Study Of Sagnac/?-OTDR Hybrid Speech Sensor And Corresponding Speech Denoising Method

Optical fiber sensing technology has been one of the most active branches in sensing technologies due to its advantages of strong electromagnetic interference resistance,small environmental impact,corrosion resistance,high sensitivity and light weight.Traditional electrical speech sensors cann ot be used in extreme environment such as strong electromagnetic interference,humidity,corrosion and so on,thus many optical fiber speech sensors have been proposed.The detecting range of optical fiber speech sensors is usually up to tens of kilometers,thus speech source location is necessary.However,existing optical fiber speech sensors can hardly locate speech source accurately,and the location method is usually complicated.Besides,speech signal is inevitably subject to noise interference in the transmition,thus speech denoising is also necessary in order to improve the quality of detected speech.Under this background,this paper desig ned Sagnac/?-OTDR hybrid speech sensor with full optical ends and corresponding speech denoising method,such sensor can detect and locate speech at the same time.Aiming at so lving the problem of existing location method that brings extra error or has com plicated calculation,algorithm based on time-domain variances was proposed,which is able to locate speech accurately.In addition,since it is necessary to denoising detected speech,a modified wavelet packet speech denoising method was designed.Experiments proved that Sagnac/?-OTDR hybrid speech sensor can detect speech successfully,the detected speech sounds clear and has some background noise.Noise can be removed after denoising noisy speech by the modified wavelet packet speech denoising method.Moreover,the system can locate the speech source,and the location error is only 30 m in the experiment.The structure of this paper is as follows.In chapter one,development of optical fiber sensing technology was introduced,several typical types of fibe r optic vibration sensors and corresponding location method were described,and development optical fiber speech sensors was also introduced.In chapter two,the principle of optical interference and phase modulation of light wave were analyzed,based on w hich principle of Sagnac fiber interferometer,signal demodulation method and location method of ?-OTDR vibration sensor was elaborated,then Sagnac/?-OTDR hybrid speech sensor was introduced,and speech demodulation method and speech source location metho d were also described.In chapter three,principle of wav elet analysis for speech denoising method was introduced,a modified wavelet packet speech denoising method was elaborated,and denoising effect based on such method was compared with denoising effect based on traditional wavelet packet denoising method.In chapter four,experimental device of Sagnac/?-OTDR hybrid speech sensor was build up,and a segment of speech was collected,effect of speech sensor was analysed through speech waveforms and spectrograms after demodulation.Then detected speech was denoised by the modified wavelet packet speech denoising method,and denoised speech has better auditory effect.Finally,speech source was located by proposed method,and the location error was only 30 m.In chapter five,innovation and deficiency of this pap er were summarizes.