Browsing by Subject "Chemometrie"

Now showing 1 - 3 of 3
  • Thumbnail Image
    Publication
    Fluorescence spectroscopy and chemometrics : an innovative approach for characterization of wheat flour and dough preparation
    (2016) Ahmad, Muhammad Haseeb; Hitzmann, Bernd
    Implementation of process analytical technologies (PAT) in food applications has attained a remarkable motivation due to higher quality and safety standards in this field. PAT applications also include rapid and non-invasive approaches which can be obtained from spectroscopic techniques. Fluorescence spectroscopy together with chemometrics is considered to be an outstanding analytical tool for fast and non-invasive technique for food analysis which can be used in various food applications on industrial scale. It is known for its sensitivity and specificity which can analyze the different foods and its ingredients while chemometrics helps to extract the useful information from the spectral data. The different chemometrics tools used for quantitative and qualitative analysis of spectral data, has increased the importance of this spectroscopic technique in generating the new ideas and hypothesis to develop new analytical methods which lead towards betterment in industrial operations for process and quality monitoring. In this doctoral project, fluorescence spectroscopic together with chemometric has been utilized to develop some new methods for determination of different parameters of wheat which provides the central idea of the thesis. First manuscript presents the potential of fluorescence spectroscopy to predict the analytical, rheological and baking parameters of different wheat flours by just taking the spectral signature without any sample preparation. Twelve different wheat flours milled from wheat cultivars were used to analyze the analytical, rheological and baking parameters using the conventional methods. These measured parameters were predicted from the spectral data taken for different wheat flours using genetic algorithm coupled with partial least square regression. The model obtained for protein, wet gluten and sedimentation value showing high R2 = 0.90, 0.92 and 0.81 respectively. Similarly, the rheological parameters like dough development time and water absorption were also predicted with low root mean square error of cross validation (RMSECV) and high R2 = 0.95 and 0.77 respectively while pasting temperature showed R2 = 0.78. Furthermore, moisture and volume of bread were predicted with high accuracy showing R2 = 0.86 and 0.95 respectively in the baking parameters. Other rheological and baking parameters like dough stability, softening, farinograph quality number, baking loss, crumb hardness and springiness were not predicted well due to poor correlation and high error. In the second paper, characterization of complex farinographic kneading process is performed by using the fluorescence spectroscopy in combination with chemometric tools. The aim of this investigation is to determine the impact of hydration of flour onto the spectral signals, classification of farinographic curve and separation of wheat flours based on their bread making performance. Secondly the middle curve of farinograph was predicted out of the fluorescence spectra using partial least square regression (PLSR) which can help to predict optimal dough development time. The spectra of the flour showed high intensities in protein, NADH and riboflavin regions which reduce to 36 %, 58 % and 61 % respectively after the hydration process depicting its influence due to structural changes in protein and oxidation of NADH. The farinographic curve was divided into four phases and principal component analysis (PCA) has been used to extract the qualitative information regarding the farinographic curve from the fluorescence spectra to categorize all farinographic phases into hydration, dough development, and stability and softening. Similarly, different pre-processing tools like standard normal variate and generalized least square weighting generate good separation of various wheat flours during the farinographic kneading process into different quality groups (E, A, B and C) on the basis of their bread baking performance from the spectral data using PCA. Additionally, PLSR was applied to predict the middle curve of farinograph out of spectral data showing a descent coefficient of determination R2 = 0.75 with RMSECV of 14 Brabender units. However, more research can lead towards the development of a sensor for determination of optimal dough development time. In another study, the nutritional parameters of 26 different types of wheat flour obtained from different vendors from the supermarket were predicted using fluorescence coupled with linear and non-linear chemometric tools. PCA applied on the spectral data for different types of the wheat flours showing a clear separation. On the other hand, PLSR was used to quantify the nutritional parameters of different types of wheat flours showing a good prediction for fat, moisture and carbohydrates using cross-validation, with a R2 of 0.88, 0.86 and 0.89, respectively whereas the protein, sucrose and salt contents presented a little correlation in PLSR. Therefore, locally weighted regression, a non-linear chemometric tool improves the prediction ability of all of the nutritional parameters by decreasing the error with an increasing R2. The energetic value, protein, fat, carbohydrate, moisture, sucrose, salt and saturated fatty acid contents showed R2 of 0.96, 0.93, 0.99, 0.99, 0.98, 0.88, 0.95, and 0.99 respectively, for different wheat flours. The aforementioned results clearly demonstrate the potential of the fluorescence spectroscopy in determination of analytical, rheological, baking and nutritional parameters of the wheat flours. They present that it can be used to characterize and categorize the farinographic kneading process, which is important in the bread-baking industry. More research in this direction can result in developing a sensor for predicting the quality parameters and processing operations in the cereal based industries rapidly and non-invasively which are important for regulatory and screening of the wheat on quality characteristics for marketing and end product evaluations.
