Prediction of protein content of winter wheat by canopy of near infrared spectroscopy (NIRS), using partial least squares regression (PLSR) and artificial neural network (ANN) models

Abstract

In this study to predict amount of protein in wheat, near infrared spectroscopy technique (NIRS) was used that is a non-destructive and fast observing method. Partial Least Squares Regression (PLSR) and Artificial Neural Network (ANN) methods were used to choose the spectral bands and the best models, respectively. To compare the efficiency of models Root-mean-square error (RMSE) and R 2 were applied. The finest consequence by cascade forward back propagation (CFBP) was related to network structure of 8-8-1 with Levenberg-Marquardt (LM), and function of TANSIG-TANSIG-PURELIN (TANSIG-TANSIG-PURELIN (R 2 =0.0289 and R 2 =0.9881 at 14 epochs). The consequences of estimation for ANN model (R 2 =0.9881) was better than the PLSR model (R 2 =0.9783). Therefore, according to the results, it can be said that NIRS has a high potential for predicting the amount of protein in wheat.