Research On The Signal Demodulation And Recognition Technology Of Interferometric Fiber Vibration Sensing

Vibration sensors are widely used in petroleum exploration,groundwater detection,civil engineering,and other fields,and have extremely important practical significance for homeland security.Because of the advantages of high sensitivity,wide bandwidth,large dynamic range,and large-scale array sensing,the interferometric optical fiber vibration sensing technology has been widely used in vibration monitoring.The signal demodulation technology directly restricts the performance of the interferometric optical fiber vibration sensors.How to develop large dynamic,high-sensitivity,and high-performance embedded optical fiber signal demodulation software and hardware is still one of the difficult problems to improve the performance.Besides,vibration signal recognition technology directly affects the practical application of optical fiber vibration sensors.But the existing signal recognition methods cannot effectively deal with actual multi-vibration aliasing scenarios.How to solve this problem is a difficult problem in the further application of optical fiber vibration sensors.Therefore,this paper focuses on the signal demodulation technology and the vibration signal recognition technology.The corresponding solutions are proposed for some problems in the existing technology or algorithms.This paper provides support for the practical application of interferometric fiber optic vibration sensors from algorithm research to engineering realization.The main research contents are as follows:Firstly,the non-zero optical path difference of the interferometer during the production of the optical fiber sensor causes increased phase noise.Therefore,it is of great significance to develop a large-range,high-precision,non-destructive interferometer optical path difference measurement method.To solve this problem,a frequency scanning interferometer method based on a variable optical path difference reference interferometer is proposed.On this basis,an interferometer optical path difference measurement system is designed and built,an full-phase Fourier transform method is used to calculate the fractional period.Because the reference interferometer optical path difference that cannot be measured accurately and changes with temperature in phase comparison method,the proposed method breaks through the measurement accuracy limitation.In addition,it improves the measurement accuracy of the short optical path length.Test results show that the proposed measurement method for the optical fiber interferometer achieves a relative uncertainty of 4.9×10^-6 in the measurement range of 75 cm～70 m.And it provides a guarantee for further suppressing the influence of the optical path noise on the minimum detectable signal of the system in the developing of optical path.Secondly,the limitation of CPU architecture in existing demodulation hardware causes insufficient processing capacity and high power consumption.This paper comprehensively considers the processing real-time performance,hardware complexity,and other indicators,and proposes a large dynamic range,high sensitivity digital demodulation technical scheme.Moreover,signal demodulation hardware including low-noise photoelectric conversion board,high-speed acquisition board,real-time signal processing board,is independently developed.Aiming at the problems of light intensity disturbance,phase modulation depth C value drift and random fluctuations,ellipse fitting algorithm is used to suppress nonlinear distortion.Then C value is estimated through ellipse correction parameters and controlled by local feedback loops to the optimal working value to further suppress nonlinear distortion.In view of the carrier phase delay problem caused by the optical path and circuit,the phase delays of the fundamental wave and the second harmonic term of the interference signal carrier are calculated based on the orthogonal lock phase method.And then the instantaneous phases of the local carriers are adjusted online through the direct digital frequency synthesizer to achieve the purpose of phase delay compensation.Test results show that the dynamic range of the developed PGC digital demodulation hardware is better than 175 d B@220Hz,and the phase self-noise is better than-110d B rad/√Hz.Thirdly,the accuracy and robustness of existing optical fiber vibration signal recognition methods needs to be improved urgently in practical application scenarios.This paper proposes an aliased fiber vibration signal recognition method based on the deep complex network.The fiber vibration signal is converted to the frequency domain through Fourier transform.The form,mode,and phase of the vibration signal in the frequency domain are consistent with that of the neural network in the complex number domain.The characteristic region of the fiber vibration signal is extracted in the frequency domain through the window-sliding in the frequency domain.Then each part of the segmented signals are recognized based on the deep complex network.Finally,a weighting vote algorithm is introduced to identify the signal type and quantity in the aliased signal.Test results show that the proposed method can recognize single fiber vibration signals,pairwise aliasing signals,and multi-aliasing signals corresponding to four events of knock,vibration,walking,and vehicle movement.The recognition accuracy rates are 0.989,0.969,and 0.922,respectively.It provides a solution for the recognition of vibration signals from multiple vibration sources.