원문정보
초록
영어
To achieve the rapid detection of soybean straw component, the key lies in establishing a quantitative analysis model with higher prediction accuracy which is rapid, stable and reliable. In order to establish the optimal Near-infrared (NIR) analysis model of cellulose and hemicellulose content in soybean straw, this paper uses NIR transmission technology by applying interval Partial Least Squares (iPLS) on the optimization of characteristic spectrum range of cellulose and hemicellulose spectrum. During the optimized characteristic spectrum range, prediction models of Partial Least Squares Regression (PLSR) and the Back Propagation Neural Network (BPNN) are built in the cellulose and hemicellulose contents respectively. The results show that the best modeling band of the Cellulose content is 5615-5731cm-1, and the optimal coefficient of determination of prediction model, PredictionR2(P-R2) reaches 0.9179266; And the best modeling band of the hemicellulose content is 5615-5731cm-1 ,the P-R2 is 0.920407. After the selection of iPLS optimal band, the quantitative analysis model of cellulose and hemicelluloses established by adopting the PLSR and BP Neural Network is more concise and has higher prediction accuracy and faster data computing speed. It also provides a theoretical basis for the optimization of characteristic spectrum range for the design of small dedicated NIR analytical instruments.
목차
1. Introduction
2. Materials and Methods
2.1 Sample Collection and Preparation
2.2 Spectral Acquisition
2.3 Chemical Analysis
3. Results and Discussion
3.1 Spectral Data Preprocessing
3.2 Interval Partial Least Squares Band Selection
3.3 Evaluation of Prediction Models
3.4 BP Neural Network Modeling
4. Conclusion
Acknowledgements
References