Design and implementation of the soft-sensing system for fermentation process

In the process of penicillin fermentation, mycelium concentration and concentration of penicillin are important process parameters. But now, many of the detection means are off-line tests. They cannot be used in real-time control. RBF neural network has strong approximation ability and simple training method in indirect measurement applications. Using easily online measuring parameters, such as substrate concentration, RBF neural network could approximate the relationship between easily online measuring parameters and the important process parameters. This method provides a good solution for online measurement of mycelium concentration and penicillin concentration. Then virtual instrument which can measure mycelium concentration and concentration of penicillin is designed and implemented based on serial communication technology. Alternative programming technology between LabVIEW and MATLAB is used in the process of implementation.