Study On Identification Technology Of Coal And Rock Based On Hyperspectral

Coal rock identification is based on the identification of coal and rock characteristics,the prominent characteristics of coal and rock differences,the effectiveness and robustness of coal and rock,and determine the accuracy and stability of coal and rock identification.It is the prerequisite and foundation of intelligent coal and rock identification technology to study the difference of characteristic properties of coal and rock and find out the difference of effective coal and rock characteristics.Meanwhile,considering the intelligent application,the recognition algorithm based on the difference of coal and rock is the "brain" of coal and rock technology,select the appropriate algorithm to improve the accuracy and real-time.In addition,the actual working conditions of the environmental factors require high practicality of coal and rock identification technology.Therefore,based on hyperspectral technology,this paper constructs a coal-rock identification technology bench based on hyperspectral spectroscopy.The hyperspectral acquisition experiment of coal and rock is carried out,and the mechanism of coal-rock hyperspectral difference and the recognition Mechanism and feasibility of coal-rock identification technology based on hyperspectral study is analyzed.The identification algorithm based on the difference of hyperspectral characteristics of coal and rock is designed,and the identification of coal and rock is carried out by using the hyperspectral data of coal and rock.The identification experiment based on the influence of environment of working conditions is carried out to explore the practical conditions.The main conclusions of this paper are as follows:1)The broad spectrum of coal and rock has significant difference,including the slope of the whole curve,1400 nm band,1900 nm band,2210 nm band and 2350 nm band,in which the rock curve has strong absorption peak near 2210 nm band and 2350 nm,Near the band and 2350 nm near the band the mine curve does not exist strong absorption peak.2)Analysis of Characteristic Difference of 2210 nm Characteristic of Coal and Rock Hyperspectral:First,there is a strong absorption peak in the band near 2210 nm in the rock hyperspectral curve,and there is no strong absorption peak in the band near 2210 nm in the rock hyperspectral curve;According to the XPS spectrum of coal and rock,after comparing the standard absorption peak,it is found that the aluminum element in the coal sample is in the form of alumina,and the content of aluminum is very small.The aluminum element in the rock sample is in the form of aluminum hydroxide and the aluminum content more;Third,there is a strong absorption peak near the 2210 nm band of the hyperspectral spectrum of aluminum hydroxide.It is shown that the strong absorption peak of the hyperspectral curve in the 2210 nm band is caused by the aluminum hydroxide in the rock,and the aluminum element in the coal is in the form of alumina and has little content,therefore the coal curve does not exist strong absorption peak near 2210 nm.3)Algorithm Design of Coal and Rock Identification: According to the generalized characteristics of the absorption peak,the identification algorithm of the coal and rock based on the absorption peak feature extraction and the recognition algorithm based on the full band matching of the absorption peak are designed respectively.The accuracy of the recognition in the laboratory environment is 100%.Based on the neural network and the absorption peak Data are designed based on the neural network supervised classification algorithm,the accuracy of the identification in the laboratory environment is 100%.Based on the real-time consideration,the simple algorithm is designed and the accuracy of the recognition in the laboratory environment is 100%.4)Impact Assessment of Working Condition on Hyperspectral Recognition of Coal and Rock: 1.For the mixed end element spectrum,the ratio of rock coal to one-ninth can correctly identify the rock samples to meet the actual working conditions.2.The dust concentration has a great influence on the spectrum of coal and rock.When the concentration of rock dust is increased,the mixed spectral curve is approached to the rock curve.When the sample of the rock sample is tested,with the increase of the coal dust concentration,the mixed spectral curve approaches the coal curve.According to the experimental results and the identification results,the dust pollution in the actual working conditions has little effect on the recognition results.For the rock samples,only when the the reflected light from the rock is almost completely shielded by coal dust concentration,the final identification is wrong.3.Water mist and humidity have little effection the recognition results.In the actual working conditions of the water mist concentration and humidity,coal and mine can be even correctly identified.4.The influence of illumination on the recognition result depends mainly on the characteristic spectral band,and the characteristic of the 2210 nm band of the rock spectrum is greatly weakened under the 500 PUX illumination,but the rock can still be identified correctly under this illumination.In order to ensure robustness,it is necessary to match the intensity of the light source so that there is a certain degree of illumination redundancy to highlight the difference between coal and rock characteristics.5.As the distance increases,the intensity of the reflected light increases first and then decreases,and the intensity belongs to the subsequent weakening state in the normal working range.the coal and rock samples can be correctly identified within 2.0m.6.The horizontal velocity and vibration have little effect on the hyperspectral recognition of coal and rock,and the recognition effect is 100% under the experimental data.In conclusion,this paper validates the feasibility mechanism of the hyperspectral identification of coal and rock,designs an efficient hyperspectral coal-rock identification algorithm,and evaluates the effect of actual conditions on the hyperspectral recognition technology of coal and rock.Which provides a new technical reference for coal-rock identification.