Visible and near infrared (Vis/NIR) spectroscopy combined with least squares-support vector machine (LS-SVM) was investigated for the determination of polysaccharides of Auricularia auricula. A total of 240 samples were prepared from four different geographical origins. The calibration set was consisted of 180 samples (45 samples for each origin) and the remaining 60 samples for the validation set. Different preprocessing methods were compared in partial least squares (PLS) models including smoothing, multiplicative scatter correction (MSC), standard normal variate (SNV), the first and second derivative. PLS analysis was employed for the calibration models as well as extraction of certain latent variables (LVs). Simultaneously, some effective wavelengths (EWs) extracted by regression coefficients of LS-SVM were used as the inputs of LS-SVM compared with LVs. The optimal prediction results were achieved by LV-LS-SVM, and the correlation coefficient (r), root mean square error of prediction (RMSEP) and bias for validation set were 0.9413, 0.6893 and -0.0729, respectively. The results were slightly better than PLS. The prediction results of EW-LS-SVM using 3 EWs were acceptable with r=0.9290, RMSEP=0.7714 and Bias=-0.1737. The results indicated that Vis/NIR spectroscopy could be utilized as an efficient way for the determination of polysaccharides of auricularia auricula based on LS-SVM method.
The determination of citric acid of lemon vinegar was processed using visible and near infrared (Vis/NIR) spectroscopy
combined with least squares-support vector machine (LS-SVM). Five concentration levels (100%, 80%, 60%, 40% and
20%) of lemon vinegar were studied. The calibration set was consisted of 225 samples (45 samples for each level) and
the remaining 75 samples for the validation set. Partial least squares (PLS) analysis was employed for the calibration
models as well as extraction of certain latent variables (LVs) and effective wavelengths (EWs). Different preprocessing
methods were compared in PLS models including smoothing, standard normal variate (SNV), the first and second
derivative. The selected LVs and EWs were employed as the inputs to develop least square-support vector machine (LSSVM)
models. The optimal prediction results were achieved by LV-LS-SVM model, and the correlation coefficient (r),
root mean square error of prediction (RMSEP) and bias for validation set were 0.9990, 0.1972 and -0.0334, respectively.
Moreover, the EW-LS-SVM model was also acceptable and slightly better than all PLS models. The results indicated
that Vis/NIR spectroscopy could be utilized as a parsimonious and efficient way for the determination of citric acid of
lemon vinegar based on LS-SVM method.
Visible and near infrared (Vis/NIR) transmission spectroscopy and chemometric methods were utilized to predict the pH
values of cola beverages. Five varieties of cola were prepared and 225 samples (45 samples for each variety) were
selected for the calibration set, while 75 samples (15 samples for each variety) for the validation set. The smoothing way
of Savitzky-Golay and standard normal variate (SNV) followed by first-derivative were used as the pre-processing
methods. Partial least squares (PLS) analysis was employed to extract the principal components (PCs) which were used
as the inputs of least squares-support vector machine (LS-SVM) model according to their accumulative reliabilities.
Then LS-SVM with radial basis function (RBF) kernel function and a two-step grid search technique were applied to
build the regression model with a comparison of PLS regression. The correlation coefficient (r), root mean square error
of prediction (RMSEP) and bias were 0.961, 0.040 and 0.012 for PLS, while 0.975, 0.031 and 4.697x10-3 for LS-SVM,
respectively. Both methods obtained a satisfying precision. The results indicated that Vis/NIR spectroscopy combined
with chemometric methods could be applied as an alternative way for the prediction of pH of cola beverages.
