当前位置: X-MOL 学术Optik › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Quantifying soluble sugar in super sweet corn using near-infrared spectroscopy combined with chemometrics
Optik ( IF 3.1 ) Pub Date : 2020-06-16 , DOI: 10.1016/j.ijleo.2020.165128
Quannu Yang , Xinhao Yang , Qianling Zhang , Yunbo Wang , Han Song , Furong Huang

Soluble sugar content is a key factor affecting super sweet corn quality, making the development of a rapid, simple, and environmentally friendly method for measuring soluble sugar content significant for successful breeding. The near-infrared (NIR) spectra of 131 sets of super sweet corn kernels with different soluble sugar contents (5.57–45.35 mg/g) were collected and preprocessed using multiple scattering correction (MSC), standard normal variate (SNV) transformation, and first and second derivatives. The first derivative spectrum, which gave the best preprocessing result, was used to construct the synergy interval partial least squares (Si-PLS) model. The PLS model developed using the 1349−1513 nm, 1842−2005 nm, 2005−2168 nm, and 2337−2500 nm wavebands gave the best result: root mean square error of the prediction set (RMSEP) =6.9199 mg/g and correlation coefficient of the prediction set (RP) = 0.7695. A competitive adaptive reweighted sampling (CARS)-Si-PLS wavelength screening algorithm was used to improve the predictive accuracy of the model further (RMSEP =5.8292 mg/g and RP = 0.8431). Compared to the original spectrum, the optimized model using CARS-SI-PLS is more concise and robust, confirming the ability of NIR spectroscopy to accurately measure soluble sugar content in super sweet corn.



中文翻译:

使用近红外光谱结合化学计量学对超甜玉米中的可溶性糖进行定量

可溶性糖含量是影响超甜玉米品质的关键因素,这使得开发一种快速,简单且环保的方法来测定可溶性糖含量对于成功育种具有重要意义。收集131组具有不同可溶性糖含量(5.57–45.35 mg / g)的超甜玉米粒的近红外(NIR)光谱,并使用多重散射校正(MSC),标准正态变量(SNV)转化和预处理一阶和二阶导数 给出最佳预处理结果的一阶导数光谱用于构建协同区间偏最小二乘(Si-PLS)模型。使用1349-1513 nm,1842-2005 nm,2005-2168 nm和2337-2500 nm波段开发的PLS模型给出了最佳结果:预测集的均方根误差(RMSEP)= 6。P)= 0.7695。使用竞争性自适应加权采样(CARS)-Si-PLS波长筛选算法进一步提高了模型的预测准确性(RMSEP = 5.8292 mg / g,R P = 0.8431)。与原始光谱相比,使用CARS-SI-PLS进行优化的模型更加简洁和强大,从而证实了近红外光谱技术能够准确测量超级甜玉米中可溶性糖含量的能力。

更新日期:2020-06-16
down
wechat
bug