کد مقاله کد نشریه سال انتشار مقاله انگلیسی نسخه تمام متن
5132145 1491508 2017 7 صفحه PDF دانلود رایگان
عنوان انگلیسی مقاله ISI
Optimized self-adaptive model for assessment of soil organic matter using Fourier transform mid-infrared photoacoustic spectroscopy
موضوعات مرتبط
مهندسی و علوم پایه شیمی شیمی آنالیزی یا شیمی تجزیه
پیش نمایش صفحه اول مقاله
Optimized self-adaptive model for assessment of soil organic matter using Fourier transform mid-infrared photoacoustic spectroscopy
چکیده انگلیسی


- A quick Optimized self-adaptive model was designed for SOM prediction.
- Five algorithms are compared in soil identification based on soil similarity.
- Euclidean distance-SAM is suitable for both different types and same soil type.
- Correlation coefficient-SAM is accurate in different types.

Advanced technologies, such as infrared spectroscopy, have been applied to develop rapid, cheap but accurate methods for the analysis of soil matter organic (SOM). However, the unsatisfied prediction accuracy resulted from heavy soil heterogeneity limits the practical application. In our previous work, soil identification based self-adaptive partial least squares model (SAM), which was built using identification algorithm and the partial least square regression (PLSR), makes it possible for a wide use. However, soil identification in the SAM needs be further optimized. In this study, we designed an advanced optimal self-adaptive partial least squares model (OPT-SAM), a more general model to predict SOM. 597 soil samples from China with large variances were collected, and the soil spectra were recorded using Fourier transform mid-infrared photoacoustic spectroscopy (FTIR-PAS). Five typical algorithms (Correlation coefficients (CC), Euclidean distance (ED), Mahalanobis distance (MD), Angle cosine (AC), and k-medoids (KM)) were considered for the identification in the SAM model. The results demonstrated that the performances of CC-SAM, ED-SAM, MD-SAM, AC-SAM were significantly improved in comparison with no identification based SAM (NI-SAM), but KI-SAM showed a poor prediction. ED-SAM (R2 = 0.8890, RMSEP = 7.00 g kg−1, RPD = 2.96) indicated the highest accuracy and robustness in all algorithms, which was an optimal model for soil identification and prediction, and CC-SAM (R2 = 0.8572, RMSEP = 7.89 g kg−1, RPD = 2.44) was an alternative choice, especially for prediction with different soil types.

ناشر
Database: Elsevier - ScienceDirect (ساینس دایرکت)
Journal: Chemometrics and Intelligent Laboratory Systems - Volume 171, 15 December 2017, Pages 9-15
نویسندگان
, , , ,