Research On OTDR Automatic Test Technology

With the development of optical fiber communications,the trend of 5G era and fiber to the home,which make the amount of optical fiber has increased significantly,and the laying situation has become more complicated.Optical Time Domain Reflectometer(OTDR)is widely used as a single-ended nondestructive optical fiber measuring instrument.Traditional OTDR requires experienced technician to perform parameters setting and curve analysis,which is inefficient and manpower-consuming.Therefore,it is important practical significance to improve the automation of OTDR testing.In view of the above problems,this paper studies a testing technology solution that can automatically select emission parameters and conduct curve analysis,and on this basis,studies a testing technology solution that can automatically select multiple emission pulse widths and identify event types.First,the original data is preprocessed is preprocessed for subsequent analysis.Logarithmic transformation is used to increase the readability of the data,and then the accumulation is used to improve the signal-to-noise ratio initially.The improved wavelet threshold filtering method is used to denoise the signal again,and finally output the preprocessed data.Three traditional event detection methods are discussed,and finally the wavelet transform method is selected as the basis of event detection in the subsequent scheme.Then,the OTDR automatic testing scheme is studied.Facing an unknown optical fiber link,a method for estimating the endpoint of the optical fiber based on probability distribution function was first designed to determine the launch pulse width and test distance.Then the signal is detected.Due to the difference between reflection and nonreflection events,a method combining wavelet transform with cell average constant false alarm rate(CFAR)is proposed for reflection events detection,and a method that combines wavelet transform with 3d B principle is proposed for non-reflective events detection.After the events are detected,the least square method is used to calculate the event parameters.In addition,aiming at the limitation of OTDR fixed amplification channels,a multi-channel synthesis algorithm based on multiple linear regression and named KGSD(K-means Gap statistic Standard Deviation)is proposed to improve the performance.The experimental results indicate that the multi-channel synthesis algorithm improves the dynamic range of the testing curve and is more conducive to event detection,and the effectiveness and accuracy of the scheme are proved by the experimental data.Finally,the multi-pulse width automatic testing and identification scheme is studied.Aiming at the contradiction between pulse width and resolution in single pulse width detection,a multi-pulse width selection method is proposed to obtain more complete link event information.To recognizes the specific types of events on the link,firstly extract the multi-dimensional features of the events through time domain analysis,wavelet analysis and wavelet packet analysis,then the Fisher criterion is used to evaluate the quality of the features,and the method to improve the recognition accuracy by optimizing the parameters of support vector machine classifier is discussed.Through experiments,the optimal feature dimension is selected,and the particle swarm optimization algorithm is finally selected as the optimization method in the genetic algorithm and the particle swarm optimization algorithm.The experimental results indicate that this scheme can effectively identify the type of link events.