基于逆系统的青霉素发酵软测量

Soft Sensor of Penicillin Fermentation Based on Inverse Systems

针对青霉素发酵过程中菌体浓度、基质浓度、产物浓度等关键参量难以直接测量的难题,将逆系统方法与动态递归模糊神经网络(DRFNN)相结合,提出一种基于动态递归模糊神经逆的青霉素发酵软测量方法。在证明了系统可逆的条件下,得到系统的逆模型;再应用DRFNN网络所具有的自学习,自适应能力以及对任意非线性的逼近能力,对该模型进行了辨识,并将辨识好的逆模型串联在发酵系统之后,能够实现发酵系统的"线性化"。仿真结果表明,该方法能够对青霉素发酵过程中不可在线测量的关键变量实现了预估,且达到了较高的测量精度。

An soft sensor based on dynamic recursive fuzzy neural network inverse system is proposed to measure the important parameters of penicillin fermentation process,such as substrate concentration,biomass concentration and production concentration.The reversibility of system is testified,and the inverse model of the system is obtained.The using dynamic recursive fuzzy neural network with the self-learning,adaptive,and arbitrary nonlinear approximation ability to identify this model,and the identified inverse model is cascaded to the fermentation system which makes the pseudo-linear system linearized.The simulation result shows that the method can precisely measure the key parameters which could not be measured online during the course of penicillin fermentation.