Published: CABEQ 37 (1) (2023) 17-32
Paper type: Original Scientific Paper
E. Emori, M. A. D. S. S. Ravagnani and C. B. B. Costa
The sugarcane crushing stage is one of the most important technologies being developed at the moment. In this paper, the control of the multiple-stage evaporation system was addressed, as it is a crucial stage in the first- and second-generation ethanol production from sugarcane. A neural network model was proposed based on a dynamic phenomenological model developed in EMSO (Environment for Modeling, Simulation and Optimization). The phenomenological model was used to build a neural network prediction model for an MPC (Model Predictive Control) scheme using a DMC (Dynamic Matrix Control) algorithm. Simulations were carried out to evaluate the performance for tracking the set-point. Also, disturbance rejection tests were performed, considering different step disturbances. The analysis demonstrated that the MPC scheme performed well in the tests and showed superiority when compared to classical PID controllers.
This work is licensed under a Creative Commons Attribution 4.0 International License
model predictive control, neural network, multiple-effect evaporation, EMSO, second-generation ethanol, dynamic matrix control