Research On Spectral Quality Feature Recognition Method Based On Weak Grating Array

In order to solve the quality problems such as broken and leaked gratings during the online preparation of fiber Bragg gratings,and to prevent mismonitoring caused by poor sensing performance when fiber Bragg gratings are applied in fields such as tunnels,power,and border defense.This paper focuses on the quality defects in the weak grating array spectrum.Starting from analyzing the sensing mechanism and key technologies of fiber Bragg grating,exploration and research are carried out on two key issues: grating spectral quality feature recognition method and grating spectral quality evaluation,in order to achieve quality detection of large capacity grating arrays.The main work of the paper includes the following aspects:Firstly,based on the research and analysis of grating spectral quality detection technology both domestically and internationally,in order to achieve grating spectral quality detection,a corresponding distorted spectral model is established by analyzing the characteristics and causes of distorted spectrum.Propose an adaptive SVM-DCAS/PS recognition algorithm,optimize the SVM recognition algorithm,and train SVM recognition algorithms with different kernel functions.Analyze and compare the recognition results of four kernel functions.Compared with the other three kernel functions,when using Poly kernel function,the recognition accuracy of grating spectral quality can reach over 99%.Then,the DCAS/PS correction algorithm was used to correct the distorted spectrum,and the validation results showed that this method has a good correction effect on the distorted spectrum.Secondly,this paper also improves the quality evaluation system for grating spectrum.In this paper,the performance of adaptive SVM-DCAS/PS recognition algorithm under different kernel functions is compared and analyzed by using Root-mean-square deviation,regression line and statistical absolute coefficient from the traditional grating spectral quality evaluation method,and good results are obtained.Finally,the DFA algorithm was cited for simulation verification of four kernel functions.When using the adaptive SVM-DCAS/PS recognition algorithm with Poly kernel function,the highest slope can reach 1.9701,with the best performance.Finally,this paper designed and built a large capacity weak grating spectral quality feature recognition system experimental testing platform to verify the feasibility of the proposed grating spectral quality feature recognition method.In the testing temperature range of 0℃ to 120℃,the adaptive SVM-DCAS/PS recognition algorithm was used to identify and correct the spectral quality of the grating array at 120℃,and the trend of the spectral characteristics of the grating array with temperature changes was analyzed.The experimental results show that the method achieves a spectral quality recognition accuracy of over 99.6% for grating arrays,and the effect of correcting distorted spectrum is obvious,with a fitting degree of over 99%.It has certain application reference value for the spectral quality detection of grating arrays.