Fourier Transform Near Infrared Spectroscopy as a Quality Control Tool for the Analysis of Lecithin and By-Products During Soybean Oil Processing

Fourier transform near infrared (FT-NIR) spectroscopy was used to analyze multiple measurement parameters in lecithin production samples and soybean oil refining by-products. For lecithin, partial least squares (PLS) calibration models were developed for acetone insolubles, acid value and moisture and leave-one-out cross validation of the calibration models yielded root mean square error of cross validation (RMSECV) values of 0.37%, 0.59 (mg KOH/g) and 0.050%, respectively. An independent test set consisting of 40% of the lecithin production samples were predicted from the PLS calibration models and a root mean square error of prediction (RMSEP) of 0.41%, 0.53 (mg KOH/g) and 0.056% were obtained for acetone insolubles, acid value and moisture, respectively. Comparison of FT-NIR predictions and corresponding reference method values of 10 lecithin samples using a two-tailed t test showed no significant difference at the p = 0.05 level. A set of 51 samples of soybean oil refining by-products, including acidulated soapstock, fatty acids and black oil, were used for developing PLS calibration models for measuring acid value, moisture and iodine value and leave-one-out cross validations for each model gave values for RMSECV of 6.59 (mg KOH/g), 0.046% and 0.42 (mg I₂/g), respectively. Overall, the results of this study demonstrate the suitability of FT-NIR spectroscopy for the routine analysis of lecithin production samples and soybean oil refining by-products for quality control purposes.     

Determining Frying Oil Degradation by Near Infrared Spectroscopy Using Chemometric Techniques

A rapid method for the determination of some important physicochemical properties in frying oils has been developed. Partial least square regression (PLS) calibration models were applied to the physicochemical parameters and near infrared spectroscopy (NIR) spectral data. PLS regression was used to find the NIR region and the data pre-processing method that give the best prediction of the chemical parameters. Calibration and validation were appropriated by leave one out cross validation and test set validation techniques for predicting free fatty acids (FFA), total polar materials (cTPM; measured by chromatographic method and iTPM measured by an instrumental method), viscosity and smoke point of the frying oil samples. For PLS models using the cross validation techniques, the best correlations (r) between NIR predicted data and the standard method data for iTPM in oils were 93.79 and root mean square error of prediction (RMSEP) values were 5.53. For PLS models using the test set validation techniques, the best correlations (r) between NIR predicted data and standard method data for FFA, cTPM, viscosity and smoke point in oils were 92.58, 94.61, 81.95 and 84.07 and RMSEP values were 0.121, 3.96, 22.30 and 8.74, respectively. In conclusion, NIR technique with chemometric analysis was found very effective in predicting frying oil quality changes.     

Multivariate calibration in Fourier transform infrared spectrometry as a tool to detect adulterations in Brazilian gasoline

In the present work, Fourier transform infrared spectroscopy (FTIR) in association with multivariate chemometrics classification techniques was employed to identify gasoline samples adulterated with diesel oil, kerosene, turpentine spirit or thinner. Results indicated that partial least squares (PLS) models based on infrared spectra were proven suitable as practical analytical methods for predicting adulterant content in gasoline in the volume fraction range from 0% to 50%. The results obtained by PLS provided prediction errors lower than 2% (v/v) for all adulterant determined. Additionally, Soft Independent Modeling of Class Analogy (SIMCA) was performed using all spectral data (650-3700 cm⁻¹) for sample classification into adulterant classes defined by training set and the results indicated that undoubted adulteration detection was possible but identification of the adulterant was subject to misclassification errors, specially for kerosene and turpentine adulterated samples, and must be carefully examined. Quality control and police laboratories for gasoline analysis should employ the proposed methods for rapid screening analysis for qualitative monitoring purposes.     

Comparison Between Linear and Nonlinear Prediction Models for Monitoring of a Paperboard Machine

Data from a paperboard machine were used to compare the performance of linear partial least squares (PLS) and nonlinear feed‐forward neural network (FFNN) modeling of a continuous process. Fifteen selected variables were used as input parameters to the models, while the quality class of the manufactured product was the output response. The models were validated with external data different to those used in the design of the models. Evaluation with root mean square error of prediction (RMSEP) showed that the FFNN models were better for prediction than the PLS models. For monitoring, however, the PLS models detected deviations from normal settings in the paperboard machine more sensitively than the FFNN models. It is suggested that these findings have general relevance to other continuous processes in manufacturing industries too.     

