PEER-REVIEW ARTICLE


 

PEER-REVIEWED ARTICLE  bioresources.com 

 

 

Wu et al. (2023). “FTIR classification of alfalfa hay,” BioResources 18(3), 5399-5416.  5399 

 

Classification of Alfalfa Hay Based on Infrared 
Spectroscopy  
 

Xiaoqing Wu,a,b Guifang Wu,a,* Bo Wang,a and Jie Li a 
 
Alfalfa hay plays a decisive role in the quality and safety of livestock 
products. Chemical analytical methods for alfalfa hays are laborious, time-
consuming, and costly. Therefore, suitable methods are required for rapid 
and accurate detection of alfalfa hay. This study evaluated the feasibility 
of infrared spectroscopy (IR) in identifying different alfalfa hays. 105 alfalfa 
hay samples under three different drying methods were analysed. Results 
indicated that the full spectra model constructed through standard normal 
variable transformation (SNV), first-derivative (FD), and second-derivative 
(SD) preprocessing by BP and SVM had the best performance. The 
accuracies were all up to 100%. Under the same preprocessing method, 
the accuracy of BP neural networks was better than that of support vector 
machine models in most cases. The characteristic wavelength-based 
SNV-SD-SPA by BP exhibited better performance than the other 
pretreatment methods, such as: SNV-SPA, SNV-FD-SPA, and SNV-GA, 
etc. The classification accuracy of moldy-dried alfalfa, sun-dried alfalfa, 
and shade-dried alfalfa in the training set were 100%, 100%, and 99.5%, 
respectively, and the accuracy of the prediction set reached 100%, 97.6%, 
and 97.4%, respectively. Thus, a better theoretical basis was obtained for 
the grading and online monitoring of alfalfa hay.  
 

DOI: 10.15376/biores.18.3.5399-5416 

 

Keywords: Alfalfa hay; Infrared spectroscopy; Machine learning; Classification 

 
Contact information: a: College of Mechanical & Electrical Engineering, Inner Mongolia Agricultural 

University, Hohhot, 010018, P.R. China; b: College of Physics and Electronic Information, Inner Mongolia 

Normal University, Hohhot, 010022, P.R. China; *Corresponding author: wgfsara@imau.edu.cn 

 
 

INTRODUCTION 
 

Alfalfa hay is an important feed for dairy cows and plays an important role in the 

healthy and stable development of the dairy market (Darabighane et al. 2020; Lorenzo et 

al. 2020). Dried alfalfa needs to be compressed and processed into a certain size bale for 

storage and transportation (Cheng et al. 2018; Vanzant et al. 1990). To ensure alfalfa 

nutrition, bundling is carried out according to a certain water content (Han et al. 2004; Lim 

et al. 2020). Improper antimildew measures are conducive to the proliferation of 

microorganisms and cause alfalfa mildew (Wang et al. 1996). The nutrient content of 

alfalfa after mildew infestation is destroyed, and its feeding value is low, which can cause 

livestock poisoning and affect the milk product quality (Coblentz et al. 1996). Therefore, 

if alfalfa mildew can be quickly identified during drying or storage, the loss can be 

effectively reduced. Traditional methods of chemical detection of mold generally have the 

characteristics of cumbersome operation, long detection period, and high cost (Gfrerer et 

al. 2004). 

Infrared spectroscopy technology is a chemical analysis method that can detect 

different absorbance frequencies of specific molecules in substances and is fast and 



 

PEER-REVIEWED ARTICLE  bioresources.com 

 

 

Wu et al. (2023). “FTIR classification of alfalfa hay,” BioResources 18(3), 5399-5416.  5400 

nondestructive (Xiong et al. 2016; Zhou et al. 2022). Because different chemical 

components contain different chemical groups, corresponding to different group 

frequencies, the positions of the generated characteristic absorbance peaks are also 

different, and moreover, for the same chemical composition, the intensity of the 

characteristic absorbance peaks reflected by the different content is not the same (Hell et 

al. 2016). Therefore, for both quantitative and qualitative analyses of substances, infrared 

spectroscopy can be utilized. Traditional mid-infrared spectroscopic analysis requires the 

production of potassium bromide tablets for solid samples. Attenuated total reflection 

(ATR) technology obtains the information through the reflection signal of the sample 

surface (Undugodage et al. 2018). It has the characteristics of high sensitivity, clearly 

characteristic bands, simple operation, and there is no need for sample preparation 

(Kuronuma et al. 2020). However, the intensity of the overall signal is hard to control when 

using ATR plate methods, since it depends on the smoothness and pressure of pressing the 

specimen onto the plate.  In recent years, infrared spectroscopy techniques have been 

widely used in food, pharmaceutical spetrochemicals, tea, wood, feed, and other fields 

(Mohebby 2010; Wallén et al. 2018; Zapata et al. 2021). There are also some research 

reports on the detection of food mildew by infrared spectroscopy. Shen and Huang 

established an online corn mold detection system using spectra and image information 

fusion technology, by collecting the spectra and image information of corn samples stored 

on days 6, 9, 12, and 15 and establishing the discriminant linear analysis model, an overall 

recognition rate of 91.1% was obtained for different degrees of mildew (Shen and Huang 

2019). Chu et al. (2014) used near-infrared spectroscopy technology to detect corn kernels 

with different degrees of mildew. They used principal component analysis to reduce the 

dimensionality of the spectral data and established a model with FDA (Fisher discriminant 

analysis), which had a classification accuracy of 91.4%. At present, most studies use 

spectroscopy and machine vision techniques to detect mildew in food, and there are few 

reports on the use of infrared spectroscopy to detect mildew in alfalfa hay.  

Infrared spectral data has the characteristic of high correlation between two 

adjacent spectra data and high dimensionality (Tanaka et al. 2011). Using full-spectra data 

to build a model will increase the computing time, and the recognition and classification 

results may not be ideal.  

