The Study On Quantitative Method Of Ceramic Raw Materials Based On Laser-induced Breakdown Spectroscopy

In the ceramic production process,the content of Si,Al,Mg,Fe,Ti and other elements in the ceramic raw materials has an important impact on the quality of the ceramic products.Laserinduced breakdown spectroscopy(LIBS)is an atomic emission spectroscopy technique used to analyze the composition of elements.The technology has the advantages of rapid analysis,remote analysis,simple pretreatment,and simultaneous detection of multiple elements.The premise of constructing a nonlinear model between high-dimensional spectral data and element content is a large number of sample data,because the composition and structure of ceramic raw materials are complex.The limited number of standard samples has seriously hindered the development of this technology in practical applications.Therefore,it is of great practical significance to study how to bulid quantitative models of key elements in ceramic raw materials based on fewer samples.On the basis of reading relevant literature,this thesis proposes a feature extraction method combining linear regression(LR),sparse under-complete autoencoder(SUAC)and neural network,aiming at the characteristics of low sample size and high dimension of spectral features of standard ceramics.The spectral features can be weighted and fused into more representative abstract features.The spectral data dimension is reduced from 8188 to 100 through the LR layer,and further reduced to 32 through the SUAC encoding layer.Further,a quantitative analysis model for the elemental composition of ceramic raw materials is established by the combination of LR+SUAC and BPNN.Since the input data dimension and redundant information are greatly reduced by LR+SUAC,the overfitting problem of BPNN is greatly reduced.The results show that LIBS technology combined with LR+SUAC+BPNN quantitative analysis method can be used for multi-element detection in ceramic raw materials.For the needs of the application and promotion requirements of the same type of equipment in different production lines,in order to reduce the labor burden of repetitive modeling,model maintenance and update for multiple equipment,this thesis proposes a flexible window piecewise standardization algorithm,which can enable different LIBS equipment to share the trained quantitative analysis model.The method reduces the RMSE between spectral lines at peak positions and improves the accuracy of model transfer,compared to piecewise direct normalization algorithms.The influence of the number of transferred samples and the number of spectral lines of the slave machine on the accuracy of the fitting are further analyzed.The experimental results showed that the flexible piecewise standardization algorithm is a feasible model migration method for quantitative analysis of ceramic raw materials.