摘要
不同种类塑料所含元素及含量不同,通过激光诱导击穿光谱技术结合机器学习算法以塑料基体元素(C、H、O、N)的原子及分子光谱峰值强度对7种塑料进行区分,避免了光谱中大量背景噪声及无关元素的干扰,分类时间较短并取得了较好的识别率,同时又以谱线积分强度对塑料进行鉴别,减弱了光谱波动和谱线线型等对分类的影响。实验结果表明利用谱线积分强度对塑料进行区分的识别正确率较谱线峰值强度有所提升。
Different kinds of plastics contain different elements and contents.Through laser-induced breakdown spectroscopy and machine learning algorithm,seven kinds of plastics are distinguished by the atomic and molecular spectral peak intensity of plastic matrix elements(C,H,O,N),avoiding the interference of a large number of background noise and irrelevant elements in the spectrum,resulting in a short classification time and a good recognition rate.At the same time,the plastic is identified by spectral line integral strength,which weakens the influence of spectral fluctuation and spectral line shape on classification.The experimental results show that the recognition accuracy of using spectral line integral strength to distinguish plastics is higher than that of spectral line peak strength.
作者
甄佳
乌日娜
付林
程德华
岱钦
李倩
李业秋
ZHEN Jia;WU Ri-na;FU Lin;CHENG De-hua;DAI Qin;LI Qian;LI Ye-qiu(School of Science,Shenyang Ligong University,Shengyang 110158,China)
出处
《激光与红外》
CAS
CSCD
北大核心
2022年第11期1587-1591,共5页
Laser & Infrared
基金
省博士科研启动基金计划项目(No.2021-BS-161)
辽宁省教育厅青年“育苗”项目(No.1030040000225)资助。
关键词
塑料
光谱
积分
分类
plastic
spectrum
integral
classification