With the development of computer science and artificial intelligence, more machine 

learning algorithms have been developed and applied to information mining of infrared 

spectra. Machine learning is a field of study that automatically detects patterns in data from 

a given database of knowledge and then uses the detected patterns to predict unknown data. 

Therefore, infrared spectroscopy combined with machine learning may be a potential 

solution for identifying the quality of alfalfa hay (Kumar et al. 2017). 

The objectives of this study were as follows: (1) to acquire spectra of alfalfa hay, 

(2) to determine the optimal wavelength using successive projection algorithm (SPA) and 

genetic algorithm (GA), (3) to construct a classification model by using the full spectra and 

optimal wavelengths, and (4) to use neural network and SVM procedures to classify the 

extracted features.  

  



 

PEER-REVIEWED ARTICLE  bioresources.com 

 

 

Wu et al. (2023). “FTIR classification of alfalfa hay,” BioResources 18(3), 5399-5416.  5401 

EXPERIMENTAL 
 
Preparation of Experimental Samples 

Samples used for this study were collected from an experimental field of Inner 

Mongolia Agricultural University in 2019. They were split into three categories of dry 

alfalfa: alfalfa dried in the shade, alfalfa naturally dried in the sun, and moldy naturally 

dried in the sun. The alfalfa moisture content was approximately 15% to 20%. Three 

different types of dry alfalfa were first ground into powder using an electric high-speed 

pulverizer, which was followed by a 1-mm mesh sieve to remove impurities. Finally, 5 

grams of the powdered alfalfa were weighed using an electronic scale into a 50-mL test 

tube and covered with plastic wrap for storage. Thirty-five samples were prepared for each 

type of dried alfalfa, and all 105 samples were prepared. 

 

Infrared Spectral Acquisition 
Infrared spectra were recorded using an attenuated total reflectance sampling 

accessory (PerkinElmer, Boston, MA, USA) and PE series software. Reflectance data were 

recorded over the wavenumber range of 400 to 4000 cm-1 with 64 scans per spectra and a 

spectral resolution of 4 cm-1. The acquisition time for a single spectra was 66 s. Background 

spectra were collected with no samples present on the crystal, and under the same 

experimental conditions. To assess repeatability and identify any problems caused by the 

sample's finite particle size, three spectra for each sample were gathered. For the statistical 

analysis, the average of these three spectra for each sample was then used. The files were 

exported as comma separated value (csv) files and imported into the MATLAB software 

(Mathworks, 2020b, Natick, MA, USA) for analysis. The first and last noise of the spectral 

data was relatively large, and finally 600 to 2000 cm-1 were retained. There were 701 

variables in each spectra for preprocessing and modelling in the study.  
 

Methods 
Pretreatment of the spectral data 

Pretreatment of the averaged spectra was required to eliminate mechanical noise 

and baseline drift. Pretreatment methods include standard normal variable (SNV), MSC 

(multiplicative scatter correction), first derivatives (FD), second derivatives (SD), and 

Savitzky-Golay convolution smoothing (SG), and so on. In order to eliminate strength 

differences between different samples and analyze data, all data were normalized before 

preprocessing. The standard normal variable transformation is primarily used for the 

surface scattering influence and light intensity changes on the spectra, and multivariate 

scattering correction is used to eliminate the influence of particle size and scattering caused 

by particle inhomogeneity (Kamruzzaman et al. 2016). A derivative operation is used to 

eliminate the shift of the baseline. The SG smoothing improves the smoothness of the 

spectra and reduces the interference of noise (Rahman et al. 2016). In this study, spectral 

data preprocessing was performed using Unscrambler 10.1 (Camo Software, Oslo, 

Norway). 

 

Characteristic Wavelength Selection 
Infrared spectral data contain hundreds of continuous wavelengths, which are 

redundant and multicollinear. Eliminating redundant wavelengths and selecting 

optimal variables not only can simplify the modelling process and reduce costs and 

running time, but also, they can improve the performance of the model. In this study, 



 

PEER-REVIEWED ARTICLE  bioresources.com 

 

 

Wu et al. (2023). “FTIR classification of alfalfa hay,” BioResources 18(3), 5399-5416.  5402 

the uninformative variable elimination (UVE)-SPA and GA methods were selected 

to extract the optimal wavelengths in MATLAB (Version 2020a, MathWorks, 

Natick, MA, USA). 

In the UVE-SPA method, UVE can remove a lot of invalid information. 

Variable modelling based on UVE selection can avoid model overfitting and 

improve its predictive ability. The SPA mainly solves the problem of collinearity, 

and it is used to select the wavenumber with the lowest redundant information and 

obtain the useful variable with the least collinearity (Mário et al. 2001). SPA has been 

widely used in the selection of spectral characteristic variables. The basic principle 

of the SPA is to simply project a set of wavelength subsets into the vector space and 

select the wavelength subset with the least redundancy. The algorithm steps are 

described below, assuming that the first wavelength k(0) and N are given. 

Step1: Before the first iteration (n=1),let 𝑥𝑗 = 𝑗 𝑡ℎ 𝑐𝑜𝑙𝑢𝑚𝑛 𝑜𝑓 𝑋𝑐𝑎𝑙; 𝑗 = 1, … , 𝐽. 

Step2: Let S be the set of wavelength which have not been selected yet. 

 S = {𝑗 𝑠𝑢𝑐ℎ 𝑡ℎ𝑎𝑡 1 j J and j {k(0), … , k(n − 1)}}.  

Step3: Calculate the projection of 𝑥𝑗 on the subspace orthogonal to 𝑥𝑘(𝑛−1)as 

P𝑥𝑗 = 𝑥𝑗 − (𝑥𝑗
𝑇𝑥𝑘(𝑛−1))𝑥𝑘(𝑛−1)(𝑥𝑘(𝑛−1)

𝑇 𝑥𝑘(𝑛−1))
−1

 

for all 𝑗 ∈ 𝑆, where P is the projection operator. 

