Advanced Soft Sensor for High-Precision Control of Growth Rate in Fed-Batch Bioprocesses

This study presents the development of an advanced control system for regulating the specific growth rate of Lactobacillus rhamnosus in fed-batch cultures using an advanced soft sensor approach. A predictive Artificial Neural Network (ANN) model was developed to estimate growth rate in real time, utilizing input variables including total cell density, oxygen uptake rate, carbon evolution rate, dissolved oxygen, and bioreactor volume. The ANN model demonstrated high accuracy, with a validation-subset RMSE of 0.073 h−1 and an R2 of 0.99. To improve control stability, the predicted growth rate was integrated into a closed-loop Proportional-Integral (PI) controller that dynamically adjusted the substrate feed rate. The controller parameters were optimized using the Ziegler-Nichols method, achieving a mean control error of 18.2% ± 5.26% across different growth rate setpoints. The implementation reduced process variability and enhanced control precision compared to traditional control methods. These findings highlight the potential of ANN-based predictive models for improving bioprocess automation, indicating the potential of ANN-based predictive models for bioprocess automation, subject to validation under industrial conditions.