Fuzzy logic application to model caffeine release from hydrogel colloidosomes

Highlights

• Analysis of a Fick’s model comprising a fuzzy model was performed in order to assess the diffusion coefficient.

• The modeling of caffeine release on hydrogel colloidosome by means of this diffusional-fuzzy model was evaluated in contrast to three other approaches.

• The fuzzy sets for the antecedent of the fuzzy system were determined by the Gustafson-Kessel clustering algorithm and approximated by membership functions of piecewise exponential form.

• More favorable results were attained from the diffusional-fuzzy model compared to the others.

Abstract

The diffusion coefficients of mass transfer in colloidosomes are sensitive to environmental changes. Also, Fick’s law of diffusion cannot be solved analytically in presence more complex geometries or non-constant active agent diffusivities. Because of the approximation property and uncertainty handing of fuzzy systems, here they are employed to model diffusion coefficients in the caffeine release from hydrogel colloidosomes. The approximated diffusion coefficients are then used in the Fick’s law equations, to create the fuzzy-diffusional model. The identification and validation of the fuzzy-diffusion model is obtained from experimental caffeine release curves of colloidosome samples in different shear rates, and at two time ranges. The proposed method is then compared with three other methods and depicts lower error and variance both in identification and evaluation. The diffusional-fuzzy model maintains the interpretability of the Fick’s model, improves the process simulation and eliminates phenomenon and property considerations.

Introduction

Microencapsulation is the most frequently used technique to accomplish controlled release in the food industry. Controlled release is a method for providing additional protection to a flavoring point. The delay in release can lessen volatilisation, oxidation or chemical reaction, as a result of which the performance of the flavoring considerably improves (Pothakamury and Barbosa-Canovas, 1995, Taylor and Hort, 2007).

While many flavor encapsulants are efficient, it is essential that the flavor molecules are released when the food is consumed; otherwise the flavor, despite being in suitable condition, will not be sensed (Roberts and Taylor, 2000).

Flavor release depends on the nature of the foodstuff itself and the way in which it is consumed. The state of the foodstuff can alter substantially as it is broken down by chewing, and mixed with saliva (Voilley and Etievant, 2006).

Flavor or any other active agents are released from the controlled release delivery systems by diffusion, degradation, swelling, osmotic pressure and melting. Diffusion is controlled by solubility and permeability of compounds in food matrices., In food systems it is the dominant mechanism in controlled release from encapsulation matrices. The principal steps in the release of a flavor compound from the matrix system include: diffusion to the surface of the matrix; partition between the matrix and the surrounding food and transportation of the flavor compound away from the matrix surface (Madene et al., 2006).

Colloidosomes as an innovative class of microcapsules are formed by the self–assembly of colloidal particles to the interface of emulsion droplet. By this means, a variety of release strategies may be attainable. Colloidosomes have a wide range of applications in different areas such as pharmaceuticals, food, flavors, fragrances, and cosmetics industries (Parthibarajan et al., 2011).

The most appealing characteristic of alginate is its versatility of gel formation simply induced by various different cations. Alginate beads have been used in microencapsulation as their preparation on a laboratory scale is undemanding because of their gentle hydrogelling process, relatively economical rate and safety for use in food formulations (Madziva et al., 2005).

Fuzzy systems are intelligent paradigms that have approximation property, uncertainty handling, and incorporating human knowledge. Modeling and control methods based on the fuzzy systems attempt to integrate numerical and symbolic processing into a single framework. Using linguistic knowledge, fuzzy systems integrate human knowledge into the control structure. In comparison with other nonlinear approximation techniques, fuzzy systems offer a more transparent representation of the nonlinear complexes that scientists handle in biological systems such as food related problems. In this manner, processed data can be translated in a model and analyzed in an approach analogous to what the public is cognizant of (Babuška and Verbruggen, 1996).

Recently, uncertainty of bioprocesses has persuaded scientists to search for new methods of solving fundamental phenomena including mass and heat transfer.

In order to design encapsulation systems especially with controlled release possibility, it is crucial to model the release data to determine the mass transfer properties of the capsules and reaching to desire release rates (Zuidam and Nedovic, 2010). In biological systems, alterations of production conditions are inevitable. In this case, the prepared spheres of hydrogel colloidosome are not uniform and equal (Amiryousefi et al., 2016). Therefore, the release characteristics of the prepared colloidosome samples do not follow a specific pattern. In fact the diffusion coefficients of caffeine release are not constant. First, this study aims to provide a simple, rapid and non-destructive method to encapsulate caffeine as a flavor model in colloidosome. Second, to handle the uncertainties in bioprocesses, here a Diffusional-fuzzy model is introduced that approximates diffusion coefficients in the caffeine release from hydrogel colloidosomes. The approximated coefficient is then used in the Fick’s law equations in order to get into more realistic solutions with fewer assumptions.

The overall structure of the paper is as follows: in Section 2 the materials and the methods are described. Results and discussions are brought in Section 3, and finally, conclusions are drawn in Section 4.