Multivariable decoupling control based on neural network inverse system in a fermentation process

Fermentation process is a time-variable, nonlinear, uncertain and multivariable coupling system, and high performance decoupling control is a target to seek. A decoupling control strategy based on neural network inverse system for a multivariable fermentation process is proposed, in which the inverse system combines with the neural networks. Based on the characteristics of fermentation process, the model of fermentation system is obtained, and the reversibility of system is testified. Constructing a neural network inverse system and combining it with fermentation process, a pseudo-linear system is completed. Then a linear close-loop adjuster is designed to obtain the good control performance. The simulation experiments demonstrate that good control performance (high accuracy and good robust) can be obtained in multivariable fermentation process based on neural network inverse system, and the disadvantages of inverse system method relied on the exact process model and parameters are overcome.