Multi-criteria Decision Analysis: Linear and Non-linear Optimization of Aqueous Herbal Extracts

Abstract

This chapter is aimed to present the multi-criteria decision analysis: Linear and non-linear optimization of aqueous herbal extracts. Modelling is an indispensable part of food production, from “farm to fork”, where it is used to optimize the initial production of food and feed as well as in the food and feed processing. Different particle sizes of olive leaves were used in extraction of biologically active components using water as a solvent. Experiment conditions varied in mixing times (5, 10, 15 min), heating treatments (40 °C, 60 °C, 80 °C), Revolutions per minute: rpm (250, 500, 750 min−1) and particle sizes (100, 300, 500 μm). Based on the measured bioactive compounds (pH, total dissolved solids, conductivity, dry matter, total polyphenols and the antioxidant capacity by ABTS method, DPPH method and FRAP method). Aim was to develop models to support the optimization of this decision-making process—find the best experiment conditions for extraction of a certain bioactive compound. Two approaches; linear and nonlinear approaches were investigated. Linear optimization is presented with two models: Response Surface Methodology and using linear programing based on the Simplex method while the nonlinear approach is presented by developing membership functions using fuzzy logic approach. Final results showed that, simple or complex, i.e. linear or nonlinear approach(es) in the search for optimal experiment conditions in extraction of bioactive compounds from olive leaves, will lead to an optimal solution, but the engineer will decide which approach is suitable for further application. Linear optimization and application of fuzzy logic resulted with the best possible offer per set limitations. But each approach resulted with other optimal extraction conditions. However, the application of fuzziness allowed the extension of the set of acceptable experiment combinations to achieve the best extraction of a bioactive component. © 2018, Springer International Publishing AG, part of Springer Nature.