RMSEP Values Research Articles

Achieving robustness across different ages and cultivars for an NIRS-PLSR model of fresh cassava root starch and dry matter content

• Starch content and dry matter content models for fresh cassava tubers were optimized. • NIR spectroscopy was used to predict starch and dry matter content in fresh cassava roots under field conditions. • Spectral variable selection was carried out using SPA and GA methods. • The starch and dry matter content was evaluated under field conditions for breeding purposes. • New spectral processing methods improved the model’s accuracy. This study used a portable near-infrared (NIR) spectrometer at wavelengths of 570–1031 nm to evaluate starch content (SC) and dry matter content (DMC) in fresh cassava tubers. An improved model was developed for the prediction of cassava tuber quality. The cassava samples were taken from four main varieties: CMR38-125-77, KU50, RY11, and RY9. The samples were obtained 4–12 months after planting (MAP). Partial least squares (PLR) regression was combined with different variable selection methods and spectral pre-treatment. Their accuracies were then compared. Variable selection methods included the successive projections algorithm (SPA) and the genetic algorithm (GA). The NIR spectra were obtained in the interactance mode under field conditions. The GA wavelengths combined with sequential pre-processing by orthogonalization (SPORT) pre-processing provided the optimum model for predicting both the SC and the DMC of cassava. The R 2 p, RMSEp, and RPD of SC were 0.91, 1.76%, and 3.26, respectively, and those of DMC were 0.75, 2.00%, and 2.00, respectively. The most effective model was tested against unknown samples of newly developed varieties obtained from different harvest seasons, yielding RMSEp and bias values of 2.37% and −9.178 × 10 -6 %, respectively, for SC. For DMC, the RMSEp and bias values were 2.67% and 4.16 × 10 -14 %, respectively. The results suggest that the calibration model could be used to monitor the internal quality of cassava tubers in the field. The variety, age, position, and section of the tubers had a slight influence on the prediction performance; however, the prediction accuracy was acceptable for in-field applications. The in-field portable NIR spectrometer could become a new tool for breeders, saving time and costs. Breeders could evaluate SC without destroying the cassava roots or stalks and could correct and inspect the behaviour of the SC and DMC accumulation.

Application of Grey Relational Analysis and Multiple Linear Regression to Establish the Cutting Force Model of Oil Peony Stalk

Oil peony is an important oil crop, which has high quality and oil content. In order to improve the cutting quality and harvest efficiency of oil peony, the cutting equipment of the pruning machine and harvester is a key component. Also, the accurate prediction of cutting force of oil peony stalk is one of the essential processes for the design of the cutting equipment. In this article, to accurately predicted the cutting force of the stalk, the physical property parameters and chemical components were considered as influencing factors, which were used to establish the model of mechanical property parameter of oil peony stalk. The physical property parameters of oil peony stalk included the stalk diameter, internode distance, fresh weight, dry weight, relative moisture content, volume, fresh density, and dry density. The chemical components of the stalk were cellulose, hemicellulose, and lignin. Besides, the modeling methods, which were the partial least squares regression (PLSR), principal component analysis (PCA) couple with multiple linear regression (MLR), and grey relational analysis (GRA) couple with MLR, were used to optimize the multiple parameters (physical property parameters and chemical components). The results showed that the internode distance and relative moisture content had significant effects on the cutting force of oil peony stalk. The R c 2 and R p 2 values of the GRA (0.5) + MLR method were 0.801 and 0.820, and RMSEC and RMSEP values were 2.862N and 4.715N, respectively. Consequently, the GRA + MLR method could be used to predict the cutting force of oil peony stalk, which was an important basis for the design of precision cutting equipment.

