Study On The Integrated Pretreatment Of Laser Induced Breakdown Spectroscopy Data

Laser-Induced Breakdown Spectroscopy(LIBS)is an emerging and promising atomic emission spectrometry detection technology.The detection speed is fast,no sample preparation is required,and simultaneous analysis of multiple elements is the biggest advantage of LIBS technology.It has broad application prospects in industrial production,environmental monitoring,biomedicine,space exploration and military fields.This thesis focuses on the comprehensive pre-processing technology of LIBS raw spectral data with the background of the defense pre-research project“Detection and Identification Technology of ***”(41426040102).The main research work of this paper is as follows:(1)This paper first introduces the research background and significance of the subject,and elaborates the basic working principle and development process of LIBS technology,analyzes and summarizes the key technologies in the current LIBS data preprocessing,including spectral noise reduction,baseline correction,and overlap.Current status of research at home and abroad such as peak identification.(2)Analyze the original spectral noise collected in the LIBS system in detail,and propose the lifting wavelet transform for LIBS noise reduction processing for its noise characteristics.Using the optimal root mean square error and entropy analysis method to determine the optimal lifting wavelet basis function and lifting the decomposition layer number,the least squares fitting method is used to fit the noise lifting wavelet coefficient in the upper and lower threshold intervals of the double threshold method.The noise is suppressed,and the threshold calculation method is optimized according to the difference in the number of decomposition layers.The improved algorithm guarantees the continuity of the threshold function and the high order guidance.The LIBS raw data of atmospheric aerosol heavy metal elements deposited in the membrane were studied by soft and hard threshold method and improved noise reduction method.The noise reduction results show that the noise reduction effect of the improved algorithm is better than the soft and hard threshold method.Great improvement.(3)Studying the composition of the original spectral signal of LIBS and the continuous background source causing the baseline drift.The correction error of the spectral line caused by the influence of the base effect is large when the blank sample subtraction method is used to correct the baseline.On this basis,the wavelet transform is adopted respectively.Two baseline correction methods are fitted to the segmentation minimum.In the wavelet transform method,the optimal wavelet basis function,the decomposition layer number and the threshold coefficient selection method are given.In the piecewise minimum value fitting,the optimal segment number is obtained by analyzing the minimum deviation of the original data and the fitted data.The experimental results show that the algorithm adopted in this paper has a more obvious baseline fitting correction effect.(4)Based on the theoretical study of the principle and type of LIBS spectral broadening,combined with the peak function to analyze the formation of overlapping peaks and the principle of peak separation.Based on the Lorentz linear overlapping peak function model,the overlapping peaks in the simulated data and experimental spectral segments are respectively resolved.The peaking results of the simulated data show that the deviation from the original discrete spectrum is smaller and the coincidence is better.It also has a good effect in the peak of the measured data.(5)Set up LIBS spectral data acquisition experimental system,use glass fiber filter to prepare concentration gradient test samples of atmospheric aerosol heavy metal elements,and use the integrated pretreatment method to obtain the original LIBS spectral data,and establish the element content by standard curve method.The calibration curve is compared to a calibration curve established from unpreprocessed LIBS data.The experimental results show that the calibration curve after comprehensive pre-processing of data has a high degree of fitness,which can improve the accuracy of subsequent analysis and detection.In summary,based on the pre-processing work of LIBS spectral data,this paper corrects the random noise,baseline drift and overlapping peaks in the original spectrum,and provides relatively "clean" spectral data for the subsequent LIBS spectroscopy work.It has important practical value for improving the accuracy of qualitative and quantitative detection and analysis of LIBS technology,and has extremely high economic benefits in the engineering application of LIBS.