In-line evaluation of average particle size in styrene suspension polymerizations using near-infrared spectroscopy

The main objective of this article is evaluating the influence of average polystyrene particle size upon the near-infrared (NIR) spectra collected during suspension polymerization experiments and observing whether NIR spectroscopy may be used for in-line monitoring and control of average particle size. It is shown that NIR spectra are sensitive to changes of the average particle size, and that standard empirical models (PLS—partial least squares—and NN—neural networks) may be built to correlate average particle size and light absorbance at certain wavelengths fairly well. Finally, it is shown that these models allow the in-line evaluation of average particle size in styrene suspension polymerizations with NIR spectroscopy. © 1998 John Wiley & Sons, Inc. J Appl Polym Sci 70: 1737–1745, 1998     

Development of calibration models for quality control in the production of ethylene/propylene copolymers by FTIR spectroscopy,multivariate statistical tools,and artificial neural networks

Principal component regression (PCR), partial least squares (PLS), StepWise ordinary least squares regression (OLS), and back‐propagation artificial neural network (BP‐ANN) are applied here for the determination of the propylene concentration of a set of 83 production samples of ethylene–propylene copolymers from their infrared spectra. The set of available samples was split into (a) a training set, for models calculation; (b) a test set, for selecting the correct number of latent variables in PCR and PLS and the end point of the training phase of BP‐ANN; (c) a production set, for evaluating the predictive ability of the models. The predictive ability of the models is thus evaluated by genuine predictions. The model obtained by StepWise OLS turned out to be the best one, both in fitting and prediction. The study of the breakdown number of samples to be included in the training set showed that at least 52 experiments are necessary to build a reliable and predictive calibration model. It can be concluded that FTIR spectroscopy and OLS can be properly employed for monitoring the synthesis or the final product of ethylene–propylene copolymers, by predicting the concentration of propylene directly along the process line. © 2008 Wiley Periodicals, Inc. J Appl Polym Sci, 2008     

Simultaneous determination of quality parameters of biodiesel/diesel blends using HATR-FTIR spectra and PLS, iPLS or siPLS regressions

Partial least-squares (PLS), interval partial least squares (iPLS) and synergy partial least squares (siPLS) regressions were used to simultaneous determination of quality parameters of biodiesel/diesel blends. Biodiesel amount, specific gravity, sulfur content and flash point were evaluated using spectroscopic data in the mid-infrared region obtained with a horizontal attenuated total reflectance (HATR) accessory. Eighty-five binary blends were prepared using biodiesel and two types of diesel, in concentrations from 0.2 to 30% (v/v). Fifty-seven samples were used as a calibration set, whereas 28 samples were used as an external validation set. All samples were characterized using the appropriated standard methods. The specific gravity values at 20 °C were in the range of 848.2-866.2 kg/m³. Flash point values lay between 47.0 and 79.5 °C. Sulfur content values varied from 312 to 1351 mg/kg. Raw spectra of the samples were corrected by multiplicative scatter correction (MSC) and were pre-processed using a mean-centered procedure. Algorithms iPLS and siPLS were able to select the most adequate spectral region for each property studied. For all the properties studied, the siPLS algorithm produced better models than the full-spectrum PLS, selecting the most important bands. The quantification of biodiesel was performed using two spectral regions between 650-1909 cm⁻¹ and 2746-3165 cm⁻¹, and an excellent correlation coefficient of R² = 0.9996 was obtained. The specific gravity was determined from the spectral region from 650 to 1070 cm⁻¹, which yielded a very good correlation coefficient of R² = 0.9987. The sulfur content was evaluated from the spectral regions of 1070-1491 cm⁻¹ and 2746-3165 cm⁻¹. A very good correlation coefficient of R² = 0.9995 was obtained, regardless of whether the samples were formulated with metropolitan or countryside diesel. Finally, the flash point was determined from the spectral region between 756 and 968 cm⁻¹ and a very good correlation coefficient of R² = 0.9982 was obtained.     