Step4: Let 𝑘(𝑛) = arg (max‖P𝑥𝑗‖, 𝑗 ∈ 𝑆). 

Step5: Let 𝑥𝑗 = P𝑥𝑗, 𝑗 ∈ 𝑆. 

Step6: Let 𝑛 = n + 1, if n<N go back to Step2. 

 Step7: The resulting wavelengths are {𝑘(𝑛): 𝑛 = 0, … , 𝑁 − 1}. 

  

The genetic algorithm is an intelligent optimization algorithm designed and 

proposed by John Holland according to the evolutionary laws of organisms in nature. 

The genetic algorithm simulates the phenomena of reproduction, crossover, and 

gene mutation that occur in natural selection and natural genetic processes (Ji et al. 

2022). In each iteration, a set of candidate solutions are retained, and a better 

individual is selected from the solution group according to a certain index. We use 

genetic operators (selection, crossover, and mutation) to combine these individuals 

to produce a new generation of candidate solution groups and repeat this process 

until a certain convergence index is met. A genetic algorithm's specific procedure is 

depicted in Fig.1. 

 



 

PEER-REVIEWED ARTICLE  bioresources.com 

 

 

Wu et al. (2023). “FTIR classification of alfalfa hay,” BioResources 18(3), 5399-5416.  5403 

 
 

Fig. 1. The flow chart of GA 
 

BP Neural Network 
The back propagation (BP) neural network is a concept proposed by scientist 

leaders Rumelhart and McClelland in 1986 (Rumelhart et al. 1986). The learning process 

of the BP network is an error correction learning algorithm that is composed of forward 

propagation and back propagation. In the forward propagation process, the input signal 

propagates from the input layer to the hidden layer and the output layer through the 

activation function. The neuron state of each layer only affects the neuron state of the next 

layer. If the desired output cannot be obtained in the output layer, it will switch to back 

propagation and return according to the original link path. The topology of the neural 

network is shown in Fig. 2. Equations 1 and 2 provide the weights, thresholds, and transfer 

functions that link the neurons in the input layer, hidden layer, and output layer, 

ℎ𝑝 = 𝑓1(∑ 𝜔𝑛𝑝𝑥𝑛 − 𝑧𝑝
3
𝑛=1 ), 𝑝 = 1,2, … 𝑚       

       (1) 

 𝑦𝑝 = 𝑓2(∑ 𝜔𝑝𝑞𝑟𝑝 − 𝑧𝑞
𝑚
𝑝=1 ), 𝑞 = 1        

       (2) 

where n is the input layer's number of neurons; p is the hidden layer's number of neurons; 

and q is the output layer's number of neurons. For Eq. 2, f1 and f2 are the activation function 

of the hidden layer and the output layer; 𝜔𝑛𝑝is the nth input neuron to the pth weights of 

hidden neurons; 𝜔𝑝𝑞is the weight from the pth hidden neuron to the qth output neuron; 𝑧𝑝is 

the threshold from the input layer to the hidden layer; 𝑧𝑞is the threshold from the hidden 

layer to the output layer; and 𝑦𝑞 𝑖𝑠 𝑡ℎ𝑒 output for the neural network. 

 



 

PEER-REVIEWED ARTICLE  bioresources.com 

 

 

Wu et al. (2023). “FTIR classification of alfalfa hay,” BioResources 18(3), 5399-5416.  5404 

 
 

Fig. 2. The structure of neural network 
 

Support Vector Machine 
Support Vector Machine (SVM) is a commonly used machine learning algorithm 

in spectral analysis, which performs well when classifying small amounts of high-

dimensional data (Chang and Lin 2007). By using different kernel functions, SVM has the 

powerful ability to handle linear and nonlinear problems. In this study, the Radial basis 

function (RBF) was selected as the kernel function, and the parameters c and g were 

determined through optimization. 
 

Model Performance Evaluation 
To test the stability of the model, Monte Carlo cross-validation was used to 

divide the dataset into 20 different points, and the average accuracy under different 

datasets was calculated as the basis for the comparison of different models. At the 

same time, the coefficient of variation was used to evaluate the stability of the model 

under different data sets. The stability of the system increases with decreasing 

coefficient of variation. Data set A (containing 84 sets of data) was used as the 

calibration set, and data set B (containing 21 sets of data) was used as the test set. 

The models were evaluated using classification accuracy as a criterion. 

 

 
RESULTS AND DISCUSSION 
 

Figure 3 shows the average spectra of the three alfalfa species. It is essential for 

describing the main traits and characteristics of the Mid-Infrared (MIR) spectra of alfalfa 

hay. The C-OH stretch of cell wall polysaccharides is responsible for the strongest band in 

the spectra, located at 1040 cm-1. The vibration peaks are CH3 symmetrical bending 

vibration peak near 1375 cm-1, N/N vibration peak near 1418 cm-1, and weak absorbance 

peaks around 1500 cm-1 to 1600 cm-1, which are absorbance peaks of benzene ring (Kaya 

and Huck 2017; Josiah et al. 2018). The average spectra of the three alfalfa species 

generally converged but differed, as shown in Fig. 4. 
 



 

PEER-REVIEWED ARTICLE  bioresources.com 

 

 

Wu et al. (2023). “FTIR classification of alfalfa hay,” BioResources 18(3), 5399-5416.  5405 

 
Fig. 3. The average spectra 

 

 
 

Fig. 4. The average spectra of three alfalfa species 
 

Spectral Characteristics 
Figure 5(a) shows the original spectral data from 105 alfalfa hay specimens. 

