Composition Analysis Of Iron Ore Raw Materials Based On LIBS

The iron and steel industry is one of the core industries of China’s national economy.Iron ore,as the main raw material of iron and steel,directly affects the smelting efficiency of the iron and steel industry,and some metal and non-metallic elements contained in iron ore will have different effects in the smelting process.Therefore,it has become a very important topic to detect the composition content of iron ores,reasonably realize the optimization of ore blending,reduce coke ratio and cost.Compared with the traditional ore analysis methods,laser-induced breakdown spectroscopy(LIBS),as a new atomic spectrum emission technology,has the characteristics of fast,real-time,online,all element analysis,etc.It is very suitable for online detection in the industrial field,and has been widely used in many disciplines.The accuracy of LIBS quantitative analysis is affected by many factors,such as the setting of system parameters,the characteristics of the sample itself,some interferences in the spectrum,and the selection of quantitative analysis models.For how to improve the accuracy of LIBS in the analysis of iron ore composition,this thesis has made the following work:(1)The internal principle of the LIBS system used in the experiment is introduced,and the effects of turntable speed,single and double laser pulses and sample moisture on spectral signals are investigated.The signal-to-noise ratio(SNR),signal-to-back ratio(SBR)and relative standard deviation(RSD)were used as evaluation indexes to select the optimal parameters.At the same time,the setting of other parameters of the system is also explained.(2)Several basic quantitative analysis models commonly used in LIBS and their mathematical principles are described in detail.The quantitative results of each model applied to iron ore raw material samples are compared to analyze the problems of each model and provide theoretical support for the next step of model improvement.(3)Due to the large amount of background and noise contained in LIBS spectrum,the accuracy of the above basic models needs to be improved when the full spectrum information is used for quantitative analysis.Therefore,we apply a PLS model which combines ridge regression cycle to select characteristic spectral lines.In the quantitative analysis of a certain element,some characteristic spectral lines related to the element are selected from the spectral data by Ridge regression model,and then quantitative analysis is carried out by PLS model to improve the accuracy of the results.The quantitative results of the model on Fe,Si,Ca and Mg elements were compared with those of the previous basic models.The RMSE of the test set and the regression coefficient R~2 of the model were used to evaluate the effectiveness of Ridge regression+PLS model in the analysis of iron ore raw materials.(4)The moisture contained in iron ore raw materials will affect the accuracy of LIBS quantitative analysis.In the experiment,the relationship between sample moisture and spectral intensity was explored by preparing samples under different moisture levels,and the standard moisture interval of the sample was determined according to this relationship.Using the spectrum within the standard moisture interval,the quantitative analysis model of four elements was rebuilded by the method of ridge regression combined with PLS model,and the spectrum outside the standard moisture interval was quantitatively analyzed with this model.It was found that moisture had a serious impact on the stability and accuracy of the results.Based on this question,a spectral standardization method based on the correlation ordering of spectral line intensity is proposed,which corrects the spectra in different moisture intervals to transform them into the spectra in the standard moisture intervals,so as to correct the influence of moisture on quantitative analysis.Compare the results of this method with the results of common used PDS＿PLS,it shows that SICMS method can jump out of the problem of"window limitation"of PDS＿PLS method,which enables more accurate spectral standardization.