Extraction of rheological-optical characteristics of rice single kernel, in order to develop an instrumental method for determining grain quality

Fazeli Burestan, Nasrollah (2020) Extraction of rheological-optical characteristics of rice single kernel, in order to develop an instrumental method for determining grain quality. Doctoral thesis, University of Mohaghegh Ardabili.

Text رساله اصلی.pdf Download (727kB)

Abstract

Research Aim: To determine the chemical properties (amylose (AC), gelatinization temperature (GT), gel consistency (GC) and protein content (PC)) and physicochemical (minimum (MinV) and peak viscosity (PV), final viscosity (FV), breakdown (BDV) and viscosity setback (SBV), time required to achieve maximum viscosity (PT) and viscosity temperature (Pa.T)) rice with the use of quality analysis is expensive. The aim of this study is the predict possibility the quality properties of rice using NIR spectroscopy, machine vision and mechanical tests methods on bulk and single kernel samples. Research method: In this study, 5 rice varieties namely Demsiah, Hashemi, Khazar, Darfak and Kadous were used via three methods of mechanical, image processing on single kernel and NIRS on single kernel and bulk samples with the aim of predicting quality properties. The data the extracted from the mechanical test was applied at two moisture levels (9 and 14% b.w) during the factorial experiment in the form of a completely randomized design using SPSS software. In the image processing test, the samples were carry out at a moisture level of 9% on a more uniform basis in three treatments: exposure from above with LED lamp (A), exposure from above with LED and fluorescent lamp (B) and exposure from below (C). Extracted data were analyzed by partial least squares regression (PLSR), multivariate regression (MLR) and response surface methodology (RSM) modeling. Also, the possibility of predicting some quality properties was investigated via spectral analysis by extracting the components of NIRS in two modes of bulk and single kernel samples. Findings: The results of analysis of variance (ANOVA) of the components extracted from compression, bending and creep tests showed that moisture and variety have a significant effect (P <0.01) On energy to failure point, maximum secanti modulus, secanti modulus at 0.05 mm deformation, maximum tangential modulus, tangential modulus at 0.05 mm deformation, maximum deformation, maximum failure force, force at 0.05 mm deformation, tangential modulus in 7 seconds, tangential modulus in 30 seconds, secanti modulus in 7 seconds, secanti modulus in 30 seconds.. Duncan's multi-amplitude test showed that the moisture content whit a increased, all the properties extracted from the tests decreased. The grain strength decreases with increasing moisture. The results of ANOVA showed that all models were using the RSM to predict significant quality properties at (P <0.01) and (P <0.05). Prediction accuracy in the compression and bending tests were obtained with R2Pre ≥ 0.78 and R2Pre ≥94, respectively, and in the creep test using multiple regression with R2 ≥ 0.70. The results of the image processing test using multivariate regression (MLR) showed that predicting quality properties for the A, B and C treatments was estimated as R2 ≥ 0.90, R2 ≥ 0.83 and R2 ≥ 0.92, respectively. Based on the results obtained from the measurement of NIRS, calibration models in flour samples for chemical properties (AC, GT and PC) and physicochemical (BDV and SBV) were obtained as (R2cal ≥ 0.83) and (R2cal≥ 0.76), and prediction accurcy of (R2pre≥ 0.78) and (R2pre≥ 0.87), respectively. All PLS models, SDR index was more than 2. Calibration models in rice single kernel samples for chemical properties (AC, GT and PC) and physicochemical (MinV, BDV and SBV) were obtained as (R2 ≥ 0.85) and (R2 ≥ 0.93), and prediction as (R2pre≥ 0.79) and (R2pre≥ 0.79), respectively. SDR index for all models was more than 2. Conclusion: The values of extracted components from rice grain strength tests at two moisture levels of 9 and 14% predicted the values of rice quality properties with an high accuracy of 0.81. The conclusion of image processing test, the most suitable model for C was obtained with R2≥ 0.92. The conclusion of NIRS showed that the most appropriate of prediction model in rice flour samples for AC, GT, PC, BDV and SBV as Kadus, Hashemi, (Khazar & Kadus), Domsiah, varieties for rice flour samples with prediction accurcy were obtained as 0.81, 0.85, 0.85, 0.89, and 0.92, respectively. In rice single kernel samples for AC, GT, PC, Min.V, BDV, and SBV as Hashemi, Dorfak, Hashemi, (Khazar & Dorfak), Damsiah and (Dorfak & Kadus) varieties with prediction accurcy were obtained as 0.77, 0.90, 0.81, 0.84, 0.90, and 0.98, respectively. The correlation coefficients between the spectral data and the quality parameters of models were more than 0.60. This study usefal, appropriate and quick using from mechanical and optical tests to predict quality properties of rice.

Actions (login required)

View Item