  • Thumbnail Image
    Publication
    Self-learning modules for spectra evaluation
    (2022) Sadeghi Vasafi, Pegah; Hitzmann, Bernd
    Monitoring milk processing is an essential task as it affects the quality and safety of the final product. The aim of this investigation was to develop and analyse the self-learning system for the supervision of the processing of milk. In the self-learning evaluation module, several algorithms for data analysis of near infrared (NIR) and Raman spectra was implemented for the prediction of sample quality and safety. In the first part of this thesis, the use of NIR spectroscopy for controlling milk processing was investigated. For this reason, a high-quality quartz flow cell with a 1 mm pathlength including temperature controlling option for liquids was implemented. For sample preparation, UHT-milk with 1.5 % fat content was measured at 5 °C and considered as the reference milk. Samples with various changes such as added water and cleaning solution, different fat content and temperature as well as milks from various suppliers were investigated as the modified samples. A data set from reference and modified samples was obtained with NIR measurements. In this study, first Savitzky-Golay derivative with second polynomial order and window size of 15 was applied. It was compared with the usefulness of raw spectrum and also the combination of raw and first derivative spectrum. For the self-learning sector, an autoencoder neural network was employed. Within this thesis, it was shown that the trained autoencoder using first derivative spectra was capable to detect 5 % added water and 9 % cleaning solution in the milk. However, by using the combination spectra, the difference of 0.1 % in fat concentration was perfectly recognized. These two procedures were able to detect milks from different suppliers and difference of 10 °C in the measurement temperature. Another part of this work was done using Raman spectroscopy. The aim of this part was to check if the previous result can be improved. In this step, the circulation method was again employed the same as in the previous part. However, because of the heat introduced to the sample by the laser using in Raman spectroscopy and the length of plastic tubes which can be affected by the temperature of the laboratory, the measurement temperature was kept at 10 °C. 1.5 % fat UHT milk was utilized as the reference sample. Milks with various changes such as different fat contents, various measurement temperatures and added water or cleaning solution were investigated as the modified samples. In this investigation, not only the autoencoder but also some chemometric models were utilized with the purpose of anomaly detection. Principal component analysis (PCA) was investigated to check if the various samples can becategorized separately. In addition, two chemometric modelling techniques such as principal component regression and Gaussian process regression were tested to check the ability for change detection. By using the results obtained by PCA, a sufficient categorization of various samples was not achieved. While the PCR did not present a promising prediction as the related R2 was 0.7, Gaussian process regression with R2 of 0.97 predicted the changes almost perfectly. The trained autoencoder and Gaussian process regression both were able to define 5 % water and cleaning solution, difference of 0.1 % fat content, and variation of 5 °C in the measurement temperature. In comparison between the autoencoder and Gaussian process regression, it should be mentioned that the Gaussian process regression was capable to determine more abnormal signals than the autoencoder, however, it must be trained with all the possible changes. In contrast, the autoencoder can be trained once just with reference signals and used in online monitoring properly. As the final part and to detect which type of anomalies happened during the milk processing, several classification approaches such as linear discriminant analysis, decision tree, support vector machine, and k nearest neighbour were utilized. While decision trees and linear discriminant analysis failed to effectively characterize the various types of anomalies, the k nearest neighbor and support vector machine presented promising results. The support vector machine presented an accuracy of 81.4 % for test set, while the k nearest neighbor showed an accuracy of 84.8 %. As a result, it is reasonable to assume that both algorithms are capable of classifying various groups of data accurately. It can help the milk business figure out whats going wrong during the processing of milk. In general, Raman spectroscopy produced better findings than NIR spectroscopy, because the typical absorption bands of milk components in NIR spectrometers may be impacted by high water absorption combined with substantial light scattering by fat globules. Additionally, the autoencoder as self-learning system was capable of correctly detecting changes during milk processing, however, classification algorithms can aid in obtaining more details.
  • Thumbnail Image
    Publication
    Ultraschallbasierte simultane Konzentrationsbestimmung der Komponenten Zucker und Ethanol in wässrigen Fermentationsfluiden
    (2014) Schöck, Thomas; Hitzmann, Bernd
    At alcoholic fermentation processes in aqueous solutions there are converted various sugars (mono-, di- and polysaccharides) into ethanol and carbon dioxide by diverse intermediate steps. In the industrial production, ultrasound based methods for the analysis of the composition of the fermentation fluid are advantageous due to their robustness, price cheapness and the possibility for the accomplishment of on line measurements. Within the scope of the present work there are presented several methods for the simultaneous determination of the sugar and ethanol content in the fermentation fluid based on the analysis of ultrasound parameters, also at the presence of dissolved carbon dioxide gas, and compared with respect to the accuracy of their predictive values. Initially there is investigated the behavior of the parameters sound velocity and adiabatic compressibility in standardized aqueous fluids in dependency of the concentration of the components sugar (2 -16 mass percent) and ethanol (1- 6 mass percent), the CO2 partial pressure (0 – 3,013E+05 Pa) and the temperature (2 – 30° C). Thereby the disaccharide saccharose acts as a model substance for the sugar fraction. From the data field of the sound velocity two polynomial calibration models for the sugar / ethanol concentration are extracted with the methods of the multiple linear regression (MLR) and the partial least squares (PLS-) analysis. The minimal accessible standard deviation of the concentration values determined by the particular model from the reference values lies for the MLR method at 0,6 mass percent for the sugar and 0,2 mass percent for the ethanol fraction. The PLS-analysis yields a standard deviation for the sugar and ethanol values of 0,36 and 0,13 mass percent respectively (fluids without a CO2 fraction), as well as 0,5 / 0,17 mass percent (fluids including a CO2 fraction). A further analytic method uses a linearized model of the adiabatic compressibility and the density for the sugar / ethanol determination. The analysis of two physical parameters at this method yields a significant increase of the model quality. For fluids without a CO2 fraction there is reached a minimal standard deviation of 0,06 mass percent for the sugar and 0,07 mass percent for the ethanol concentration. For CO2 containing fluids the corresponding values results to 0,06 / 0,13 mass percent.