Through observing the original spectra, it can be seen that the overall spectra data 

tends to be consistent. However, the degree of dispersion is high, and it is impossible 

to distinguish the samples dried in three ways through spectral data. Figure 5(b 

through f) shows the spectral curves for the various pretreatments, including SNV, 

FD, SD, SNV+MSC, and SNV+SG. After pretreatments, all these models kept their 

original spectral features. The processing results show that various processing 

techniques cause the spectral data to be processed differently and become smoother. 



 

PEER-REVIEWED ARTICLE  bioresources.com 

 

 

Wu et al. (2023). “FTIR classification of alfalfa hay,” BioResources 18(3), 5399-5416.  5406 

                                                                   

 
(a)                                                             (b) 

       

 
          
                                     (c)                                                                    (d)    
        

 
 

(e)                                                                         (f) 
                                              

Fig. 5. The raw and pretreated spectral curves of all alfalfa via different methods: (a) raw; (b) 
SNV; (c) FD; (d): SD; (e) SNV+MSC; and (f) SNV+SG 

 



 

PEER-REVIEWED ARTICLE  bioresources.com 

 

 

Wu et al. (2023). “FTIR classification of alfalfa hay,” BioResources 18(3), 5399-5416.  5407 

Characteristic Wavelength Selection  
To simplify the classification model, UVE-SPA and the GA were used for 

characteristic wavelengths. The screening results are shown in Fig. 6 and listed in Table 1. 

Figure 6 shows the characteristic wavelength extraction results for the SNV, SG, MSC, 

SG-SNV, and SG-MSC models, respectively. The characteristic spectra after 4 and 5 cycles 

of dimensionality reduction are shown in Fig. 7. 

As presented in Fig. 6 and Table 1, after variable selection the number of 

characteristic wavelengths selected by the SPA, was reduced 96%, 96.3%, 97.1%, 97.9%, 

and 96.7% when the SNV, MSC, SD, FD, and SG pretreatment methods were employed, 

respectively. These results indicate the effectiveness of the SPA in dimension reduction. 

After wavelength selection, spectral reflection values at specific wavelengths were 

extracted, and a simplified classification model was constructed to replace the full spectra 

as the input for the subsequent classification mode. 

As shown in Fig. 7 and Table 1, after variable selection, the number of characteristic 

wavelengths selected by the GA was reduced to 96.7%, 96.9%, 96.4%, and 97.4%.  

        
(a)                                                             (b)                                                                    

 
(c)                                                                        (d) 

 



 

PEER-REVIEWED ARTICLE  bioresources.com 

 

 

Wu et al. (2023). “FTIR classification of alfalfa hay,” BioResources 18(3), 5399-5416.  5408 

 
                                                                                      (e) 

 

Fig. 6. Wavelength selection results on the pretreated spectral data via the UVE-SPA method: (a) 
SNV; (b) MSC; (c) FD; (d) SD; and (e) SG 
 

The specific parameter settings for feature spectra extraction using GA algorithm 

were as follows: the initial population was 701 cm-1 composed of 0, 1, and a total of 30. In 

the experiment, the genetic algorithm was used to reduce the dimensionality, which 

reduced the number of wavenumbers from 701 in the whole band to 20 to 40 waves in 5 

cycles. The precise band selected in each cycle was half of the original wavenumber. The 

genetic algorithm uses the fitness function to evaluate the quality of individual solutions. 

When the value of the fitness function is larger, the quality of the solution is much better. 

In this paper, the error sum of squares was used as the fitness function, and GAOT pachage 

was used. The specific genetic operator settings were as follows: normGeomSelect was 

selected for selection operator, simpleXover for the crossover operator, and boundary 

Mutation selected for the mutation operator. 
 

Table 1. Wavelength Selection for Classification 

Pre-processing 
Technique  

Method Number  Wavelength (cm-1) 

SNV UVE-SPA 28 616 , 648 , 668 , 724 , 780 , 1200 , 1244 , 1288 , 
1324 , 1504 , 1548 , 1564 , 1588 , 1656 , 1668 , 
1684 , 1704 , 1730 , 1760 , 1824 , 1840 , 1852 , 
1860 , 1880 , 1884 , 1992 

GA 23 600 , 636 , 720 , 730 , 744 , 820 , 866 , 994 , 998 , 
1080 , 1126  , 1132 , 1146 , 1214 , 1218 , 1364 , 
1382 , 1498 , 1564 , 1652 , 1738 ,  1954 , 1980 

SNV+MSC UVE-SPA 26 672 , 692 , 708 , 816 , 1028 , 1200 , 1220 , 1288 , 
1328 ,1436 , 1512 , 1560 , 1576 , 1592 , 1608 , 
1656 , 1700 , 1740 , 1760 , 1824 , 1848 , 1856 , 
1864 , 1944 , 1948 , 1972 

GA 22 614 , 746 , 756 , 934 , 956 , 1050 , 1086 , 1114 , 
1170 , 1182 , 1350 , 1470 , 1490 , 1494 , 1592 , 
1698 , 1736 , 1766 , 1830 , 1862 , 1938 , 1984  

SNV+SG UVE-SPA 20 720 ,816 ,868 ,1196 ,1288 ,1496 ,1540 ,1576 , 
1596 ,1628 ,1652 ,1672 ,1700 ,1740 ,1816 ,1836 
,1852 ,1864 ,1872 ,1888  

GA 22 656 , 702 , 726 , 742 , 964 , 980 , 1000 , 1094 , 
1232 , 1328 , 1358 , 1414 , 1428 , 1582 , 1656 , 
1684 , 1806 , 1814 , 1828 , 1936 , 1942 , 2000 



 

PEER-REVIEWED ARTICLE  bioresources.com 

 

 

Wu et al. (2023). “FTIR classification of alfalfa hay,” BioResources 18(3), 5399-5416.  5409 

