Study On Spectral Identification Method And Adaptability Of Coal And Rock In Fully Mechanized Caving Face Of Tashan Mine

Coal rock identification is an important prerequisite for realizing intelligent coal caving in fully mechanized caving face.Near infrared spectroscopy（NIR）is a method to identify coal and rock from the essential properties of materials.In this paper,the coal and rock drilling cores of the fully-mechanized caving face of Tashan Mine of Datong Coal Mine Group are drilled and sorted into 27 kinds of coal and rock samples according to the difference of their occurrence positions.The reflectance spectra of 27kinds of core-drilling samples at 1100-2500 nm were obtained by spectrograph,and the mineral composition,element content and industrial component content of the samples were determined by X-ray diffraction,X-ray fluorescence and industrial analysis.Twenty-two kinds of coal and rock samples from Shanxi,Shandong,Ningxia Hui Autonomous Region and Inner Mongolia Autonomous Region were collected to obtain the reflection spectrum curve and industrial composition content of coal and rock.The optical path system is composed of lens group,collimating mirror and halogen tungsten lamp,and the calibration part of light source is composed of cosine corrector and optical path switcher.The spectral acquisition device is made up of the optical path system and the calibration part of light source.Within the scope of the 1100-2500 nm band cosine corrector radiation flux of the original data,and the continuum removal（CR）,standard normal variable transformation（SNV）and multiple scatter correction（MSC）three algorithms after preprocessing of tungsten halogen light source radiation flux data as independent variables,respectively,the 1100-2500 nm wavelength range of whiteboard collected collimating lens reflection radiation flux data as the dependent variable,tungsten halogen light source calibration model is set up.The results show that:（1）Compared with roof rock,the absorption characteristics and absolute slope of the spectral curve of coal seam gangue are smaller,mainly because the relatively large amount of opaque carbonaceous material in coal seam gangue obscures the absorption characteristics of structural water,quartz inclusion water and Al-OH at their characteristic wavelengths.（2）even though the coal matrix（mainly organic matter）composed of C,H,O element is 5.3 times of its mineral,but organic groups had no obvious absorption at near infrared band,and a large number of opaque carbonaceous material to cover the absorption characteristics of the inorganic ions in near infrared wave band,makes the spectral curve of coal is relatively smooth and is on the rise.（3）The coal and rock qualitative identification models based on the sum of the absorption valley depth H total of the coal and rock drilling cores in Tashan Mine,the curve slope K’in the band of 1100-2500nm,the curve slope K’in the band of2250-2500nm and the curve convx H’were established to identify the coal and rock in the mining areas of Shanxi Province,Shandong Province,Ningxia Hui Autonomous Region and Inner Mongolia Autonomous Region,respectively,with the accuracy of90.91%,68.18%,81.82%and 86.36%.Based on the ash content and classification of Tashan drilling core,the SVM regression and classification model can identify coal rocks in Shanxi Province,Shandong Province,Ningxia Hui Autonomous Region and Inner Mongolia Autonomous Region with 95.45%and 90.91%accuracy,respectively.（4）Within the designed detection range of the spectral acquisition device,the irradiance and illumination intensity of the light source decrease with the increase of detection distance.The infrared irradiance varies from 440-1987W*m^-2,and the illumination intensity varies from 144*10²-697*10²Lux.The spectral acquisition area is always within the irradiation area of the light source.Through the field test of fully mechanized caving working face,the spectral acquisition device can obtain the spectrum of coal and rock released to the scraper conveyor,which meets the demand of real-time and accurate spectral acquisition in the spectral identification technology of coal and rock.（5）using continuum removal（CR）after processing of light radiation flux（cosine corrector acquisition）as the independent variables,using the whiteboard the reflected radiation flux（collimating lens acquisition）as dependent variable,to establish the forecast model of whiteboard reflection radiation flux has a better ability of inversion and stability,the model calibration set and prediction set decision coefficient is 0.96,0.97,root mean square error is 114.08,91.88,and the relative analysis of error of 7.92.Effectively reduce the drift trend of reflectivity in two-dimensional space.This paper contains 49 figure,16 table and 155 references.