The feasibility of visible and near infrared (Vis/NIR) spectroscopy, in combination with a hybrid multivariate methods of
partial least squares (PLS) analysis and BP neural network (BPNN), was investigated to identify the variety of rice
vinegars with different internal qualities. Five varieties of rice vinegars were prepared and 225 samples (45 for each
variety) were selected randomly for the calibration set, while 75 samples (15 for each variety) for the validation set. After
some pretreatments with moving average and standard normal variate (SNV), partial least squares (PLS) analysis was
implemented for the extraction of principal components (PCs), which would be used as the inputs of BP neural network
(BPNN) according to their accumulative reliabilities. Finally, a PLS-BPNN model with sigmoid transfer function was
achieved. The performance was validated by the 75 unknown samples in validation set. The threshold error of prediction
was set as ±0.1 and an excellent precision and recognition ratio of 100% was achieved. Simultaneously, certain effective
wavelengths for the identification of varieties were proposed by x-loading weights and regression coefficients. The
prediction results indicated that Vis/NIR spectroscopy could be used as a rapid and high precision method for the
identification of different varieties of rice vinegars.
Three different chemometric methods were performed for the determination of sugar content of cola soft drinks using
visible and near infrared spectroscopy (Vis/NIRS). Four varieties of colas were prepared and 180 samples (45 samples for each variety) were selected for the calibration set, while 60 samples (15 samples for each variety) for the validation set. The smoothing way of Savitzky-Golay, standard normal variate (SNV) and Savitzky-Golay first derivative transformation were applied for the pre-processing of spectral data. The first eleven principal components (PCs) extracted by partial least squares (PLS) analysis were employed as the inputs of BP neural network (BPNN) and least squares-support vector machine (LS-SVM) model. Then the BPNN model with the optimal structural parameters and LS-SVM model with radial basis function (RBF) kernel were applied to build the regression model with a comparison of PLS regression. The correlation coefficient (r), root mean square error of prediction (RMSEP) and bias for prediction were 0.971, 1.259 and -0.335 for PLS, 0.986, 0.763, and -0.042 for BPNN, while 0.978, 0.995 and -0.227 for LS-SVM, respectively. All the three methods supplied a high and satisfying precision. The results indicated that Vis/NIR spectroscopy combined with chemometric methods could be utilized as a high precision way for the determination of sugar content of cola soft drinks.
For rapid detection of soluble solid content (SSC) in beer, visible/near infrared (Vis/NIR) spectra of 360 beer samples were collected by using Vis/NIR spectroradiometer. Principal component analysis (PCA) was applied for reducing the dimensionality in order to decrease the overlapped information of raw spectral data, 6 principal components (PCs) were selected. The samples were randomly separated into calibration set and validation set, and least squares support vector machine (LS-SVM) algorithm was used to build calibration model of SSC in beer, then the model was employed for the prediction of the validation set. Correlation coefficient (r) of prediction and root mean square error prediction (RMSEP) were used as evaluation standards, and the results indicated that r and RMSEP for the prediction of SSC were 0.9829 and 0.1506. The precision of prediction was obviously higher than that of back-propagation artificial neural network (BP-ANN) and partial least squares (PLS) models, hence PCA and LS-SVM algorithm model with high prediction precision could be applied to the determination of SSC in beer.
KEYWORDS: Spectroscopy, Calibration, Principal component analysis, Near infrared spectroscopy, Chemical analysis, Autoregressive models, Near infrared, Statistical modeling, Absorbance, Visible radiation
Visible and near infrared (Vis/NIR) transmission spectroscopy and a hybrid chemometrics method were applied to
determine the pH of rice wines. A spectroradiometer with a wavelength region of 325-1075 nm was used for spectral
scanning. The calibration set was composed of 240 samples and 60 samples were used in the validation set. The
smoothing way of Savitzky-Golay and standard normal variate (SNV) were used as data pre-processing methods.
Principal components analysis (PCA) was employed to extract the principal components (PCs) which were used as the
inputs of Least squares-support vector machine (LS-SVM) model. Then LS-SVM with radial basis function (RBF)
kernel function was applied to build the regression model with a comparison of partial least squares (PLS) regression.
The correlation coefficient (r), root mean square error of prediction (RMSEP) and bias of LS-SVM were 0.964, 2.62x10-4
and 8.83x10-4, respectively. Significant wavelengths for pH were proposed according to x-loading weights. The results
indicated that Vis/NIR spectroscopy with the combination of LS-SVM could be utilized as an alternative way for the
determination pH of rice wines.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.