中国矿业
中國礦業
중국광업
China Mining Magazine
2015年
11期
71-76
,共6页
张东%塔西甫拉提·特依拜%张飞%阿尔达克·克里木
張東%塔西甫拉提·特依拜%張飛%阿爾達剋·剋裏木
장동%탑서보랍제·특의배%장비%아이체극·극리목
高光谱%砷%分数阶微分%重金属%偏最小二乘回归
高光譜%砷%分數階微分%重金屬%偏最小二乘迴歸
고광보%신%분수계미분%중금속%편최소이승회귀
hyperspectral%arsenic%fractional derivative%heavy metals%PLSR
分数阶微积分将整数阶微积分的概念推广到分数阶。以准东煤田五彩湾矿区为研究区,测定土壤样本的砷含量及光谱反射率。对土壤光谱、均方根、倒数、对数变换进行对应的0~2阶(间隔0.2阶)微分处理,利用偏最小二乘回归进行建模,并比较不同模型的反演效果。结果表明:均方根变换1.8阶微分数据所建立的模型,RMSEC=1.996,R2C=0.821,RMSEP=1.973,R2P=0.890,RPD=2.367为最优模型,能较好的预测土壤砷含量。该结果可为分数阶微分在高光谱遥感监测土壤重金属污染中的应用提供参考。
分數階微積分將整數階微積分的概唸推廣到分數階。以準東煤田五綵灣礦區為研究區,測定土壤樣本的砷含量及光譜反射率。對土壤光譜、均方根、倒數、對數變換進行對應的0~2階(間隔0.2階)微分處理,利用偏最小二乘迴歸進行建模,併比較不同模型的反縯效果。結果錶明:均方根變換1.8階微分數據所建立的模型,RMSEC=1.996,R2C=0.821,RMSEP=1.973,R2P=0.890,RPD=2.367為最優模型,能較好的預測土壤砷含量。該結果可為分數階微分在高光譜遙感鑑測土壤重金屬汙染中的應用提供參攷。
분수계미적분장정수계미적분적개념추엄도분수계。이준동매전오채만광구위연구구,측정토양양본적신함량급광보반사솔。대토양광보、균방근、도수、대수변환진행대응적0~2계(간격0.2계)미분처리,이용편최소이승회귀진행건모,병비교불동모형적반연효과。결과표명:균방근변환1.8계미분수거소건립적모형,RMSEC=1.996,R2C=0.821,RMSEP=1.973,R2P=0.890,RPD=2.367위최우모형,능교호적예측토양신함량。해결과가위분수계미분재고광보요감감측토양중금속오염중적응용제공삼고。
Fractional calculus extends the conception of integer calculus to the fractional order .This study set Wucaiwan open coalmine area in the Eastern Junggar Basin in Xinjiang Uygur Autonomous Region as the study area ,the arsenic (As) element content and spectral reflectance of the soil samples were measured .The paper treated the hyperspectral reflectance data ( R ) with 3 mathematical transforms such as R ,1/R ,ln(R) ,then the formula of Grünwald‐Letnikov fractional derivative was used to calculate their 0~2 order derivative(interval 0 .2 order) ,and the inversion effects were compared after the models between different transforms and the As content in soil were set up by PLSR .Results show that the model based on R 1 .8 order derivative transform (RMSEC= 1 .996 ,R2C = 0 .821 ,RMSEP= 1 .973 ,R2P = 0 .890 ,RPD=2.367) is much better than others ,and has a better capability to predict As content in soil .The results of this research would provide a reference basis for the application of fractional order derivative in monitoring heavy metal contamination by using hyperspectral data .