摘要
Asanimportantnon-ferrousmetalstructuralmaterialmostusedinindustryandproduction,aluminum(Al)alloyshowsitsgreatvalueinthenationaleconomyandindustrialmanufacturing.HowtoclassifyAlalloyrapidlyandaccuratelyisasignificant,popularandmeaningfultask.Classificationmethodsbasedonlaser-inducedbreakdownspectroscopy(LIBS)havebeenreportedinrecentyears.AlthoughLIBSisanadvanceddetectiontechnology,itisnecessarytocombineitwithsomealgorithmtoreachthegoalofrapidandaccurateclassification.Asanimportantmachinelearningmethod,therandomforest(RF)algorithmplaysagreatroleinpatternrecognitionandmaterialclassification.ThispaperintroducesarapidclassificationmethodofAlalloybasedonLIBSandtheRFalgorithm.TheresultsshowthatthebestaccuracythatcanbereachedusingthismethodtoclassifyAlalloysamplesis98.59%,theaverageofwhichis98.45%.ItalsorevealsthroughtherelationshiplawsthattheaccuracyvarieswiththenumberoftreesintheRFandthesizeofthetrainingsamplesetintheRF.Accordingtothelaws,researcherscanfindouttheoptimizedparametersintheRFalgorithminordertoachieve,asexpected,agoodresult.TheseresultsprovethatLIBSwiththeRFalgorithmcanexactlyclassifyAlalloyeffectively,preciselyandrapidlywithhighaccuracy,whichobviouslyhassignificantpracticalvalue.
出版日期
2019年03月13日（中国期刊网平台首次上网日期，不代表论文的发表时间）