Rancidity and moisture estimation in shelled almond kernels using NIR hyperspectral imaging and chemometric analysis

The current work focused on rapid and non-destructive determination of moisture content (MC), free fatty acids (FFA) and peroxide value (PV) in shelled almonds, using reflectance NIR-hyperspectral imaging (HSI). Sample set of 354 and 235 almond kernels were treated to vary the MC, rancidity (FFA and PV), respectively and used for model development. The sample set for each response was divided into calibration and testing set in the ratio of 70:30. Reference MC, FFA and PV were measured using wet analysis methods and their association with the spectral data were modelled using partial least squares regression. Superior models were obtained from the full-spectrum (900–1700 nm) data with Rp2 values of 0.957, 0.970, 0.955 and RMSEP values of 0.41%, 0.108%, 0.453 mEq for MC, FFA and PV, respectively. Competitive adaptive reweighted sampling method was used to select the feature wavelengths for rapid quantification. Multiple linear regression models were developed using the feature wavelengths having good predictability (Rp2 values: 0.941, 0.903, 0.886 and RMSEP values: 0.494%, 0.162%, 0.658 mEq for MC, FFA and PV, respectively). The current findings demonstrated great feasibility in industrial deployment of HSI technique for non-destructive estimation of moisture and rancidity indices in almonds and other nuts.

Quantitative analysis of blended corn-olive oil based on Raman spectroscopy and one-dimensional convolutional neural network

Blended vegetable oil is a vital product in the vegetable oil market, and quantifying high-value vegetable oil is of great significance to protect the rights and interests of consumers. In this study, we established a one-dimensional convolutional neural network (1D CNN) quantitative identification model based on Raman spectra to identify the amount of olive oil in a corn-olive oil blend. The results show that the 1D CNN model based on 315 extended average Raman spectra can quantitatively identify the content of olive oil, with R2p and RMSEP values of 0.9908 and 0.7183 respectively. Compared with partial least squares regression (PLSR) and support vector regression (SVR), although the index is not optimal, it provides a new analytical method for the quantitative identification of vegetable oil.

The Application of FTIR Spectroscopy and Chemometrics for the Authentication Analysis of Horse Milk.

Expensive milk such as horse's milk (HM) may be the target of adulteration by other milk such as goat's milk (GM) and cow's milk (CM). FTIR spectroscopy in combination with chemometrics of linear discriminant analysis (LDA) and multivariate calibrations of partial least square regression (PLSR) and principal component regression (PCR) was used for authentication of HM from GM and CM. Milk was directly subjected to attenuated total reflectance (ATR) spectral measurement at midinfrared regions (4000-650 cm−1). Results showed that LDA could make clear discrimination between HM and HM adulterated with CM and GM without any misclassification observed. PLSR using 2nd derivative spectra at 3200-2800 and 1300-1000 cm−1 provided the best model for the relationship between actual values of GM and FTIR predicted values than PCR. At this condition, R2 values for calibration and validation models obtained were 0.9995 and 0.9612 with RMSEC and RMSEP values of 0.0093 and 0.0794. PLSR using normal FTIR spectra at 3800-3000 and 1500-1000 cm−1 offered R2 for the relationship between actual values of CM and FTIR predicted values of >0.99 in calibration and validation models with low errors of RMSEC of 0.0164 and RMSEP of 0.0336 during authentication of HM from CM. Therefore, FTIR spectroscopy in combination with LDA and PLSR is an effective method for authentication of HM from GM and CM.

Genetic algorithm based artificial neural network and partial least squares regression methods to predict of breakdown voltage for transformer oils samples in power industry using ATR-FTIR spectroscopy

The current study proposes a novel analytical method for calculating the breakdown voltage (BV) of transformer oil samples considered as a significant method to assess the safe operation of power industry.Transformer oil samples can be analyzed using the Attenuated total reflection Fourier transform infrared (ATR-FTIR) spectroscopy combined with multivariate calibration methods. The partial least squares regression (PLSR) back propagation–artificial neural network (BP-ANN) methods and a genetic algorithm (GA) for variable selection are used to predict and assess breakdown voltage in transformer oil samples from various Iranian transformer oils.As a result, the root mean square error (RMSE) and correlation coefficient for the training and test sets of oil samples are also calculated. In the GA-PLS-R method, the squared correlation coefficient (R2pred) and root mean square prediction error (RMSEP) are 0.9437 and 2.6835, respectively. GA-BP-ANN, on the other hand, had a lower RMSEP value (0.2874) and a higher R2pred function (0.9891).Considering the complexity of transformer oil samples, the performance of GA-BP-ANN has resulted in an efficient approach for predicting breakdown voltage; consequently, it can be effectively used as a new method for quantitative breakdown voltage analysis of samples to evaluate the health of transformer oil..

