|
|
DisplayTitle
Page Content
Predicting rheological behavior of wheat dough based on machine learning and front-face fluorescence spectroscopy on wheat flour
L. Rhazi (1), J. P. Bonhoure (1), T. Aussenac (1), L. LAKHAL (1). (1) Institut Polytechnique LaSalle Beauvais, Beauvais, France
The milling and baking quality of wheat dough is commonly measured by its rheological properties assed using internationally accepted standard rheological techniques such Farinographs, Mixoraphs, Extensographs and Alveographs. The drawback of these measurement methods is that they are time consuming and costly. Hence, there is a global thrust towards the development of more time and cost efficient methodologies for rapid and accurate determination of wheat flour dough and final products qualities. Front-face fluorescence spectroscopy provides a good alternative as it is rapid, timely, less expensive, non-destructive and straightforward. The aim of this work is to develop a fast and reliable devise for wheat and flour quality control. Rheological quality of wheat dough prepared from 130 cultivars wheat flour samples was assessed with Alveograph indices (W, P, L, P/L and G). Unsupervised fuzzy C-means clustering algorithm is then used to classify alveographic indexes data into four rheological groups based on similarities among the individual data items. Fluorescence excitation and emission spectra of all samples were measured on Horiba Jobin Yvon spectrofluorometer. Using a pattern recognition technique, MOLMAP approach coupled with Bi-Directional Kohonen network, rheological groups were predicted. Despite the small number of available training samples, the estimated correct classification rates were 67 %, 81% and 87 % when the samples were divided into four, three and two rheological groups respectively.
View Presentation |
|
|
|
|
