Multivariable decoupling control based on fuzzy-neural network ath-order inverse system in fermentation process

Abstract

Fermentation process is a time-variable, nonlinear, uncertain and multivariable coupling system, and high performance decoupling control is a target to seek. An a th-order inverse decoupling control strategy based on fuzzy-neural network inverse system for a multivariable fermentation process is proposed, in which the inverse system combines with the fuzzy-neural network. According to the nonlinear identification theory of fuzzy-neural network, the nonlinear offline inverse model of the plant was built by fuzzy inference system with cascade-forward backpropagation neural network, and the reversibility of system is testified. Based on the theory of inverse system method, the fuzzy-neural network inverse model was cascaded before the fermentation system to decouple a comples nonplex nonlinear multivariable system into several relatively independent single input single output pseudo-linear sub-systems, and used expert controller to carry on the optimization to the control system. The simulation experiments demonstrate that good control performance(high accuracy and good robust)can be obtained in multivariable fermenation process, and can be easily implemented.