On-line Quantitative Research Of Powder Mineral Based On Laser-Induced Breakdown Spectroscopy

Powder,as a very common physical form in the production and processing of industrial sites,often requires to detect the component,such as the cement raw material powder in the cement production and the coal powder in the fire power industry,etc.To give helpful evidence and hints for the industrial production process,to further save the resource and energy and product the environment,there is a demand for a fast and online method for the elements' analyzing.At present,the conventional detection methods for coal and cement are not conducive to the sustainable development in the future for the reason of long analyzing period,limited elements can be detected,and the disadvantages of radiation sources.Laser-induced breakdown spectroscopy,LIBS,as a promising chemical analyzing technology,has the advantages of no or simple need foe sample preparation,fast analysis,end the ability to analyze almost all elements.Due to these advantages,LIBS has great potential and application value in more and more fields,and also includes the powder cement and coal.In this thesis,we introduce the LIBS system which is optimized to online analysis of in industrial field,and design the experiment.In our experiment,the powder sample in the two-dimensional removable box are detected directly,and a fan putted above the sample is aimed to blowing away the flying sample.Then we optimize the laser energy,pulse frequency and the delay time to the best state for the detection of powder cement and coal.In the analysis of cement powder,the univariate model of Ca,Si,Al,Fe and Mg in cement are first established using the row and column normalization to preprocessing the original data,and however,the results are not satisfactory.Then,the pretreatment method was improved,and we use the intensities of two characteristic lines with the wavelength at 315.89nm and 317.89nm of Ca as the internal standard to do the normalization,meanwhile compared with row-by-row and column-by-column normalization.Next we build multivariate quantitative analysis models which based on partial least squares(PLS)and support vector regression(SVR),and these two models are used to be compared.The results show that when the internal standard normalization combined with SVR quantitative analysis,the model has a higher precision for the analysis of cement.Finally,the comparison between tablet-type cement and powder cement is carried out to further prove the feasibility of using LIBS technology to measure powder cement directly.For the analysis of C and ash in coal,we use the method consistent with the cement analysis.Firstly we build the univariate method for C in powder coal and prove the shortcomings.Then we choose the characteristic line intensity at 247.9nm of C element as the internal standard,next compare the preprocessing methods and model kinds described in the analysis of cement.Finally,the type of detecting the powder coal directly also be demonstrated.Through the quantitative analysis of the cement and coal components in this thesis,it is verified that the powder state of both samples can use the LIBS technology to detect directly,as for the difference of the adapted methods between the two samples is that cement and coal powder have different properties,and there is no common method to achieve the best results for both samples simultaneously.