Assessment of geographical origin of virgin coconut oil using inductively coupled plasma mass spectrometry along with multivariate chemometrics

Recently, Virgin coconut oil (VCO) has emerged as one of the most favorable edible oils because of its application in cooking, frying as well as additive used in food, pharmaceuticals, and cosmetic goods. These qualities have established VCO in high consumer demand and there is a great need of establishing a reliable method for the identification of its geographical origin. Through this present study, for the first time, it has been established that Inductively Coupled Plasma-Mass-Spectrometry (ICP-MS) combined with multivariate chemometrics can be used for the identification of the geographical origin of the VCO samples of various provinces. Principal Component Analysis (PCA), and Linear Discriminant Analysis (LDA) were able to differentiate and classify the VCO samples of different geographical origins. Further, calibration models (Principal Component Regression and Partial Least Square Regression) were developed on the calibration dataset of the elemental concentration obtained from the ICP-MS analysis. An external dataset was used to develop the prediction model to predict the geographical origin of an unknown sample. Both PCR and PLS-R models were successfully able to predict the geographical origin with a high R2 value (0.999) and low RMSEP value 0.074 and 0.075% v/v of prediction respectively. In conclusion, ICP-MS combined with regression modelling can be used as an excellent tool for the identification of the geographical origin of the VCO samples of various provinces. This whole technique is the most suitable as it has high sensitivity as well as provides easy multi-metal analysis for a single sample of edible oil.

Improved modelling for low-correlated multiple responses by common-subset-of-independent-variables partial-least-squares

In this study, a new approach for PLS modelling for low-correlated multiple responses, called Common-Subset-of-Independent-Variables Partial-Least-Squares, denoted as CSIV-PLS1, is proposed and evaluated. In CSIV-PLS1, for each response vector, individual PLS1 models with individual model complexities are developed, based on one common set of independent variables, obtained after variable selection by the Final Complexity Adapted Models method, using the absolute values of the PLS regression coefficients, denoted as FCAM-REG.CSIV-PLS1 combines a common variable set for all response vectors, which is a characteristic of PLS2, with the individual model complexity for each response, which is a characteristic of PLS1. These characteristics make CSIV-PLS1 more flexible than PLS2.The selective and predictive abilities of the proposed CSIV-PLS1 method are investigated using one simulated and four real data sets with low-correlated multiple responses from different sources. The simulated data set is used to test the general applicability of the CSIV-PLS1 method. The predictive abilities, measured by the RMSEP values, resulting from CSIV-PLS1 models, are statistically compared with those of the corresponding PLS1 and PLS2 models, using one-tailed paired t-tests.The selective ability of the CSIV-PLS1 method is good, because mostly variables with an informative meaning to the responses are selected. The RMSEP values resulting from the CSIV-PLS1 method are (i) significantly lower at the 95% confidence level than those of the corresponding PLS2 method, and (ii) borderline significantly lower at the 90–95% confidence level than those of the corresponding PLS1 methods. In case of low-correlated multiple responses, the predictive ability of the CSIV-PLS1 method is significantly better than that of the PLS2 method, and borderline significantly better than those of the corresponding PLS1 methods. Therefore, CSIV-PLS1 modelling may be an alternative for PLS1 or PLS2.

