introduced to the original phase unwrapping method to improve the response consistency to the disturbance intensity of the system.In the experiments,the pulse repetition frequency of the ?-OTDR system is 4 k Hz,the pulse width is 80 ns,the rising edge / falling edge is less than 6 ns,and the length of the sensing fiber is about 24.6 km.The frequency response range of the measured system is 5 Hz to 1.715 k Hz,the spatial resolution is about 9.2m,the signal-to-noise ratio is greater than 23.5 d B,the phase intensity response is consistent and has a good linear relationship with the amplitude of the disturbance intensity,and the linear fitting coefficient is 0.9215.(3)A time-frequency domain signal analysis method based on MEEMD-Hilbert transform is proposed for the feature extraction of non-stationary random disturbance signals,which improves the efficiency of signal feature extraction.At the same time,the improvement of the feature extraction saves a lot of time for training and classification work of feature classifier which enhanced the system ability of real-time response.Different analysis methods include time domain analysis method;frequency domain analysis method and time-frequency domain analysis method of original signal are studied and compared,resulting in the conclusion that the time-frequency domain analysis method is more suitable for the feature extraction of the non-stationary random disturbance signal in the sensing fiber.In addition,two kinds of signal time-frequency domain analysis methods,including wavelet packet decomposition method and Hilbert yellow transform HHT method are studied intensively.The EMD method of empirical mode decomposition and the related improved methods EEMD and CEEMD in HHT were improved.In the process of signal decomposition,the entropy signal anomaly evaluation mechanism is introduced to make the introduction of random white noise reach the best level.By contrast and analysis of simulation,the MEEMD method has ability in solving the inherent algebraic problem of EMD method and the stochastic pseudo component problem of EEMD and CEEMD.In addition,it also improves the signal decomposition effect and reduces the natural modal decomposition time.Then,the time-frequency signal analysis method based on MEEMD-Hilbert transform was applied to the feature extraction of 1500 time series of three kinds of disturbance events including continuous trampling,heavy fall and continuous catenary.The BP neural network is used for feature training and feature classification.The result came out that the accuracy of the final disturbance signal recognition is 98.3%.The performance of feature extraction,feature extraction efficiency,feature training efficiency,feature classification efficiency and frequency adaptability are obviously better than wavelet packet decomposition under the same experimental conditions.(4)Taking underground pipe gallery in the construction of smart cities as technology application's proving ground;the performance and engineering practicality of the new ?-OTDR system based on single-source dual heterodyne detection scheme are verified,providing a complete standard design reference for engineering application of ?-OTDR system in other application fields.Combined with the experience of practical engineering application,according to the hierarchical structure,the concrete realization of ?-OTDR system in Heng Qin underground pipe gallery is divided into physical perception layer,data processing layer,logic processing layer and business presentation layer.According to the content of hierarchical deployment,the selection basis and laying methods of vibrating sensitive cable,the hybrid programming of upper system software and the designation of mechanical packaging and realization of GIS were introduced respectively.The upper system software is implemented by multiple programming techniques including NI measurement Studio,Visual C++ and Matlab Engine.The adoption of techniques is decided by the advantages of each programming technique.NI measurement studio are capable of drawing plots with large amount of data rapidly and the visualization is friendly,Visual C++ are capable of processing data rapidly in multithreaded way and Matlab Engine have absolutely advantages in the designation of data processing algorithm.The combination of the three techniques greatly improves the speed of the development works and enhances the real-time response ability which will be a useful guidance for the development of other relevant industrial control software.A geographical information system based on the web techniques and Cloud GIS techniques is designed and implemented.The intrusion alarm data stored in the database with a unique format can be presented in real time.The presentation visualization views are improved by Data Table plug-ins which can provide retrieval and statistical analysis methods for use,providing a good user experience with the system and improving the management level of underground pipe gallery.The system acquires a relatively accurate test result and the recognition accuracy of the intrusion signal has reached 98.8%. |
PDF Full Text Request