In this study, an integrated optimization framework that combined artificial neural network-genetic algorithm (ANN-GA) modeling, multi-sensor online monitoring, and metabolic profiling was developed. The ANN-GA model outperformed the traditional methods in terms of prediction accuracy, exhibiting superior fitting capability and enhanced performance of gentamicin C1a production. Real-time fermentation control was achieved via integrated near-infrared and Raman spectroscopy, yielding a 54.6 % titer increase (385.3 mg/L) over control group (249.2 mg/L). Metabolomic and metabolic flux analyses revealed a 36.2 % and 18.4 % reduction in glycolysis and tricarboxylic acid cycle fluxes, respectively, with an 11.3 % increase in pentose phosphate pathway flux and enhanced NADPH availability. Carbon flux was redirected toward biosynthetic intermediates, with key precursor pools such as glucose-1-phosphate and gentamicin X2 increasing by 52.8 % and 61.7 %, respectively. Overall, the approach significantly enhanced gentamicin C1a titer and laid the foundation for a scalable paradigm and theoretical framework for the intelligent optimization of industrial bioprocesses.
Keywords: Artificial neural network; Fermentation optimization; Genetic algorithm; Metabolic flux analysis; Multi-sensor monitoring.
Copyright © 2025 Elsevier Ltd. All rights reserved.