SNV+FD UVE-SPA 15 818 , 1154 , 1178 , 1542 , 1610 , 1634 , 1686 , 
1710 , 1730 , 1750 , 1766 , 1810 , 1826 , 1838 , 
1854 

GA 25 654 , 742 , 820 , 880 , 940 , 984 , 1014 , 1036 ,  
1070 , 1152  , 1186  , 1246 , 1332 , 1430 , 1474 , 
1496 , 1504 , 1520 , 1538 , 1568 , 1604 , 1636 ,   
1786 , 1952 , 1982  

SNV+SD UVE-SPA 23 862 , 922 , 946 , 974 , 998 , 1010 , 1022 , 1030 , 
1038 , 1062 , 1098 , 1154 , 1234 , 1242 , 1370 ,  
1386 , 1398 , 1406 , 1422 , 1430 , 1434, 1438 ,  
1954  

GA 18 646 , 770 , 778 , 788 , 920 , 932 , 1138 , 1194 ,  
1246 , 1360 , 1450 ,1554 , 1732 , 1746 , 1866 ,  
1888 , 1944 , 1986 

 

The dimensionality reduction by the genetic algorithm reduced the number 

of 701 waves to less than 50. Figure 7 is the characteristic band spectra after the 

fourth and fifth optimizations. After the fourth optimization, 47 feature bands were 

extracted, and after the fifth optimization, it was reduced to 23 characteristic bands. 

The specific characteristic spectra is shown in the Table. 

 

Table 2. Wavelength Selection after GA 

Method Cycle 
Times 

Number Wavelength (cm-1) 

GA 4 46 1980 , 1970 , 1954 , 1938 , 1920 , 1902 , 1776, 
1738 , 1704 , 1692 , 1652 , 1598 , 1584 , 1564, 
1514 , 1498 , 1494 , 1432 , 1382 , 1378 , 1364, 
1306 , 1238 , 1218 , 1214 , 1182 , 1146 , 1132, 
1126 , 1108 , 1106 , 1080 , 1030 , 998 , 994, 990 
, 960 , 944 , 866 , 820 , 808 , 744 , 730 , 720 , 
716 , 636 ,  600  

GA 5 23 1980 , 1954 , 1738 , 1652 , 1564 , 1498 , 1382, 
1364 , 1218 , 1214 , 1146 , 1132 , 1126 , 1106, 
1080 , 994 , 866 , 820 , 744 , 730 , 720 , 636 and 
600  

 
 

 
(a)                                                                         (b) 

 

Fig. 7. Result of characteristic variables by GA (a) After 4 cycles; (b) After 5 cycles 



 

PEER-REVIEWED ARTICLE  bioresources.com 

 

 

Wu et al. (2023). “FTIR classification of alfalfa hay,” BioResources 18(3), 5399-5416.  5410 

Construction of the Full Spectra Model 
The appropriate full spectra BP neural network and SVM models were 

created after pretreating the original spectra using various techniques. When using 

the full band for training, the specific parameters of the neural network were set as 

follows: the hidden layer was set to 10; the learning rate was set to 0.0001, and the 

number of iterations was set to 30. The results of this model are listed in Table 3. 

Using a back propagation neural network to classify full band alfalfa hay was able 

to achieve 100% classification results, when using SNV, FD, and SD pretreatment, 

but it took a long time for the model to function. The other two pretreatment 

techniques also produced positive outcomes.  

The classification results of the training set were 100%, 98.8%, and 98.2%, 

while the classification results of the prediction set were 100% because of the MSC 

preprocessing technique. With the exception of the natural drying method sample, 

which has a classification result of 98.7% when the SG processing method is 

applied, the training set's classification result was 100%. Table 4 shows the full 

spectrum classification results using support vector machines. In the prediction set, 

the classification accuracy of moldy dried alfalfa hay reached 100%, while the 

classification accuracy of the other two drying methods was not particularly good, 

with the minimum accuracy of 80.6%. 

 

Characteristic Wavelength Model 
Spectral data have high dimensionality and great correlation, so the model 

takes a long time. The authors used GA and SPA to extract the characteristic 

wavelengths from the entire spectrum using multiple pretreatment techniques to 

simplify the model, decrease the model’s running time, and improve classification 

results. The BP neural network and SVM model were then built using the extracted 

characteristic wavelengths. Table 5 and 6 provides the model's performance data. 

It can be seen from Tables that both models established by the characteristic 

band obtained relatively good classification results. The moldy alfalfa in the 

calibration set were both 100% recognized. In addition to SNV-FD-GA and SNV-

SD-GA, other moldy alfalfa in the prediction set were 100% identified with using 

BP model. In the SVM model, over half of the moldy alfalfa hay classification 

accuracy also reached 100% in the prediction set. All naturally dried alfalfa in the 

validation set, excluding SNV-MSC-UVE-SPA, SNV-FD-UVE-SPA, and SNV-

SG-UVE-SPA, were correctly identified at 100%; however, the recognition results 

of the prediction set were not very good. The recognition rate of most of the 

preprocessing methods reached more than 90%. In the model using SVM, the 

accuracy of the four preprocessing methods reached 100%, while the accuracy of 

the other methods remained above 80%. Most of the recognition performances of 

the shady dried alfalfa samples in the training set reached 100%, while most of the 

validation sets were also remained above 90%. In the SVM model, only the training 

set under SNV-FD-SPA processing achieved 100%, while the others were above 

89%, and most of the accuracy of the validation sets were above 80%. 