Novel,Rapid Identification,and Quantification of Adulterants in Extra Virgin Olive Oil Using Near‐Infrared Spectroscopy and Chemometrics

A new, rapid Fourier transform near infrared (FT‐NIR) spectroscopic procedure is described to screen for the authenticity of extra virgin olive oils (EVOO) and to determine the kind and amount of an adulterant in EVOO. To screen EVOO, a partial least squares (PLS1) calibration model was developed to estimate a newly created FT‐NIR index based mainly on the relative intensities of two unique carbonyl overtone absorptions in the FT‐NIR spectra of EVOO and other mixtures attributed to volatile (5280 cm^?1) and non‐volatile (5180 cm^?1) components. Spectra were also used to predict the fatty acid (FA) composition of EVOO or samples spiked with an adulterant using previously developed PLS1 calibration models. Some adulterated mixtures could be identified provided the FA profile was sufficiently different from those of EVOO. To identify the type and determine the quantity of an adulterant, gravimetric mixtures were prepared by spiking EVOO with different concentrations of each adulterant. Based on FT‐NIR spectra, four PLS1 calibration models were developed for four specific groups of adulterants, each with a characteristic FA composition. Using these different PLS1 calibration models for prediction, plots of predicted vs. gravimetric concentrations of an adulterant in EVOO yielded linear regression functions with four unique sets of slopes, one for each group of adulterants. Four corresponding slope rules were defined that allowed for the determination of the nature and concentration of an adulterant in EVOO products by applying these four calibration models. The standard addition technique was used for confirmation.     

Rapid Determination of Total Polar Compounds in Frying Oil Using ATR‐FTIR Combined with Extended Partial Least Squares Regression

A method of rapidly determining the total polar compounds (TPCs) in frying oils using attenuated total reflectance‐Fourier transform infrared spectroscopy combined with partial least squares (PLS) regression is developed. Oils of various types and geographic origins are used to ensure that the proposed model is robust. The first derivative spectrum is selected as the spectral processing method. The interval PLS, forward interval PLS, and backward interval PLS algorithms are compared in terms of their performance. A correlation coefficient (R²) of 0.9942, a root mean square error of calibration (RMSEC) of 1.1, a root mean square error of prediction (RMSEP) of 2.30, a residual predictive deviation (RPD) of 4.1, and a limit of detection (LOD) of 1.65% are obtained by the fiPLS33 model with fewer latent variables and a lower spectral interval number. In addition, sub‐models using a single type of oil showed higher performance (R² 0.9957–0.9998, RMSEC 0.12–0.92, RMSEP 0.79–1.58, RPD 4.79–9.64, LOD 0.66–1.26%) than the general model. The TPC models developed are accurate, stable, and adaptable, and they can be used to analyze general frying oil samples quickly, regardless of the oil type, and to analyze samples of specific oil types accurately. Practical applications: The content of TPCs is an important indicator of whether the oil has been overused and whether it will be harmful during the frying process. However, traditional chemical methods are time‐consuming, and they have not been used to determine large‐sized samples. In addition, due to a lack of regional optimization, most studies on determining TPCs with FTIR give unsatisfactory model performance. A general TPC model that incorporates several oil types and regional optimization is expected to improve prediction performance. Therefore, the proposed method represents a rapid and accurate tool for measuring TPCs in edible fats and oils.     

Discrimination of various poly(propylene) copolymers and prediction of their ethylene content by near‐infrared and Raman spectroscopy in combination with chemometric methods

Near‐infrared (NIR) diffuse reflectance (DR) spectra and Fourier‐transform (FT) Raman spectra were measured for 12 kinds of block and random poly(propylene) (PP) copolymers with different ethylene content in pellets and powder states to propose calibration models that predict the ethylene content in PP and to deepen the understanding of the NIR and Raman spectra of PP. Band assignments were proposed based calculation of the second derivatives of the original spectra, analysis of loadings and regression coefficient plots of principal component analysis (PCA) and principal component regression (PCR) (predicting the ethylene content) models, and comparison of the NIR and Raman spectra of PP with those of linear low‐density polyethylene (LLDPE) with short branches. PCR and partial least squares (PLS) regression were applied to the second derivatives of the NIR spectra and the NIR spectra after multiplicative scatter correction (MSC) to develop the calibration models. After MSC treatment, the original spectra yield slightly better results for the standard error of prediction (SEP) than the second derivatives. A plot of regression coefficients for the PCR model shows peaks due to the CH₂ groups pointing upwards and those arising from the CH₃ groups pointing downwards, clearly separating the bands due to CH₃ and CH₂ groups. For the Raman data, MSC and normalization were applied to the original spectra, and then PCR and PLS regression were carried out to build the models. The PLS regression for the normalized spectra yields the best results for the correlation coefficient and the SEP. Raman bands at 1438, 1296, and 1164 cm^?1 play key roles in the prediction of the ethylene content in PP. The NIR chemometric evaluation of the data gave better results than those derived from the Raman spectra and chemometric analysis. Possible reasons for this observation are discussed. © 2002 Wiley Periodicals, Inc. J Appl Polym Sci 87: 616–625, 2003     

Characterization of petroleum using near-infrared spectroscopy: Quantitative modeling for the true boiling point curve and specific gravity

