A Flow-Through Multi-Wavelength Sensor with Machine-Learning Calibration for Real-Time Monitoring of Microalgae Cultivation Processes

Abstract

Acquiring real-time biological data is essential for effective control of microalgae cultivation processes, yet routine monitoring still depended on laborious offline analyses that relied on time-consuming wet-chemical techniques. This study introduces a flow-through, multi-wavelength visible-light (VIS) sensor for real-time monitoring of biomass and pigment concentrations in microalgae cultivation processes. Based on 209 experimental data points, six machine-learning regression models were developed to estimate dry biomass, chlorophyll α, chlorophyll β, total chlorophyll, total carotenoid, and astaxanthin concentrations. Validation under realistic continuous operation showed that potential remains within improved hydrodynamics and operation automatics to reduced biofilm formation. The independent dataset demonstrated that biomass and astaxanthin predictions were within ±10% of offline reference measurements. The proposed low-cost and versatile multi-wavelength sensor platform, together with machine-learning-based calibration, provides a practical soft-sensor concept for real-time monitoring of microalgal bioprocesses and offers a foundation for future integration of model-based and predictive control strategies.