Preference-based multi-objectivization applied to decision support for High-Pressure Thermal processes in food treatment

Abstract

Food industry aims to provide healthy products that must satisfy the quality requirements of the considered legislation. To do so, food is treated by using some processing techniques, such as High-Pressure Thermal (HPT) treatments. In this work, we propose a preference-based multi-objectivization methodology to design HPT processes for food treatment. This approach is based on formulating a multi-objective optimization problem, instead of a constrained mono-objective problem, where the constraints are reformulated as separate objective functions. The multi-objective problem is then solved by using preference-based evolutionary optimization algorithms (PMOEAs). PMOEAs focus the search of a numerical solution inside a region of interest defined by the food engineer, avoiding exploring HPT designs that are out of interest. The proposed methodology is validated by considering several particular mono-objective and multi-objective optimization problems related to HPT processing. In particular, we compare the results obtained by two competitive state-of-the-art PMOEAs, called WASF-GA and R-NSGA-II, with the ones returned by a mono-objective algorithm called MLS-GA. As part of this study, the influence of the optimization algorithm parameters on the solutions, their quality and the computing time are discussed. Finally, the best solutions returned by the algorithm that shows a better performance for our problems, which is WASF-GA, are analyzed from a food engineering point of view and a sensitivity analysis regarding the impact of design parameters on the performances of those solutions is carried out.