Application of SWIR hyperspectral imaging coupled with chemometrics for rapid and non-destructive prediction of Aflatoxin B1 in single kernel almonds

Almonds are highly susceptible to Aflatoxin B1 (AFB1) contamination, which can result in significant economic losses. Current detection techniques are destructive, time consuming and unfit for in-line application. This study investigated the potential of hyperspectral imaging in the near infrared (NIR) range (900–1700 nm) to develop a rapid and non-destructive protocol for determination of AFB1 content of single almond kernels. Almond kernels were treated to varying AFB1 concentration by artificial infection using standard AFB1 solution and used for Experimentation. Reference AFB1 concentration and their association with the spectral data were modelled using partial least squares regression (PLSR) combined with suitable spectral preprocessing techniques. Superior models were obtained with full-spectrum regression models with R2 and RMSEP values of 0.958 and 0.089 μg/g, respectively. Competitive-adaptive reweighted sampling (CARS) was used to select the feature wavelengths for rapid quantification of AFB1 at commercial scale. Multiple linear regression (MLR) models were developed using the selected wavelengths for its applicability in multispectral imaging systems. Results showed that the MLR model achieved good prediction capabilities with R2 and RMSEP values of 0.948 and 0.090 μg/g, respectively. The results demonstrated that hyperspectral imaging had potential for rapid and nondestructive determination of AFB1 concentration in single kernel almonds.

UV-Vis Spectrophotometry and UPLC–PDA Combined with Multivariate Calibration for Kappaphycus alvarezii (Doty) Doty ex Silva Standardization Based on Phenolic Compounds

The algae Kappaphycus alvarezii is considered an important raw material for industrial practices, producing high economic value of various derived products. However, the quality of this commodity, which can be indicated by the level of phenolic compounds, may vary due to growth factors, including cultivation sites. An analytical UV-Vis spectrophotometry method coupled with chemometrics was proposed to standardize the red alga based on the content of phenolic compounds. The correlation between the UV-Vis spectra and UPLC–PDA results, combined with a multivariate calibration of the K. alvarezii extracts, was analyzed. The extracts were prepared using an ultrasound-based technique and subsequently subjected to UV-Vis spectral measurements at 200–800 nm and UPLC–PDA at 260 and 330 nm. Chemometric techniques and partial least squares (PLS) were applied to the acquired data to build a reliable analysis of the phenolics in the K. alvarezii extracts. The result showed that the wavelength combination of 200–450 and 600–690 nm provided a valid method for quantitative analysis of the studied phenolics that belong to hydroxybenzoic acid, hydroxycinnamic acid, and flavonoid with a coefficient of regression (R2) > 0.96 in the calibration and validation models, along with an RMSEC and RMSEP value < 8%. The method was then employed to characterize the K. alvarezii samples from 13 different cultivation areas. Principal component analysis (PCA) generated principal components that produced a clear distribution among the samples of K. alvarezii based on phenolic compounds corresponding to the geographical origin.

Variable selection in near-infrared spectra: Application to quantitative non-destructive determination of protein content in wheat

With the development of information technology, online non-destructive quantitative testing technology for wheat has attracted widespread attention. Protein content is an important indicator for wheat. A near infrared (NIR) spectroscopy system was utilized for the quantitative detection of protein in wheat. Firstly, the spectra of wheat were obtained by the NIR spectroscopy system. Then, variable selection methods were used to optimize the characteristic wavelengths, including competitive adaptive reweighted sampling (CARS), variable combination population analysis (VCPA), Monte Carlo variable combination population analysis (MCVCPA), and automatic weighting variable combination population analysis (AWVCPA). AWVCPA algorithm was effective in eliminating uninformative and inference wavelength variables of NIR spectra and reduced the complexity of quantitative prediction model. The characteristic wavelengths selected by AWVCPA were highly related to the absorbance of protein. Finally, the partial least squares (PLS) was employed to construct quantitative detection models based on different variable selection method. The experimental results showed that the AWVCPA-PLS prediction model is the best among the four tested models. The R2, RMSEP, RPD and RER values of AWVCPA-PLS were 0.9753, 0.0934, 10.15 and 38.54 respectively, indicative of high prediction accuracy and robust performance. This study provides an important theoretical basis for evaluating wheat in food engineering region.

Predicting Drug Release from 3D Printed Oral Medicines Based on the Surface Area to Volume Ratio of Tablet Geometry.

