From traditional methods to intelligent control – a comparison of algae cultivation strategies with a focus on reinforcement learning

Algae cultivation is gaining significance in biotechnology, particularly in biofuel production, CO 2 capture, wastewater treatment, and the creation of high-value bioproducts. The efficiency and scalability of cultivation systems depend on both the chosen strategy and environmental control methods. This paper critically compares several control approaches: constant-condition systems, rule-based techniques, model predictive control, and adaptive feedback. A central focus is placed on reinforcement learning (RL), highlighted as a promising solution for dynamic and real-time optimization. Unlike traditional methods based on static rules or complex mechanistic models, RL can autonomously learn and adapt control policies through interaction with the environment. This enables responsive management of key variables such as light intensity, temperature, pH, and dissolved oxygen. The results indicate that RL can effectively optimize algae cultivation under changing conditions, positioning it as a strong candidate for the development of intelligent, self-optimizing cultivation systems.