Research Of Baseline Correction Method Based On Laser-induced Breakdown Spectroscopy

Laser-induced breakdown spectroscopy(LIBS)is a promising detection technology of atomic emission spectroscopy developed in the 1960 s.LIBS technology has become a hot research topic due to its characteristics,such as no sample preparation,hardly no damage to samples,high sensitivity and low detection limit.Based on the national defense preresearch project ‘* * * research on rapid detection and identification technology',in order to improve the spectrum of LIBS and the accuracy of quantitative analysis,the baseline correction methods were discussed in this thesis.The main content is as follows:(1)The factors which are connected with LIBS spectral data are discussed in detail through studying the history of LIBS and literature review,and the parameters of the instrument in the LIBS system built by lab are adjusted.The plasma will continue to radiate light outward after exited,so the best observation time of the plasma spectrum can be obtained by adjusting the acquisition time of the spectrometer;the best plasma spectral intensity is obtained during the dual-pulse generation by adjusting the time interval between two laser pulses.(2)By summarizing the baseline correction methods of other spectra,the basic principle and technical characteristics of baseline correction methods are clear.Aiming at the problem of periodic arch protrusion in laser-induced breakdown spectrum,an improved baseline correction algorithm is proposed.The algorithm is based on the penalty least square method.By changing the local penalty coefficient,it is possible to fit the periodic arch protrusion and reduce the influence from analyze line.Compared with other baseline correction methods,the root mean square error of the algorithm in fitting the simulated baseline gets smaller,and the correlation coefficient of the calibration curve of Barium whose data is corrected by the algorithm reaches 0.9972.The results show that,compared with the other baseline correction methods,the baseline of spectral data is well fitted through the new algorithm while the effective information of spectral data is still preserved.(3)Five different approaches are integrated into the analysis software designed by the laboratory,such as IAs LS,As LS,as PLS,air PLS and optimized approach.Combined with neural network,the effects of different baseline correction methods on neural network training model and prediction results are compared.The experimental results show that the baseline correction of spectral data can improve the stability of neural network training model to a certain extent.Besides,it is verified that the baseline correction plays a positive role in neural network quantitative analysis.As was stated above,this thesis aims at improving the accuracy of LIBS quantitative analysis,the study is involved from the LIBS system hardware parameters to the baseline correction algorithm.The results of this study can provide some valuable references for LIBS data processing.