Study On Data Processing And Analysis Method Of Spectral Measurement

This paper is based on the data processing and analysis of spectral measurement.The common spectral preprocessing methods are studied from four aspects.that are spectral data screening and processing,the spectral data denoising,eliminating the influence of spectral line overlap,solid particles and other factors and calibration of spectral characteristic lines.A wavelet adaptive threshold denoising method and the selection of spectral data de-noising evaluation index are proposed.The main contents of the paper are as follows:The first chapter that is the introduction introduces the research background,the development of spectroscopic technology,the structure of spectrometer,and characteristics of spectral technology.And then introduces the research tools,the research content of this topic,and then discuss the research tools,the research content and the innovations.The second chapter summarizes the existing spectral preprocessing methods,and analyzes the spectral pretreatment methods from four aspects: data selection,noise filtering,elimination of various factors and calibration of characteristic lines.The data selection algorithm includes the analysis and treatment of abnormal data and the data enhancement algorithm to increase the diversity of data.Noise filtering algorithm includes smoothing algorithms,Fourier transform and wavelet transform.The algorithms to eliminate various factors include derivative algorithm,SNV and detrending algorithm and multiple scattering correction algorithms.The algorithm of characteristic spectral line calibration includes the automatic peak seeking of spectral lines and the calibration of the corresponding elements in the spectrum line.The third chapter studies the wavelet denoising method based on the spectrum signal processing.Firstly,the principle of wavelet denoising method is introduced.Then the three commonly used wavelet denoising methods are summarized.They are denoising method of wavelet transform modulus maxima,denoising method of wavelet coefficient correlation and wavelet threshold denoising method respectively.Next,the wavelet threshold denoising method is studied emphatically.The Wavelet threshold denoising is divided into forced threshold denoising,default threshold denoising,given threshold denoising and adaptive threshold denoising.In this paper,wavelet adaptive threshold denoising is further studied.The most critical part of wavelet threshold denoising is threshold selection.Therefore,a new threshold selection method is proposed in this paper.The genetic algorithm is applied to threshold selection in wavelet adaptive threshold denoising.The genetic algorithm for threshold selection of wavelet adaptive threshold denoising is very much obvious advancer than the existing four threshold selection methods by the experimental verification.The fourth chapter mainly studies the evaluation index of de-noising.The existing denoising quality evaluation methods include mean square error,signal to noise ratio,smoothness correlation coefficient and so on.Based on the existing denoising evaluation index,this paper puts forward a comprehensive evaluation index Q for the evaluation of spectral signal de-noising effect,which is an evaluation index based on signal-to-noise ratio and peak relative error.The result of experiment shows that this evaluation index has certain advantages and stability and can be used for the evaluation of spectral signal denoising effect in the future.The fifth chapter summarizes the research work and the outlook for the future.