Artificial Neural Network and Regression Models to Evaluate Rheological Properties of Selected Brazilian Honeys

Abstract

The relationships between physico-chemical and rheological properties are considered complex nonlinear systems. Thus, the artificial neural network (ANN) and regression models were used for the rheological characterization of Brazilian honeys, based on low-cost measurements of water content and temperature. The steady shear viscosity (η) performed well when measured in the test phase in a 2-12-1 neuron multilayer perceptron (MLP) ANN (model 1) with a root mean square error (RMSE) and correlation coefficient (r) equal to 0.0430 and 0.9681, respectively. The parameter loss modulus (G″), storage modulus (G′) and complex viscosity (η*) were predicted in the temperature sweep test by small amplitude oscillatory shear (SAOS) measurements during heating and cooling, and the MLP ANNs with architectures of 2-9-3 (model 2) and 2-3-3 (model 3) showed RMSE values equal to 0.0261 and 0.0387 in the test phase, respectively. For all the determined parameters, non-linear exponential models showed similar results to models 1, 2 and 3. An ANN with 3-9-3 architecture (model 4) showed RMSE and r for G′ equal to 0.0158 and 0.7301, for G″ equal to 0.0176 and 0.9581, and for η* equal to 0.0407 and 0.9647, respectively, in the test phase for date of the frequency sweep test obtained by SAOS. These results were far superior to those obtained by second-order multiple linear models. The acquisition of all models is an important application for the processing of honey and honey-based products, since these properties are essential in engineering calculations and quality control of products.

Analytical Rheology of Honey: A State-of-the-Art Review

“…The worst results observed for G expressed the high complexity and nonlinearity in its relationships with the input variables, in a similar behaviour already noticed by other authors [68]. Model 4 presented a superior predictive power when compared to a second-order polynomial regression model [71]. Overall, the obtained results pointed to the potential of ANN models to be applied for the processing of honey and honey-based products in engineering calculations and quality control domains [71].…”

“…Model 4 presented a superior predictive power when compared to a second-order polynomial regression model [71]. Overall, the obtained results pointed to the potential of ANN models to be applied for the processing of honey and honey-based products in engineering calculations and quality control domains [71].…”

“…Once the linear viscoelastic region (LVR) is defined through the strain sweep test, a food's structure can be characterized by a frequency sweep test at a strain below the critical value providing understanding about the effect of the interactions forces between colloidal particles [21]. Furthermore, heating or temperature fluctuations applied to a material promote structural changes, which are determined with great sensitivity by temperature sweep tests [29,71]. The oscillation creates deformations small enough to be in the LVR, and the elastic and loss moduli are both independent of the strain amplitude [9,70].…”

“…Rheological analysis combined with multivariate data analysis and machine learning techniques have been performed to predict rheological properties as a function of several variables (temperature, frequency, chemical composition) and to forecast the botanical and geographical origin of honey, for the purpose of authenticity certification [51,56,64,68,69,71,75]. Yilmaz et al [56] undertook the first rheological study on detection of honey adulteration, investigating the steady, dynamic shear and creep behaviour of natural Turkish honey adulterated with saccharose and fructose syrups at different ratios (0-50% w/w).…”

“…In another study [71], the authors employed ANN, namely MLP, to predict the rheological properties of the aforementioned Brazilian honeys [39,52], based on temperature and moisture content. Four independent ANN models were developed to predict η from steady shear measurements (model 1), G , G and η*, either from temperature scans during heating and cooling (models 2 and 3, respectively) or from frequency sweeps (model 4).…”