Prediction of quality parameters of flour by using near and mid infrared spectroscopy

Grain has a big importance in human diet and wheat is the first row among grains. Wheat, which is used in the industry, quality can show significant variance depending on climate, genetic difference, environment, and storage conditions. Flour, which is obtained from wheat, can be used to produce bread, pasta, and biscuit. The obtained these end-products quality depends on flour quality. Quality in flour can be stated as that measurable parameters determine the utility of flour for a certain aim. These parameters can be carried out by physical (hectoliter, hardness etc.) chemical (moisture, ash, protein etc.) physicochemical (sedimentation, gluten index, gluten etc.) rheological (farinograph, extensograph etc.) and technological (bread, pasta, and biscuits) analysis. However experienced staff is needed to carry out these analysis. Also these analysis can take long time and cause usage of hazardous chemicals. Spectroscopic methods such as Near Infrared and Middle Infrared don't cause environmental pollution because not to use any chemicals. Relationship between spectra and reference analysis is researched by using chemometric and mathematical methods. This thesis is aimed at built calibration models which can be predict the quality flour parameters without reference analysis and with short-time spectroscopic methods. The most accurate calibration models was obtained for protein, moisture and gluten parameters. R2 value of protein (%) parameter for bread wheat, durum wheat, biscuit wheat and watery region bread wheat was 0.963, 0.979, 0.956 and 0.983, respectively. R2 value of moisture (%) parameter for bread wheat, durum wheat, and watery region bread wheat was 0.942, 0.941 and 0.955, respectively. R2 value of gluten (%) parameter for bread wheat, durum wheat, biscuit wheat and watery region bread wheat was 0.917, 0.934, 0.954 and 0.962, respectively. Key words: flour, wheat, chemometric, quality parameters, NIR, MIR, ANN.