Research On Analysis Methods Of Some Metal Elements In Iron Ore Based On LIBS Technology

Steel production plays an important role in the country's economic development and is a pillar industry of the national economy,while iron ore is the most important raw material for the production of steel.The quality of its quality,to a large extent,affects the smelting process and the economic benefits it can produce.Traditional elemental analysis methods take a long time and require complex sample preparation.There is an urgent need to find a technical method that can be quickly and quantitatively analyzed.Laser Induced Breakdown Spectroscopy(LIBS),which does not require complex sample pretreatment,in-situ,real-time and other advantages,provides an effective method for rapid analysis of iron ore.However,LIBS spectroscopy has a large number of high-dimensional,complex characteristics,which seriously affects the accuracy of quantitative analysis.Chemometrics can provide an effective tool for data analysis of complex high-dimensional LIBS spectra.In this paper,starting from the actual needs of rapid analysis of iron ore,the research on quantitative analysis methods of iron ore based on LIBS spectrum is carried out,focusing on the quantitative analysis methods based on RF and PLS to improve the accuracy of LIBS technical analysis results.Mainly conducted the following research:(1)Taking iron ore as the research object,establish a quantitative analysis method of Fe element in iron ore based on LIBS technology combined with partial least squares(PLS)algorithm.First,based on the LIBS technology,the 16 iron ore samples prepared were spectrum collected,then the LIBS spectrum was preprocessed based on different pretreatment methods,and finally the PLS model was constructed based on the optimal input variables to achieve the determination of Fe elements in the ore Compare with the predicted performance of the RF correction model.The results show that the PLS correction model has the best prediction performance for the Fe element in iron ore.(2)Taking iron ore as the research object,establish the determination method of Al,Mg and Ca in iron ore based on LIBS technology combined with WT-RF method.First,preprocess the LIBS spectrum based on wavelet transform,then filter the processed spectrum based on variable importance,and finally build an RF correction model based on the filtered variables to achieve quantitative analysis of Al,Mg,Ca elements in iron ore,And compared with the predicted performance of the PLS correction model.The results show that the RF model has the best prediction performance for the three elements.(3)Taking iron ore acidity as the research object,an iron ore acidity determination method based on LIBS technology and RF algorithm was established.Firstly,based on cross-validation,the best pretreatment method is determined as the second derivative.Then,the pre-processed spectrum is screened based on the importance of variables.Finally,an RF correction model is constructed based on the optimal input variables to achieve iron ore acidity Measured and compared with the predicted performance of the PLS calibration model.The results show that the RF model based on variable importance extraction has the best prediction performance.