3D printing offers the advantage of being able to modify dosage form geometry, which can be exploited to modify release characteristics. In this study, we investigated the influence of the surface area to volume ratio (SA/V) to change and predict release profiles of 3D printed dosage forms. Geometries with varying SA/V and dosages were designed and printed, and drug dissolution was investigated. Three drug substances were used: pramipexole, levodopa (both BCS I) and praziquantel (BCS II). Two polymers were chosen as matrix formers: polyvinyl alcohol (water-soluble) and ethylene vinyl acetate (inert). Drug release was characterized using the mean dissolution time (MDT) and established equations that describe complete dissolution curves were applied. Predictions were validated with previously un-printed dosage forms. Based on an identified MDT-SA/V correlation, the MDT can be predicted with a deviation of ≤5 min for a given SA/V. Using correlations of fit parameters and SA/V, RMSEP values of 0.6–2.8% and 1.6–3.4% were obtained for the BCS I formulations and RMSEP values of 1.0–3.8% were obtained for the BCS II formulation, indicating accurate prediction over a wide range of dissolution profiles. With this approach, MDT and release profiles of dosage forms with a given SA/V can be precisely predicted without performing dissolution tests and vice versa, the required SA/V can be predicted for a desired release profile.

Direct quantification of red wine phenolics using fluorescence spectroscopy with chemometrics

Phenolic compounds are secondary metabolites known to play crucial roles in important chemical reactions impacting the mouthfeel, colour and ageing potential of red wine. Their complexity has resulted in a number of advanced analytical methods, which often prevent routine phenolic analysis in winemaking. Fluorescence spectroscopy could be an alternative to current spectrophotometric techniques and its combination with chemometrics was investigated for its suitability in directly quantifying phenolic content of unaltered red wine and fermenting samples. Front-face fluorescence was optimised and used to build predictive models for total phenols, total condensed tannins, total anthocyanins, colour density and polymeric pigments. Machine learning algorithms were used for model development. The most successful models were built for total phenols, total condensed tannins and total anthocyanins with coefficient of determination (R2cal) and RMSECV of 0.81, 0.89, 0.80 and 5.71, 104.03 mg/L, 60.67 mg/L, respectively. The validation results showed R2val values of 0.77, 0.8 and 0.77, and RMSEP values of 7.6, 172.37 mg/L and 76.57 mg/L, respectively. A novel approach for the classification of South African red wine cultivars based on unique fluorescent fingerprints was also successful with an overall cross validation score of 0.8. The best classification ability (validation score = 0.93) was shown for the data set containing only fermenting wines for the most widely represented cultivars (>20 samples). This approach may provide a useful tool for authentication and quality control by regulatory bodies.

Chemical versus infrared spectroscopic measurements of quality attributes of sun or oven dried fruit leathers from apple, plum and apple-plum mixture

This study tested the effect of convective and sun drying on quality parameters of apple, plum and a blend of apple and plum (50:50, w/w) leathers. Moreover, physical, sensory and chemical attributes determined by chemical methods were compared with an infrared spectroscopic method. All products had optimum values of L* (brightness), huge angle and chroma within naturally occurrence range. Convective drying significantly (P < 0.05) caused reduced L* and huge angle and increased chroma, only in plum leather. All products have high general acceptability, while the texture of apple leather was highly preferable. Convective drying lead to significantly decreased dry matter, brix, pH and titratable acidity (TA) and increased aw, total phenolic compound (TPC) and antioxidant capacity (AOC), particularly for plum leather. An overall increase of 10–25% in AOC is found valuable. Peak absorbance values at a band of 1800 to 800 cm−1 in infrared spectra were significantly correlated with TA, pH, TPC and AOC contents. Having compared with chemical methods, infrared spectroscopic method with higher coefficient of determinant, precision and ratio of prediction determinant values as well as lower bias and RMSEP values is recommended for a robust and fast determination of physico-chemical attributes in fruit leather production.

