Research on Multi-Objective Recipe Optimization Design Based on Genetic Algorithm and NSGA-II
Caoyan Wang, Haorui Tan, Yu‐Sheng Lai, Yixuan Wang · 2024
This paper presents a study on the optimization of recipes based on genetic algorithms and NSGA-II for multi-objective design. To address recipe optimization under different objectives, the study divides the optimization into single-objective and multi-objective categories, establishing a single-objective optimization model based on GA and a multi-objective optimization model based on NSGA-II. For the single-objective model, the objective functions include maximizing the protein amino acid score and minimizing meal costs, while the multi-objective optimization aims to simultaneously satisfy both objectives. For the maximization of amino acid scores, men improved from 93.5 to 98.1077, and women from 89 to 96.1483. In the case of minimizing meal costs, men improved from 18.5 to 6.5, and women from 16 to 6. The multi-objective optimization model achieved notable results, increasing the amino acid scores for men from 80 to 98.5064 and reducing meal costs from 28 to 6, while for women, the scores improved from 81 to 97.0679 and the costs decreased from 45 to 3. The comprehensive superior evaluation values were calculated as [99.2373, 90.9347], [97.2297, 88.9271], and [90.2272, 81.9246], with the multi-objective optimization model demonstrating the most effective performance.