Agentic Artificial Intelligence-Driven Neutrosophic Multi-Objective Optimization of Oxygen Transfer Systems in Smart Bioreactors

Oxygen transfer control in industrial bioreactors becomes more complex when cell metabolic changes, nonlinear process dynamics, energy constraints, and biological and operational uncertainties simultaneously affect control decisions. This study presents an innovative framework for intelligent oxygenation management that integrates a synchronized digital twin, agent-based artificial intelligence, fuzzy-notrosophic modeling, and multi-objective decision-making in a closed, uncertainty-aware loop. The proposed framework enables predictive and safe decision-making under variable conditions by continuously monitoring and reconstructing the process state, predicting short-term behavior, generating and screening control options, and virtually evaluating them before implementation. Evaluation under normal operating conditions and biological and process disturbances showed that the complete framework achieved 98.04% dissolved oxygen stability and reduced the recovery time after oxygen demand shock to 7.26 min. The findings show that the main value of this architecture lies not in the single maximization of oxygen transfer, but in creating an adaptive balance between oxygen delivery, energy consumption, shear stress, biological stability, and operational risk. This framework provides an extensible platform for moving toward autonomous, resilient, and sustainable bioreactors.