|
Quantitative Determination of Fenthion Using Laser Induced Breakdown Spectroscopy with CARS Variable Selection Method |
|
View Full Text Download reader |
DOI: |
KeyWord:laser induced breakdown spectroscopy partial least square competitive adaptive reweighted sampling fenthion |
Author | Institution |
LIU Jin,GAN Lan-ping,SUN Tong,LIU Mu-hua |
1.江西农业大学工学院,江西省高校生物光电技术及应用重点实验室;2.江西省果蔬采后处理关键技术及质量安全协同创新中心 |
|
Hits: 2468 |
Download times: 1062 |
Abstract: |
In this study,double pulse laser induced break down spectroscopy(LIBS) technique was used to detect the content of fenthion in solution.The spectra of different concentrations’ samples in range of 206.28-481.77 nm band were collected using a two channel spectrometer with high precision.And then several preprocessing methods such as multiplicative scatter correction(MSC),standardized normal variate(SNV) and 3 point smoothing were conducted on the spectra.The optimal pretreatment method was confirmed according to partial least square(PLS) modeling.On this basis,competitive adaptive reweighted sampling(CARS) was used to screen the important variables related to fenthion.Then,the quantitative analysis model for fenthion in solution was established by PLS regression.Finally,the CARS-PLS quantitative analysis model was compared with the single variable quantitative analysis model and the PLS quantitative analysis model without variable selection.The results indicated that the CARS-PLS quantitative analysis model has a better performance compared with single variable quantitative analysis model and PLS model,and its determination coefficient and average relative error of calibration set and prediction set are 0.969 4,15.537% and 0.995 9,5.016%,respectively.Furthermore,the CARS-PLS model adopted only 1.9% of the wavelength variables,but the average error of the prediction set was decreased from 9.829% to 5.016%.Thus it is found that LIBS technology has a certain feasibility to detect the content of fenthion in solution.And CARS could simplify the quantitative analysis model and improve the prediction accuracy of the model. |
Close |
|
|
|