This work describes a new approach to predict the true boiling point (TBP) curve and to estimate the API gravity in order to characterize the petroleum processed in refineries by using the information present in its absorbance spectrum obtained in the near-infrared region (NIR). The absorbance spectra were obtained in the range from 3700 to 10000 cm⁻¹ employing a CaF₂ transmittance cell with a 0.5 mm light path. Three spectral regions were evaluated for modeling purpose: 5000–3900 cm⁻¹, 6000–3700 cm⁻¹, and 9000–700 cm⁻¹. The spectral region corresponding to the combination of C–H vibrations produces absorption spectra with very good quality while the region above 6500 cm⁻¹ is dominated by scattering of the radiation. The absorbance spectra of a total of 122 samples of petroleum and petroleum blends coming from various producing regions in Brazil and abroad were obtained and pre-processed to correct for base line shift and for the integrated area. Two approaches were employed to obtain the models: one using artificial neural networks (ANN) and the other using partial least squares (PLS). The results showed that PLS gives better predictions than ANN for the API gravity and the TBP curve. The best results were obtained using the 5000–3900 cm⁻¹ spectral range. In an external validation, the average RMSEP for the volume of distillate along the TBP curve employing PLS model was 1.13V% while that for API gravity was 0.24. A comparison between the results obtained by a simulator used by the refinery and the PLS model revealed a better performance for the model based on NIR spectrometry.     

Transfer of a nanoparticle product between different mixers using latent variable model inversion

An experimental nanoparticle preparation process by solvent displacement in passive mixers is considered. The problem under investigation is to estimate the operating conditions in a target device (Mixer B) in order to obtain a product of assigned properties that has already been manufactured in a source device of different geometry (Mixer A). A large historical database is available for Mixer A, whereas a limited historical database is available for Mixer B. The difference in device geometries causes a different mixing performance within the devices, which is very difficult to capture using mechanistic models. The problem is further complicated by the fact that Mixer B can only be run under an experimental setup that is different from the one under which the available historical dataset was obtained. A joint‐Y projection to latent structures (JY‐PLS) model inversion approach is used to transfer the nanoparticle product from Mixer A to Mixer B. The Mixer B operating conditions estimated by the model are tested experimentally and confirm the model predictions within the experimental uncertainty. Since the inversion of the JY‐PLS model generates an infinite number of solutions that all lie in the so‐called null space, experiments are carried out to provide (to the authors' knowledge) the first experimental validation of the theoretical concept of null space. Finally, by interpreting the JY‐PLS model parameters from first principles, the understanding of the system physics is improved. © 2013 American Institute of Chemical Engineers AIChE J, 60: 123–135, 2014     

Comparative study of estimation methods of NO<Subscript><Emphasis Type="Italic">x</Emphasis></Subscript> emission with selection of input parameters for a coal-fired boiler

This study focuses on estimation of NO_x emission and selection of input parameters for a coal-fired boiler in a 500 MW power generation plant. Careful selection of input parameters is required not only to improve accuracy of the estimation, but also to reduce the model dimensionality. The initial operating input parameters are determined based on operation heuristics and accumulated operation knowledge; the essential input parameters are selected by sensitivity analysis where the performance of the estimation model is assessed as one or some input parameters are successively eliminated from the computation while all other input parameters are retained. From the sequential input selection process, less than ten input parameters survived out of 36 initial input parameters. Auto-regressive moving average (ARMA) model, artificial neural networks (ANN), partial least-squares (PLS) model, and least-squares support vector machine (LSSVM) algorithm were proposed to express the relationship between the operating input parameters and the content of NO_x emission. Historical real-time data obtained from a 500 MW power plant coal-fired boiler were used to test the proposed models. It was found that principal components analysis (PCA) enhances the estimation performance of each model. Among the four proposed estimation models, the LSSVM model coupled with PCA scheme showed the minimum root-mean square error (RMSE) and the best R-square value.     

Determination of octane numbers for clean gasoline using dielectric spectroscopy

In the present work, dielectric spectroscopy (DES) in association with partial least squares (PLS) multivariate calibration method was employed to determine octane numbers (research octane number or RON and motor octane number or MON) of clean gasoline samples. The factor number included in PLS model was obtained according to the lowest sum of squares of predicted residual error (PRESS) in calibration set. The performance of the final model was evaluated according to PRESS and correlation coefficient (R). The optimal factor numbers are 9 in both RON and MON PLS calibration models, which were achieved with PRESS = 2.74 and R = 0.9598 in RON calibration set and the lowest PRESS = 2.72 and R = 0.8983 in MON calibration set. In validation set, PRESS = 1.00 and R = 0.9552 for RON and PRESS = 0.47 and R = 0.9105 for MON were obtained. Results indicated that PLS multivariate calibration models based on DES data were proven suitable as a practical analytical method for predicting octane numbers of clean gasoline.     

