Software sensor development for product concentration monitoring in fed-batch fermentation process using dynamic principal component regression

Monitoring and control of batch processes is more complicated than that of a continuous process. This is due to the fact that the properties change with time in a batch process. Penicillin production using fed-batch fermentation technique is one such process which is dynamic and highly non-linear in nature. In thsi research, a dynamic pricnipal component regression based soft-sensor model is proposed for continuous monitoring of penicillin concentration in the fed-bath fermentation reactor. The available data (generated using pensim simulator) were divided into training and validation data. The model was developed from the training data and accuracy testing was done by simulation of the model with validation data. Results show that the dynamic PCR model proposed in this work is able to capture the collinearity and dynamic nature of the data quite effectively and is able to predict the product concentration with good accuracy.