欢迎来到天天文库
浏览记录
ID:56182308
大小:616.80 KB
页数:5页
时间:2020-06-04
《基于近红外光谱的小麦品质分类研究.pdf》由会员上传分享,免费在线阅读,更多相关内容在行业资料-天天文库。
1、中国麈学通报2013,29(36):386—390ChineseAgriculturalScienceBulletin基于近红外光谱的小麦品质分类研究毛晓东,孙来军,戴常军,惠光艳,徐璐璐(黑龙江省电子工程省高校重点实验室(黑龙江大学),哈尔滨150080)摘要:为了快速、简便、准确地鉴别小麦品质的类别,本研究提出了应用近红外光谱分析技术结合BP神经网络的鉴别方法对小麦进行品质分类。研究过程中对小麦样品的光谱数据进行了详细分析,采用马氏距离剔除了光谱数据中异常数据,并通过主成分分析说明利用近红外光谱鉴别小麦品质分类的可行性。为了提高所建模型的性能,采用SPXY算法对小麦样品进行
2、合理的划分。并选取了一阶微分加归一化的预处理方法来处理光谱数据,消除无关信息和噪声对小麦光谱数据的影响。运用偏最小二乘法压缩光谱数据,减少了数据量,节省建模时间。最后采用BP神经网络方法建立了小麦品质分类模型。实验结果显示:模型的鉴别效果较好,对强筋样品识别的准确率高达94.4%,弱筋样品识别的准确率高达100%。实现了快速、准确地对小麦品质强筋和弱筋两类的鉴别,对小麦生产、市场交易及食品加工有着非常重要的意义。关键词:小麦;近红外光谱;偏最小二乘;BP神经网络中图分类号:S1文献标志码:A论文编号:2013.1332StudyonClassificationofWheatQu
3、alityBasedonNearInfraredSpectroscopyMaoXiaodong,SunLaijun,DaiChangjun,HuiGuangyan,Xululu(Key'LaboratoryofElectronicsEngineering,CollegeofHeilongfiangProvince(HeilongfiangUniversity),Harbin150080)Abstract:Inordertoquick,easyandaccuratelyidentifytheclassificationofwheatquality,thispaperputforw
4、ardusingidentificationmethodofnearinfraredspectralanalysistechnologycombinedwithBPneuralnetworkforclassificationofwheat.Samplesofwheatwerecarriedoutadetailedanalysisofspectraldataintheprocessofresearch.Firstofall,Mahalanobdistancewasappliedonspectraldatafiltering,whichcouldeliminateabnormals
5、pectrum.Andprincipalcomponentanalysisexplainedthatnearinfraredspectroscopyidentifiedthefeasibilityoftheclassificationofwheatquality.Inordertoimprovetheperformanceofmodel,SPXYalgorithmwasusedforreasonabledivisionofsamplesofwheat.ThenthedataprocessingmethodoffirstderivativeandSNVwereusedtodeal
6、withspectraldata,whichcouldeliminateirrelevantinformationandnoiseonwheatspectraldata.Themethodofpartialleastsquareswasusedtocompressspectraldata,whichcouldreducetheamountofdataandsavemodelingtime.Finally,BPneuralnetworkasthemodelingmethodwasusedtoestablishidentificationmodelofwheatquality.Ex
7、perimentalresuhshowedthat:themodelidentificationeffectwasgood,therecognitionaccuracyofstrongglutensampleswasashighas94.4%,theidentificationaccuracyofweakglutensampleswasashighas100%,whichhadrealizedquicklyandaccuratelyclassificationbetweenstrongglu
此文档下载收益归作者所有