Monitoring PV in corn and soybean oils by NIR spectroscopy

NIR spectroscopy was used successfully in our laboratory to monitor oxidation levels in vegetable oils. Calibration models were developed to measure PV in both soy and corn oils, using partial least squares (PLS) regression and forward stepwise multiple linear regression, from NIR transmission spectra. PV can be measured successfully in both corn and soy oils using a single calibration. The most successful calibration was based on PLS regression of first derivative spectra. When this calibration was applied to validation sample sets containing equal numbers of corn and soy oil samples, with PV ranging from 0 to 20 meq/kg, a correlation coefficient of 0.99 between titration and NIR values was obtained, with a standard error of prediction equal to 0.72 meq/kg. For both types of oil, changes occurred in the 2068 nm region of the NIR spectra as oxidation levels increased. These changes appear to be associated with the formation of hydroperoxides during oxidation of the oils.     

Multivariate calibration of fourier transform infrared spectra for determining thiobarbituric acid-reactive substance content in palm oil

Fourier transform infrared (FTIR) spectra of palm oil samples between 2900 and 2800 cm⁻¹ and 1800 and 1600 cm⁻¹ were used to compare different multivariate calibration techniques for quantitative determination of their thiobarbituric acid-reactive substance (TBARS) content. Fifty spectra (in duplicate) of palm oil with TBARS values between 0 and 0.25 were used to calibrate models based on partial least squares (PLS) and principal components regression (PCR) analyses with different baselines. The methods were compared for the number of factors, coefficients of determination (R ²), and accuracy of estimation. The standard errors of prediction (SEP) were calculated to compare their predictive ability. The calibrated models generated three to eight factors, R ² of 0.9414 to 0.9803, standard error of estimation (SEE) of 0.0063 to 0.0680, and SEP of 1.20 to 6.67.     

Refractive index and lattice parameter prediction for garnets (A3B2C3X12)

The validity of prediction methods for garnet lattice parameters was tested, and a modified model with best fit to a lattice parameter database with over 1000 garnet compositions was developed. The lattice parameter predictions were used in Shannon and Fischer's model to predict garnet refractive indices. The predictions were compared to refractive index measurements reported for 143 garnet compositions. After calibration of electron overlap factors used in the model, the average prediction error was 1.13%. Sources of error in the models are discussed, as well as applications of the predictive methods.     

Simultaneous determination of molecular weight and crystallinity of recycled HDPE by infrared spectroscopy and multivariate calibration

An attempt of correlating molecular weight (M_n) of recycled high‐density polyethylene (HDPE) as measured by size‐exclusion chromatography (SEC) with diffuse reflectance near and mid‐infrared spectroscopy (NIR/MIR) was made by means of multivariate calibration. The spectral data obtained was also used to extract information about the degree of crystallinity of the recycled resin. Differential scanning calorimetry (DSC) was used as the reference method. Partial least‐squares (PLS) calibration was performed on the MIR and NIR spectral data for prediction of M_n. Four PC factors described fully the PLS models. The root‐mean‐square error of prediction (RMSEP) obtained with MIR data was 360, whereas a RMSEP of 470 was achieved when calibration was carried out on the diffuse reflectance NIR data. A PLS calibration for prediction of degree of crystallinity was performed on the NIR data in the 1100–1900‐nm region, but the ability of prediction of this model was poor. However a PLS calibration in the region 2000–2500 nm yield better results. Four PC factors explained the most of the variance in the spectra and the RMSEP was 0.4 wt %. © 2002 Wiley Periodicals, Inc. J Appl Polym Sci 85: 321–327, 2002     

Multiloop PID controller design using partial least squares decoupling structure

The goal of this paper is to identify and control multi-input multi-output (MIMO) processes by means of the dynamic partial least squares (PLS) model, which consists of a memoryless PLS model connected in series with linear dynamic models. Unlike the traditional decoupling MIMO process, the dynamic PLS model can decompose the MIMO process into a multiloop control system in a reduced subspace. Without the decoupler design, the optimal tuning multiloop PID controller based on the concept of general minimum variance and the constrained criteria can be directly and separately applied to each control loop under the proposed PLS modeling structure. Several potential applications using this technique are demonstrated.