| Aerosols are the main form of propagation of atmospheric pollution,toxic heavy metal elements contained in aerosols are the culprit causing environmental pollution and human health hazards.New technologies are urgently needed for the detection of elements in aerosols.Laser-induced breakdown spectroscopy(LIBS)is a highly sophisticated new spectral analysis technology.It uses a pulsed laser to strike a sample to produce a plasma,and analyzes the collected plasma spectrum to achieve qualitative and quantitative elements analysis.This paper takes the requirements for the detection of heavy metal elements in atmospheric aerosols as the background,and focuses on the development process of LIBS analysis system,based on the previous work of the research group,we studied the issue for peak identification,quantitative models and software systems development of LIBS system.The main research work of this paper are as follows:(1)On the basis of introducing the research background and significance of the subject,this paper expounds the basic principles and development status of LIBS technology and analyzes and summarizes the key technologies of current LIBS system development,that is,spectral peak detection,quantitative models,and analysis software,and launched a review of its research status at home and abroad.(2)In order to improve the accuracy and automation of LIBS peak detection,a model based on scale adaptation of continuous wavelet transform(SA-CWT)is proposed.The algorithm is based on the idea of quadratic optimization.On the basis of continuous wavelet transform,the scale range is automatically adjusted by introducing a scale adaptive model to improve the degree of automation of the program.and two times continuous wavelet transforms are used to improve the recognition performance of weak and overlapping peaks.The experimental results show that compared with the gaussian curve fitting method and ridge peak finding method,SA-CWT has significant advantages in overlapping peak resolution and weak peak detection.The error between the peak detection result and the NIST standard spectral library is about 0.0143 nm,and the accuracy of peak detection is high,the algorithm can be applied to LIBS data processing.(3)Aimed at the application of machine learning algorithms to quantitative analysis of LIBS,and based on the introduction of the concept of kernel extreme learning machine(KELM),a PCA-GA-KELM algorithm model is proposed by combining with the advantages of principal component analysis(PCA)and genetic algorithm(GA),which uses PCA to complete the dimensionality reduction process,and uses the GA algorithm to determine the regularization coefficient C and kernel parameter s of the kernel extreme learning machine,and improved the recognition accuracy and automation of algorithm.Comparative experiments show that it has a certain effect on the prediction of LIBS element concentration.(4)The software designed and developed by the name of "LIBS spectral data analysis and processing system",which combined with the LIBS hardware platform designed by the research group to form a fully functional LIBS analysis and detection system,and it achieves the purpose of LIBS analysis software autonomy.At the same time,the effectiveness of the software system was verified by comparison with the foreign professional LIBS analysis software.In summary,based on the development of LIBS detection system,focuses on the research of spectral peak identification and qualitative and quantitative analysis,combines with wavelet analysis and machine learning theory,this paper proposes improved algorithms and has contributed its power for the development of LIBS.Besides,the software systrm has also promoted the progress of domestic and independent of LIBS analyzing software. |
PDF Full Text Request