Online Intelligent Controllers for an Enzyme Recovery Plant : Design Methodology and Performance

Abstract EN

This paper focuses on the development of intelligent controllers for use in a process of enzyme recovery from pineapple rind.

The proteolytic enzyme bromelain (EC 3.4.22.4) is precipitated with alcohol at low temperature in a fed-batch jacketed tank.

Temperature control is crucial to avoid irreversible protein denaturation.

Fuzzy or neural controllers offer a way of implementing solutions that cover dynamic and nonlinear processes.

The design methodology and a comparative study on the performance of fuzzy-PI, neurofuzzy, and neural network intelligent controllers are presented.

To tune the fuzzy PI Mamdani controller, various universes of discourse, rule bases, and membership function support sets were tested.

A neurofuzzy inference system (ANFIS), based on Takagi-Sugeno rules, and a model predictive controller, based on neural modeling, were developed and tested as well.

Using a Fieldbus network architecture, a coolant variable speed pump was driven by the controllers.

The experimental results show the effectiveness of fuzzy controllers in comparison to the neural predictive control.

The fuzzy PI controller exhibited a reduced error parameter (ITAE), lower power consumption, and better recovery of enzyme activity.