Modeling and Parameter Updating for Nosiheptide Fed-Batch Fermentation Process

详细信息    查看全文

• 作者：Dapeng Niu ; Long Zhang ; Fuli Wang
• 刊名：Industrial & Engineering Chemistry Research
• 出版年：2016
• 出版时间：August 3, 2016
• 年：2016
• 卷：55
• 期：30
• 页码：8395-8402
• 全文大小：356K
• 年卷期：0
• ISSN：1520-5045

文摘

Nosiheptide is a sulfur-containing peptide antibiotic obtained through fermentation. It can be used as feed additives because of its relative safety and good effect. However, nosiheptide fermentation does not have a high yield. Keeping the fermentation environment or operating conditions optimum through optimization is an effective way to improve nosiheptide’s yield, while accurate and reliable process models are the basis to achieve process optimization. Based on the reaction mechanism of the nosiheptide fed-batch fermentation process, we establish its mechanism models. Fermentation processes have slow time-varying characteristics and the conditions usually change due to disturbances during the production process, so the accuracy of established models tends to decline. Thus, models do not match the actual process gradually, leading to model mismatch, which have bad effects on the optimization and control of the process. Therefore, it is necessary to update the process models in time. In this paper, we update process models through identification of model parameters. Because there are many parameters in nosiheptide fed-batch fermentation process models, the calculation cost is quite large to correct all the parameters simultaneously. Considering that different parameters have different effects on process model output, we propose a model updating method based on parameter sensitivity analysis. During the model updating process, we identify and correct the parameters having a main impact on the models, while ignoring the parameters having secondary effects. Thus, we realize updating of the mechanism models for nosiheptide fed-batch fermentation process. Simulation results show that the proposed model updating method improves the models’ accuracy.