Vibrational Spectroscopy for the Quantification of Hydrotreated Vegetable Oil (HVO) Advanced Biofuels in Petroleum-Derived Fuel Blends: A Minireview

Hydrotreated vegetable oil (HVO), also called hydrotreated esters and fatty acids (HEFA), is an advanced biofuel composed of a mixture of hydrocarbons, similar to petroleum-derived diesel oil and kerosene jet fuel. Due to this chemical similarity, the quantification of this type of biofuel in mixtures with petroleum-derived fuels is a challenging analytical task, whose solution requires carefully developed methods for practical applications. In this review are discussed analytical methods based on vibrational spectroscopy (near infrared and mid infrared absorption as well as Raman scattering) found in the literature, highlighting their advantages in relation to radiocarbon-based reference methods described in ASTM D6866-18, such as the possibility of discrimination and quantification of two different biofuels in mixtures with petroleum-derived fuels. In addition, these approaches are simpler, faster, and less expensive. Vibrational spectroscopy methods provide good accuracy and RMSEP values below 0.8% for quantification of HVO or HEFA in fuel mixtures.

Comparison of PLS and SVM models for soil organic matter and particle size using vis-NIR spectral libraries

In this study a systematic comparison was carried out to assess differences on the accuracy between partial least squares (PLS) and support vector machine (SVM) regression algorithms in soil organic matter and particle size determinations using vis-NIR spectroscopy. The comparison consisted in investigating the influence on the size of calibration set on the external validation set accuracy. For this purpose, three vis-NIR soil libraries containing 14,212, 15,330 and 42,471 soil samples were used to determine sand, clay, and SOM content, respectively. To increase the variability of the results obtained, each calibration subset was randomly generated 49 times and for each iteration a PLS, SVM-Linear and SVM-RBF (radial basis function) regression models were built. These calibration subsets were composed by 250, 1000, 2000, 5000 and 8000 or 10,000 samples.In all situations the SVM-Linear obtained the worst accuracy results. For sand and clay determinations, SVM-RBF models shows a significant improvement on the accuracy, compared to PLS, when the calibration model was built using at least 1000 samples, resulting in a reduction of ~14–29% on the RMSEP. For SOM determinations the difference in RMSEP values of SVM-RBF and PLS starts to be significant when 2000 or more samples were used in calibration set, presenting a reduction of ~8–22% on the RMSEP values. In addition, for all soil attributes investigated between 20 and 27% of the external validation set (1173–2241 samples) were considered outliers and excluded by the PLS regression models.This loss of PLS performance for large calibration sets, indicates the correlation between the vis-NIR spectra and clay, sand and SOM contents tends to be more complex by increasing the variability/number of samples. Requiring the use of machine learnings models with high generalization capacity, such as the SVM-RBF, which increased the performance as the number of samples that compose the calibration set increased.

The Use of Chemometrics for Classification of Sidaguri (&lt;i&gt;Sida rhombifolia&lt;/i&gt;) Based on FTIR Spectra and Antiradical Activities

Sidaguri (Sida rhombifolia) is one of the herbal components used in traditional medicine. The application of chemometrics in the standardization of herbal medicine is common. The objective of this study was to classify Sidaguri from different regions based on FTIR spectra with chemometrics of principal component analysis (PCA) and to correlate the antioxidant activities with FTIR spectra using the multivariate calibration of partial least square regression (PLSR). The extraction of Sidaguri powder was performed using ultrasound-assisted extraction (UAE) at optimum conditions. The obtained extracts were subjected to antiradical scavenging activities using DPPH (2,2’-diphenyl-1-picrylhydrazyl) and ABTS (2,2′-azinobis-3-ethylbenzothiazoline-6-sulfonic acid) radicals. The PCA result shows that Sidaguri from different regions could be separated using 14 wavenumbers of FTIR spectra based on the PCA's loading plot. PLSR regression using the second derivative FTIR spectra at wavenumbers of 3662–659 cm–1 could predict radical scavenging activities (RSA) of Sidaguri with R2 values of 0.9636 and 0.9024 for calibration and validation models, with RMSEC and RMSEP values of 1.45% and 2.65%, respectively. It can be concluded that FTIR spectra treated by PCA were reliable for classifying Sidaguri from different regions. At the same time, PLSR was accurate and precise enough to predict the RSA of Sidaguri.

