Optimization of fed-batch fermentation processes with bio-inspired algorithms

• Published in: Expert systems with applications Vol. 41; no. 5; pp. 2186 - 2195
• Main Authors: Rocha, Miguel, Mendes, Rui, Rocha, Orlando, Rocha, Isabel, Ferreira, Eugénio C.
• Format: Journal Article
• Language: English
• Published: Amsterdam Elsevier Ltd 01.04.2014; Elsevier
• Subjects: Algorithmics. Computability. Computer arithmetics; Algorithms; Applied sciences; Artificial intelligence; Biological and medical sciences; Biotechnology; Computer science; control theory; systems; Control theory. Systems; Convergence; Differential Evolution; Evolution; Evolutionary algorithms; Exact sciences and technology; Expert systems; Fed-batch fermentation; Feeding trajectory optimization; Fermentation; Fundamental and applied biological sciences. Psychology; Mathematical models; Methods. Procedures. Technologies; Microbial engineering. Fermentation and microbial culture technology; Optimization; Particle Swarm Optimization; Process control. Computer integrated manufacturing; Swarm intelligence; Theoretical computing; Evolutionary algorithms; Feeding trajectory optimization; Differential Evolution; Particle Swarm Optimization; Fed-batch fermentation; Validation; Statistical analysis; Probabilistic approach; Evolutionary algorithm; Process control; Fermentation; Particle swarm optimization; Convergence speed; Case study; Biomimetics; Heuristic method; Swarm intelligence; Metamodel; Semicontinuous
• ISSN: 0957-4174; 1873-6793; 1873-6793
• DOI: 10.1016/j.eswa.2013.09.017

•We optimize feeding trajectories in fed-batch fermentation processes.•We use several case studies from the literature.•We compare algorithms using statistical validation of the results.•We conclude that Differential Evolution is the best overall.•We provide a computational application to allow users to apply the proposed methods. The optimization of the feeding trajectories in fed-batch fermentation processes is a complex problem that has gained attention given its significant economical impact. A number of bio-inspired algorithms have approached this task with considerable success, but systematic and statistically significant comparisons of the different alternatives are still lacking. In this paper, the performance of different metaheuristics, such as Evolutionary Algorithms (EAs), Differential Evolution (DE) and Particle Swarm Optimization (PSO) is compared, resorting to several case studies taken from literature and conducting a thorough statistical validation of the results. DE obtains the best overall performance, showing a consistent ability to find good solutions and presenting a good convergence speed, with the DE/rand variants being the ones with the best performance. A freely available computational application, OptFerm, is described that provides an interface allowing users to apply the proposed methods to their own models and data.