نوع مقاله : مقاله پژوهشی
1 گروه مهندسی بیوسیستم، دانشکده کشاورزی و منابع طبیعی، دانشگاه محقق اردبیلی، اردبیل، ایران
2 گروه مهندسی بیوسیستم، دانشکده کشاورزی دانشگاه نامیک کمال، ترکیه
3 گروه مهندسی ماشین های کشاورزی و فناوری، دانشکده کشاورزی دانشگاه هجده مارس چاناکاله، ترکیه
4 دانشجوی دکتری، گروه مهندسی بیوسیستم، دانشکده کشاورزی و منابع طبیعی، دانشگاه محقق اردبیلی، اردبیل، ایران
عنوان مقاله [English]
Potato tubers are one of the most important sources of nutrition in most countries. This product is the world's fourth important food crop after wheat, rice and maize because of its higher yield potential along with high nutritive value.
Soluble solids content (SSC) is one of the quality parameters of potatoes and it play an important role in quality and commercial value of potato. The amount of SSC in potatoes varies greatly depending on the cultivars, growing environment and storage conditions.
Today, the development of fast, non-destructive, accurate and online technique identification methods to determine product quality is strongly felt. One method for this purpose is Near-infrared spectroscopy. The high speed and accuracy of near-infrared spectroscopy to determine of vegetables and fruits quality in high tonnage, has led to the use of this method in many grading and quality control systems.
Shao, et al., Using reflection Near-infrared spectroscopy for investigating the qualitative properties of tomatoes, including firmness, SSC and treatable acidity, and predicted these properties to be non-destructive with a high correlation coefficient. A study was conducted by Saad, et al. to evaluate the non-destructive quality of stored tomatoes using vis/NIR absorption spectroscopy in wavelength range from 350 to 1050 nm. In another study, the main compounds of potatoes were determined by near-infrared spectroscopy by Ainara Lopez. The sugar content of potato tubers was determined using visible and near infrared spectroscopy by Ji Yu Chen et al.Khodabakhshian et al., using reflection Near-infrared spectroscopy for classify the maturity stage and to predict the quality attributes of pomegranate, Principal component analysis was used to distinguish among different maturities and several preprocessing methods were utilized including centering, smoothing (Savitzky–Golay algorithm, median filter), normalization (multiplicative scatter correction and standard normal variate) and differentiation (first derivative and second derivative). The results of this study concluded that different preprocessing techniques had effects on the classification performance of the model using the principal component analysis method.
The aim of this study was to investigate the possibility of using near-infrared spectroscopy (vis/NIR) in estimating changes of SSC in three potato cultivars during storage under different storage conditions.
Three potato cultivars, including SANTE, MARABEL and GRANA, which were different in terms of yield, crop quality and shelf life, were prepared from a farm near the University of Onsekiz Mart Çanakkale in Turkey and were stored under different storage conditions. Required values of each cultivar was stored under three different conditions including 4 ̊C temperature with relative humidity of 90%, 7 ̊C temperature with relative humidity of 80-90% and 22-28 ̊C variable temperature with relative humidity of 70-90%, for 65 days.
Spectroscopy was performed using a multi-purpose analyzer (MPA) FT-NIR spectrometer in the range of 800-2500 nm.
To measure the SSC in the samples, 7 samples from each potato cultivar were randomly selected and kept in the laboratory for 2 hours at 22 ̊C before analysis. After peeling the potatoes, the juice of the samples was extracted by using a common juicer and then filtered through filter paper to remove any remaining pulps. The amount of SSC in the samples was measured by using a digital refractometer at 25 ̊C as a ̊Brix.
In order to remove the outlier data, principal component analysis (PCA) was used before any processing on the data. Moving averages, Multiple Gaussian Fitting Regressions, Median filter, Savitzky-Golay smoothing, normalization, Multiplicative Scatter Correction (MSC) and Standard Normal Variety Transformation (SNV) were applied to the data and compared.
Partial least squares regression (PLS) models for all Pre-processed data were extracted and the statistical indicators include correlation coefficient (R2) and Root Mean Square Error (RMSE) were used to find the best model. To extract the models, the data were randomly divided into two parts: 80% of the data was used for training and cross-validation and the rest of the data was used for independent validation.
To select the effective wavelengths and reduce the wavelengths to a limited number, the regression coefficients of the best calibration model obtained from (PLS) model were used.
In order to find the best fitting model for the relationship between effective wavelengths and SSC changes of potatoes during storage, Partial least squares regression (PLS), Principal Component Regression (PCR), Multiple Linear Regression (MLR), Support Vector Machine (SVM) and Artificial Neural Network (ANN) were employed.
This research were performed with factorial design on a completely randomized design (CRD) base include cultivar in three levels (SANTE, MARABEL and GRANA), storage temperature in three levels (4, 7 and 20 ̊C) and storage time in four levels (20, 34 , 46 and 62 days) with five replications. The statistical analysis was performed using the Unscrambler X 10.4 statistical software.
According to results, the amount of SSC was increases during storage. This increasing trend during storage can be attributed to the increase in sugar content during the ripening process and the hydrolysis of starch to maintain the physiological activities of the crop in the postharvest period.
The results also show that the raw and all Pre-processed spectral data are able to predict SSC with acceptable accuracy. The best results were obtained from the Median filter Pre-processed model with RMSEC = 0.410, R2c = 0.875, RMSECV = 0.420, R2CV = 0.867.
Based on the regression coefficients of best the (PLS) model, 12 wavelengths in the range of 1400 to 2500 nm were identified as effective wavelengths. Based on distribution of main overtone bond, the discriminated of potato samples according to SSC quality index can be attributed second overtones of C-H, CH2 and CH3 in the wavelength range from 1400 to 1500 nm, the first overtones of C-H, CH2 and CH3 in the wavelength range from 1500 to 2100 nm and composition overton of these bonds in the wavelength range from 2200 to 2500 nm which all this overtone are present in the structure of hydrocarbons such monomeric sugars (glucose and xylose) and oligosaccharides (starch).
According to the results, all models are able to predict SSC of potatoes during storage based on effective wavelengths with acceptable accuracy but among these models, the Artificial Neural Network (ANN) model with RMSEC = 0.215, R2c = 0.974, RMSECV = 0.927, R2CV = 0.280 has the highest accuracy to prediction of SSC of potatoes during storage.