Neural Networks Calibration Model Optimizing of Wheat Protein Based on Successive Projections Algorithm

Abstract:

In order to reduce computational complexity of modeling and improve the model's robustness and prediction accuracy, successive projections algorithm is used in the Neural Networks calibration modeling of wheat protein. Firstly, the spectral data is pretreated with first-order differential method and SNV method，and then a representative set of calibration samples are selected by SPXY algorithm. Secondly, making use of successive projections algorithm(SPA) to extract sensitive wave points of the original spectrum and the pretreated spectrum, and then the neural networks calibration model of wheat protein is established. The results show that the calibration model based on successive projections algorithm has a fast convergence speed and high accuracy, both of which are better than the calibration model established with the original data. Root mean square error of prediction(RMSEP) and prediction correlation coefficient(r) are 1.3332 and 0.94319 respectively, which can basically complete the grain reserves and the food processing profession division and the breeding preliminary generation.

• Periodical: Applied Mechanics and Materials (Volumes 401-403)
• Pages: 1565-1570
• DOI: https://doi.org/10.4028/www.scientific.net/AMM.401-403.1565
• Online since: September 2013
• Authors: Xiao Dong Mao, Lai Jun Sun, Gang Hao, Lu Lu Xu, Guang Yan Hui
• Keywords: Near Infrared Spectroscopy (NIRS), Neural Network (NN), Protein, Successive Projectionsalgorithm, Wheat

References

[1] Xiaoyan Liang, Haiyan Ji, Applications of Near Infrared Spectroscopy Technology in Analyzing the Quality of Crops, J. Chinese Agricultural Science Bulletin. 22 (2006) 366-371.

[2] Yanlu yan, Longlian Zhao, Donghai Han, et al, Fundamentals and Applications of Near Infrared Spectroscopy, China Light Industry Press, Beijing, 2005, pp.32-33.

[3] M. Zeaiter, J-M. Roger, V, Bellon-Maurel. Robustness of models developed by multivariate calibration. Part II: The influence of pre-processing methods, J. Trends in Analytical Chemistry. 24 (2005) 437-445.

DOI: 10.1016/j.trac.2004.11.023

[4] Xiaori Zhan, Xiangrong Zhu, Xinyuan Shi, et al, Determination of Hesperidin in Tangerine Leaf by Near-Infrared Spectroscopy with SPXY Algorithm for Sample Subset Partitioning and Monte Carlo Cross Validation, J. Spectroscopy and Spectral Analysis. 29 (2009).

[5] Hongzhi Gao, Qipeng Lu, Haiquan Ding, et al, Choice of Characteristic Near-Infrared Wavelengths for Soil Total Nitrogen Based on Successive Projection Algorithm, J. Spectroscopy and Spectral Analysis. 29 (2009) 2951-2954.

[6] Bin Chen, Xianglong Meng, Hao Wang, Application of Successive Projections Algorithm in Optimizing Near Infrared Spectroscopic Calibration Model, J. Journal of Instrumental Analysis. 26 (2007) 66-69.

[7] Di Wu, Hongxi Wu, Jingbo Cai, et al, Classifying the Species of Exopalaemon by Using Visible and Near Infrared Spectra with Uninformative Variable Elmination and Successive Projections Algorithm, J. Journal Infrared Millim. Waves. 28 (2009).

DOI: 10.3724/sp.j.1010.2009.00423