A cheap handheld NIR spectrometric system for automatic determination of methane, ethane, and propane in natural gas and biogas

In this study, a new NIR detection system using a cheap handheld NIR spectrometer, and a tungsten lamp coupled to a lab-made NIR flow cell was developed for the successful automatic quantification of methane, ethane, and propane in natural gas and biogas samples. To prepare standard gaseous mixtures for full-spectrum PLS multivariate calibration modeling and to record the NIR spectra of these mixtures and samples, this NIR detection system was connected to a flow automatic system and a gas chromatograph. The raw and second-derivative NIR spectra of the calibration set were then employed to construct PLS models using full cross-validation leave-one-out. All models presented high correlation coefficients (>0.9) and root mean square error (<5% mol mol−1) values. However, the RMSEP values using raw and preprocessed datasets were statistically similar according to the F-test at the 95% confidence level. The advantage of using raw data is the elimination of the preprocessing step for future prediction of samples to be analyzed. Once the NIR detection system has been previously calibrated via an automatic or batch process, it can be used to monitor light hydrocarbons or other gases inline (as in pipelines).

A robot system for the autodetection and classification of apple internal quality attributes

Soluble solids content (SSC) is an important index of apple internal quality. To invent a more flexible and efficient method of apple internal quality detection and classification, a robot system for the autodetection and classification of apple internal quality attributes was developed. Visible and near infrared (Vis/NIR) spectroscopy is a promising technology for the nondestructive detection of the internal quality attributes of apples. The end effector of the robot system mainly carried the Vis/NIR spectra collection module and gripping mechanism. The Vis/NIR spectrum was collected when the end effector gripped the apple. Single shot multibox detector (SSD) target detection algorithm was applied to process the images and calculate the position of the apple, which greatly reduced the low accuracy of apple identification caused by light intensity and complex backgrounds, and the speed was approximately 0.055 s per frame. In comparing different modeling results, the normalized spectral ratio (NSR) pretreatment combined with the competitive adaptive reweighted sampling algorithm (CARS) obtained the best modeling result, with Rc and Rcv values of 0.979 and 0.969 and RMSEC and RMSECV values of 0.335 % and 0.385 %, respectively. The classification accuracy of independent validation was 90.0 % with Rp and RMSEP values of 0.952 and 0.393 %. The robot system required approximately 5.200 s to complete a classification for each sample. The results showed feasibility of the robot system to detect the internal quality attributes of apples.

A novel approach for rapid discrimination of common and durum wheat flours using spectroscopic analyses combined with chemometrics

Spectroscopic methods were used in this study for the discrimination of durum and common wheat samples since they are rapid, reliable, easy to use, low cost, environmentally friendly, and non-destructive. For this purpose, 120 common and 119 durum wheat samples with different genotypes were collected from various regions in Turkey and analysed using Raman spectroscopy, near-infrared spectroscopy (NIR), synchronous fluorescence spectroscopy (SFS), and attenuated total reflectance Fourier-transform infrared spectroscopy (ATR-FTIR). Data analysis was performed using the principal component analysis (PCA) and partial least squares-discriminant analysis (PLS-DA).These spectroscopic tools, combined with chemometric analysis, were generally successful in distinguishing common and durum wheat flour samples. It was found that the best method was SFS with a discrimination rate of 100% based on high sensitivity (1.000) and specificity (1.000) values. The effectiveness of the models in which NIR and ATR-FTIR spectroscopies were used was found to be highly similar in terms of the discrimination of durum and common wheat samples. Data obtained from Raman Spectroscopy demonstrated that the method was less sensitive in discriminating between common and durum wheat flour samples than the other spectroscopic techniques with a quite high RMSEP value (0.441). SFS, ATR-FTIR, and NIR spectroscopies proved to be more sensitive and applicable tools than Raman spectroscopy in the discrimination of common and durum wheat samples.