 

 

 

 

 



 

PEER-REVIEWED ARTICLE  bioresources.com 

 

 

Wu et al. (2023). “FTIR classification of alfalfa hay,” BioResources 18(3), 5399-5416.  5411 

Table 3. Result Using the Full Spectra by BP 

Pretreatment Method Number Calibration Set Prediction Set  

A B C A B C 

SNV N=701 100 100 100 100 100 100 

SNV-MSC N=701 100 98.8 98.2 100 100 100 

SNV-FD N=701 100 100 100 100 100 100 

SNV-SD N=701 100 100 100 100 100 100 

SNV-SG N=701 100 98.7 100 100 100 100 

* A represents mold-dried alfalfa, B represents sun-dried alfalfa, and C represents the 
shade-dried alfalfa 

 

Table 4. Result Using the Full Spectra by SVM 

Pretreatment Method Number Calibration Set Prediction Set  

A B C A B C 

SNV N=701 100 100 100 100 100 90.9 

SNV-MSC N=701 100 100 97.3 100 90 90.5 

SNV-FD N=701 100 100 100 100 100 85 

SNV-SD N=701 100 100 100 100 85.6 91.8 

SNV-SG N=701 100 100 100 100 80.6 97.6 

* A represents mold-dried alfalfa, B represents sun-dried alfalfa, and C represents the 
shade-dried alfalfa 

 

Table 5. Characteristic Wavelength by BP Model Data 

Pretreatment Method Parameter  Number Calibration Set Prediction Set 

A B C A B C 

SNV UVE-SPA 28 100 100 100 100 94.4 95.2 

GA 23 100 100 100 100 79.8 83.6 

SNV-MSC UVE-SPA 26 100 97.6 100 100 94.5 93.3 

GA 22 100 100 100 100 70.7 91 

SNV-FD UVE-SPA 20 100 98.9 98.8 100 97.4 93 

GA 22 100 100 100 94.7 94.7 96.7 
SNV-SD UVE-SPA 15 100 100 99.5 100 97.6 97.4 

GA 25 100 100 100 95 91.2 77.3 

SNV-SG UVE-SPA 23 100 98.7 97.3 100 94.9 87.8 

GA 18 100 100 100 100 93 78.1 

* A represents moldy dried alfalfa, B represents sun-dried alfalfa, and C represents the 
shade-dried alfalfa 

 

  



 

PEER-REVIEWED ARTICLE  bioresources.com 

 

 

Wu et al. (2023). “FTIR classification of alfalfa hay,” BioResources 18(3), 5399-5416.  5412 

Table 6. Characteristic Wavelength by SVM Model Data 

Pretreatment Method Parameter  Number Calibration Set Prediction Set 

A B C A B C 

SNV UVE-SPA 28 100 100 97.7 100 88.4 95.2 

GA 23 100 91.7 89.4 100 73.9 80.1 
SNV-MSC UVE-SPA 26 100 100 98.7 100 93.9 82 

GA 22 100 88.6 93.8 100 71.4 91 

SNV-FD UVE-SPA 20 100 100 100 97.4 96 85.2 

GA 22 100 88.5 97.8 94.8 91.7 94.4 

SNV-SD UVE-SPA 15 100 100 96.6 100 96.6 89.3 
GA 25 100 87.2 97 94.9 75 80.3 

SNV-SG UVE-SPA 23 100 100 96.4 100 94 82.6 

GA 18 100 83.9 89.1 94 76.9 78.6 

* A represents moldy dried alfalfa, B represents sun-dried alfalfa, and C represents the 
shade-dried alfalfa 

 

The characteristic band screened by SPA was more accurate than the one 

screened by the GA in the experiment, as demonstrated by the fact that the model 

designed using SPA performed higher than the design model after GA characteristic 

wavelength extraction, the minimum classification accuracy was 2.7% higher and 

the maximum was 23.8% higher. 

 

Model Comparison and Discussion 
In this study, different preprocessing methods were used to classify the full 

spectra data and characteristic spectra data by neural network, as presented in Tables 3 to 

6. The following results were obtained. 

The full-spectra model showed that most of the categories reached 100%, and 

the smallest reached 98.2% by BP and 80.2% by SVM. However, because of the 

time-consuming and laborious modelling procedure, the full spectra model is 

inappropriate for practical implementation. Under the same preprocessing method, 

comparing the classification results of full spectra and characteristic spectra, we will 

find that the results of full spectra are better because the information is relatively 

complete, but the classification accuracy of most characteristic spectra also reached 

more than 93% by BP, the classification accuracy of SVM is relatively poor, but 

most of it remained above 80%, and it was able to achieve the purpose of real-time 

processing. Among all the characteristic models studied in this study, under SNV-

SD-UVE-SPA pretreatment method both models displayed best performance, 15 

characteristic spectra were extracted. The classification accuracy of moldy-dried 

alfalfa, sun-dried alfalfa, and shade-dried alfalfa in the training set were 100%, 

100%, and 99.5%, etc., and the classification results of the prediction sets were 

100%, 97.6%, and 97.4%, etc. The classification results for the SVM prediction set 

were 100%, 100%, and 96.6%, etc., and the classification results of the prediction 

sets were 100%, 96.6%, and 89.3%, etc. Under the same preprocessing mode, the 

classification accuracy of most neural networks was higher than that of support 

vector machine. 

 Researchers have also used spectral technologies to conduct several studies 

on the quality of alfalfa hay for a long time. For example, this approach has been 

used to predict the nutritional components of alfalfa hay, such as crude protein, 



 

PEER-REVIEWED ARTICLE  bioresources.com 

 

 

Wu et al. (2023). “FTIR classification of alfalfa hay,” BioResources 18(3), 5399-5416.  5413 

neutral detergent fiber, and acid detergent fiber, etc. (Zapata et al.  2021;Guo et al.  

2020). Up to now, there have been few studies on the classification of alfalfa hay 

with different dried methods by mid-infrared spectroscopy. 

