Hybridized multi-objective optimization approach (HMODE) for lysine fed-batch fermentation process

Abstract

A new hybrid multi-objective differential evolution (MODE) algorithm is proposed that combines the MODE algorithm for the global space search with a dynamical local search (DLS) method for the local space search. HMODE-DLS algorithm was validated using the tri-objective DTLZ7 test problem and the results were compared with MODE algorithm with respect to four performance metrics. In addition to HMODE-DLS, another three algorithms were used to solve two multi-objective optimization cases in an industrial lysine bioreactor at different feeding conditions. Case 1 considers maximizing lysine’s productivity and yield. While case 2 studies the maximization of productivity along with minimization of total operating time. In all cases, theoretical and industrial, HMODE-DLS showed a better performance with a better quality Pareto set of solutions. The Pareto front of case 1 found by HMODE-DLS was compared with a recent study trade-off, and the current non-dominated solutions values were found to be improved. This indicates that the lysine production process is enhanced. For case 2, the switching time from fed-batch to batch was found to be the key decision variable. Generally, these findings indicate the effectiveness of HMODE-DLS for the studied cases and its potential in solving real world complex problems.