Chemical process modeling with multiple neural networks

It is difficult to identify some chemical processes which are operated in complex environments and the operation conditions are changed frequently. In this paper we combine the two effective identification tools, multiple models and dynamic neural networks, and propose a new class of identification approach. A hysteresis switching algorithm is used to select the best model in each time. The convergence of the multiple neuro identifier is proved. The simulation results show that the multiple neuro identifier has a better performance for the pH neutralization and the fermentation process.

[1]  J. Gauthier,et al.  A simple observer for nonlinear systems applications to bioreactors , 1992 .

[2]  José Ragot,et al.  Non-linear dynamic system identification: A multi-model approach , 1999 .

[3]  Kumpati S. Narendra,et al.  Adaptation and learning using multiple models, switching, and tuning , 1995 .

[4]  Jirí Benes,et al.  On neural networks , 1990, Kybernetika.

[5]  A. Morse,et al.  Applications of hysteresis switching in parameter adaptive control , 1992 .

[6]  Alexander S. Poznyak,et al.  Dynamic multilayer neural networks for nonlinear system on-line identification , 2000, Proceedings of the 2000 IEEE International Symposium on Intelligent Control. Held jointly with the 8th IEEE Mediterranean Conference on Control and Automation (Cat. No.00CH37147).

[7]  Manolis A. Christodoulou,et al.  Adaptive control of unknown plants using dynamical neural networks , 1994, IEEE Trans. Syst. Man Cybern..