 

 
CONCLUSIONS 
 

1. In this study, moldy dried alfalfa, natural sun-dried alfalfa, and shade-dried alfalfa were 

used as the research objects. Through the use of standard normal variable 

transformation (SNV), as well as versions with multiplicative scatter correction (SNV-

MSC), first derivatives (SNV-FD), second derivatives (SNV-SD), and Savitzky-Golay 

convolution smoothing (SNV-SG) pretreatment methods, a back propagation (BP) 

neural network and SVM  were constructed based on the full spectra. To improve the 

performance of the model, successive projection algorithm (SPA) and genetic 

algorithm (GA) routines were used to extract the characteristic wavelengths for the 

model. The SNV-SD-SPA-BP model had the best performance, with classification 

accuracy of calibration set of 100%, 100%, and 99.5%, and prediction set of 100%, 

97.6%, and 97.4%, respectively using BP model. 

2. Both calibration set and prediction set were able to identify 100% of the moldy 

alfalfa. The findings of this study offer a theoretical framework for alfalfa hay 

quality control and real-time identification. In the future, more Artificial 

Intelligence methods can be used to improve the classification accuracy of the 

system. 

 

 

Author Contributions 
Xiaoqing Wu is responsible for conceptualization, data curation, formal 

analysis, methodology, software, and writing-original draft. Guifang Wu is 

attributed to funding acquisition, resources.  Bo Wang helped with data curation. Jie 

Li helped with the formal analysis. 

 

Conflicts of interest 
The authors declare that there are no conflicts of interest regarding the 

publication of this study. 

 

ACKNOWLEDGEMENTS 
 

This study was supported by the National Natural Science Foundation of 

China (32060414), National Natural Science Foundation of China (62061037), Inner 

Mongolia Natural Science Foundation (2018MS06012), and Research Fund for the 

Doctoral Program of Higher Education (Project No: 20111515120004). 

 

  



 

PEER-REVIEWED ARTICLE  bioresources.com 

 

 

Wu et al. (2023). “FTIR classification of alfalfa hay,” BioResources 18(3), 5399-5416.  5414 

REFERENCES CITED 
 

Chang, C. C., and Lin, C. J. (2007). “LIBSVM: A library for support vector machines,” 

CM Transactions on Intelligent Systems and Technology 2(27), 1-27. DOI: 

10.1145/1961189.1961199 

Cheng, Q. M., Ge, G. T., and Yin, Q. (2018). “Effect of moisture, bale density and CaO 

on storage life and quality of alfalfa hay,” Acta Prataculturae Sinica 27(5), 190-200. 

DOI: 10.11686/cyxb2017262 

Chu, X., Wang, W., and Zhang, L. D. (2014). “Determination of aflatoxin on the surface 

of corn grain by hyperspectral optimal wavelength selection and Fisher discriminant 

analysis,” Spectroscopy and Spectral Analysis (07), 1811-1815. 

Coblentz, W. K., Fritz, J. O., and Bolsen, K. K. (1996). “Quality changes in alfalfa hay 

during storage in bales,” Dairy Science 79(5), 873-885. DOI: 10.3168/jds.S0022-

0302(96)76436-6 

Darabighane, B., Aghjehgheshlagh, F. M., and Mahdavi, A. (2020). “Replacing alfalfa 

hay with dry corn gluten feed alters eating behavior, nutrient digestibility, and 

performance of lactating dairy cows,” Italian Journal of Animal Science 19(1), 1266-

1276. DOI: 10.1080/1828051X.2020.1830722 

Gfrerer, M., Chen, S., and Lankmayr, E. P. (2004). “Comparison of different extraction 

techniques for the determination of chlorinated pesticides in animal feed,” Analytical 

and Bioanalytical Chemistry 378, 1861-1867. DOI: 10.1007/s00216-004-2492-6 

Guo, T., Huang, Y. Q., Guo, L., Li, F. D., Pan, F. M., Zhang, Z. J., and Li, F. (2020). 

“Rapid prediction of nutrient content of alfalfa hay by near infrared spectroscopy,” 

Pratacultural Science 37(11), 2374-2381. 

Han, K. J., Collins, M., and Vanzant, E. S. (2004). “Bale density and moisture effects on 

alfalfa round bale silage,” Crop Science 44(3), 914-919. DOI: 

10.2135/cropsci2004.0914 

He, Y., Zhang, L., Wu, X. J., Zheng, A. R., Liu, W., He, Y. H., Niu, Y., Wang, Y. X. and 

Zhang, X. X. (2019). “Establishment of near infrared prediction model for 

conventional nutrient content of alfalfa hay,” Chinese Journal of Animal Nutrition 

31(10), 4684-4690. DOI: 10.3969/i.issn.1006-267x.2019.10.033 

Hell, J., Prückler, M., and Danner, L. (2016). “A comparison between near-infrared 

(NIR) and mid-infrared (ATR-FTIR) spectroscopy for the multivariate determination 

of compositional properties in wheat bran samples,” Food Control 60, 365-369. DOI: 

10.1016/j.foodcont.2015.08.003 

Ji, R., Jiang, Z., Wang, X., Han, Y., Bian, H., Yang, Y., Zhuang, L., and Zhang, Y. 

(2022). “Detection of captan residues in apple juice using fluorescence spectroscopy 

combined with a genetic algorithm and support vector machine,” Applied Optics 

61(12), 3455-3462. DOI: 10.1364/AO.451831 

Josiah, D. C., Ellie, J., Patrick, C. H. M., Paul, R. K., and Mark, R. W. (2018). “Mid-

infrared reflectance spectroscopy as a tool for forage feed composition prediction,” 

Animal Feed Science and Technology 241, 102-111. DOI: 

10.1016/j.anifeedsci.2018.04.022 

Kamruzzaman, M., Makino, Y., and Oshita, S. (2016). “Hyperspectral imaging for real-

time monitoring of water holding capacity in red meat,” LWT - Food Science and 

Technology 66, 685-691. DOI: 10.1016/j.lwt.2015.11.021 



 

PEER-REVIEWED ARTICLE  bioresources.com 

 

