Soft-sensing Modeling Based on PSO-FNN Inversion for Penicillin Fermentation Process

• Main Authors: B. Wang, X.F. Ji, Z.K. Zhuang
• Format: Article
• Language: English
• Published: AIDIC Servizi S.r.l. 2015-12-01
• Series: Chemical Engineering Transactions
• Online Access: https://www.cetjournal.it/index.php/cet/article/view/4392

As for the problem that the crucial parameters of penicillin fermentation process are difficult to be online measured, a soft-sensing modeling method is proposed based on PSO-FNN inversion. Firstly, a dynamic system model is developed based on the material balance relation of the penicillin fed-batch fermentation process, and existence of inverse system is analyzed. And then, an inverse model has been developed offline by use of fitting capacity of fuzzy neural network (FNN); online optimization is made by use of particle swarm optimization (PSO) algorithm on the basis of deviation information. Finally, to connect the inverse model and original fermentation process in serial into a compound pseudo-linear system, and based on the inverse system theory, realize the prediction of the crucial parameters. The simulation experiment shows that a better prediction for the crucial parameters of penicillin fermentation process is obtained based on PSO-FNN inversion algorithm.