 

Wu et al. (2023). “FTIR classification of alfalfa hay,” BioResources 18(3), 5399-5416.  5415 

Kaya, T., and Huck, C. W. (2017). “A review of mid-infrared and near-infrared 

imaging: Principles, concepts and applications in plant tissue analysis,” Molecules 

22(1), Article Number 168. DOI: 10.3390/molecules22010168 

Kumar, R., Kumar, V., and Sharma, V. (2017). “Fourier transform infrared spectroscopy 

and chemometrics for the characterization and discrimination of writing/photocopier 

paper types: Application in forensic document examinations,” Spectrochimica Acta 

Part A: Molecular & Biomolecular Spectroscopy 170, 19-28. DOI: 

10.1016/j.saa.2016.06.042 

Kuronuma, K., Kawamorita, T., and Yagi, T. (2020). “Application of peripheral near 

infrared spectroscopy to assess risk factors in patient with coronary artery disease: 

Part 2,” Advances in Experimental Medicine and Biology 1232, 355-360. DOI: 

10.1007/978-3-030-34461-0_45 

Lim, J., Watanabe, N., Yoshitoshi, R., and Kawamura, K. (2020). “Simple in-field 

evaluation of moisture content in curing forage using normalized difference 

vegetation index (NDVI),” Grassland Science 66, 238-248. DOI: 

10.1111/grs.12275 

Lorenzo, C. D., García-Gagliardi, P., Antonietti, M. S., Sánchez-Lamas, M., Mancini, E., 

and Dezar, C. A. (2020). “Improvement of alfalfa forage quality and management 

through the down-regulation of MsFTa1” Plant Biotechnology Journal 18(4), 944-

954. DOI: 10.1111/pbi.13258 

Mário, C. U. A., Saldanha, T. C. B., and Galvo, R. K. H. (2001). “The successive 

projections algorithm for variable selection in spectroscopic multicomponent 

analysis,” Chemometrics and Intelligent Laboratory Systems 57(2), 65-73. DOI: 

10.1016/S0169-7439(01)00119-8 

Mohebby, B. (2010). “Application of ATR infrared spectroscopy in wood acetylation,” 

Journal of Agricultural Science & Technology 10(3), 253-259. DOI: 

10.1016/j.compag.2008.01.008 

Rahman, A., Kondo, N., Ogawa, Y., and Kanamori, K. (2016). “Determination of K 

value for fish flesh with ultraviolet-visible spectroscopy and interval partial least 

squares (iPLS) regression method,” Biosystems Engineering 141, 12-18. 

Rumelhart, D. E., Hinton, G. E., and Williams, R. J. (1986). “Learning representations by 

back-propagating errors,” Nature 323(6088), 533-536. DOI: 10.1038/323533a0 

Shen, F., and Huang, Y. (2019). “On-line detection of corn mildew degree based on 

spectral and image information fusion,” Food Science 40(16), 274-280.  

Tanaka, K., Yamaguchi, N., and Baba, T. (2011). “Rapid enumeration of low numbers of 

moulds in tea based drinks using an automated system,” International Journal of 

Food Microbiology 145(1), 365-369. DOI: 10.1016/j.ijfoodmicro.2011.01.012 

Undugodage, Don, Nuwan, Perera, Koichi and Nishikida. (2018). “Development of 

infrared library search prefilters for automotive clear coats from simulated attenuated 

total reflection (ATR) spectra,” Applied Spectroscopy 72(6), 886-895.  

DOI: 10.1177/0003702818759664 

Vanzant, E. S., Cochran, R. C., and Jacques, K. A. (1990). “Influence of level of 

supplementation and type of grain in supplements on intake and utilization of 

harvested, early-growing-season, bluestem-range forage by beef steers,” Journal of 

Animal Science 68(5), 1457-1468. DOI: 10.1016/0168-1591(90)90143-2 

Wallén, S. E., Prestløkken, E., Meuwissen, T. H. E., McParland, S., and Berry, D. P. 

(2018). “Milk mid-infrared spectral data as a tool to predict feed intake in lactating 



 

PEER-REVIEWED ARTICLE  bioresources.com 

 

 

Wu et al. (2023). “FTIR classification of alfalfa hay,” BioResources 18(3), 5399-5416.  5416 

Norwegian red dairy cows,” Journal of Dairy Science 101, 1-12. DOI: 

10.3168/jds.2017-13874 

Wang, Y. R., Hampton, J. G., and Sun, J. (1996). “Effect of topography and phosphorus 

on seed yield and quality of alfalfa in China,” Journal of Applied Seed Production 14, 

53-57.  

Xiong, Y. T., Li, Z. P., and Wang, J. (2016). “The near infrared spectral bands optimal 

selection in the application of liquor fermented grains composition analysis,” 

Spectroscopy and Spectral Analysis 36(1), 84-90. DOI: 10.3964/j.issn.1000-

0593(2016)01-0084-07 

Zapata, F., López-Fernández, A., and Ortega-Ojeda, F. (2021). “Introducing ATR-FTIR 

spectroscopy through analysis of acetaminophen drugs: Practical lessons for 

interdisciplinary and progressive learning for undergraduate students,” Journal of 

Chemical Education 98(8), 2675-2686. DOI: 10.1021/acs.jchemed.0c01231 

Zhou, J., Yang, Y., and Yao, Y. (2022). “Application of mid-infrared spectroscopy in the 

analysis of key indexes of strong flavour Chinese spirits base liquor,” Spectroscopy 

and Spectral Analysis 42(3), 764-768. DOI: 10.3964/j.issn.1000-0593(2022)03-

0764-05 

 

Article submitted: February 21, 2023; Peer review completed:  June 3, 2023; Revised 

version received and accepted: June 23, 2023; Published: June 28, 2023. 

DOI: 10.15376/biores.18.3.5399-5416