Decoupled Neuro-Fuzzy Controller for the Cooking-Pressing Process in a Fishmeal Plant

Gutarra A.; Martins G.S.; Alegria E.J. · 2025 16th IEEE International Conference on Industry Applications, INDUSCON 2025 - Proceedings (2025) · DOI: 10.1109/induscon66435.2025.11241346

Resumen

This paper presents the improvement of the cooking-pressing system in a fishmeal plant to produce high-quality meal with low moisture and fat content. Due to limitations in accessing the plant, a computational model based on a Nonlinear AutorRegressive model with eXogenous inputs (NARX) neural network was developed using experimental process data. A decoupled neuro-fuzzy control strategy was proposed, employing Adaptive Neuro-Fuzzy Inference System (ANFIS)-tuned membership functions to optimize the fuzzy controllers. Different fuzzy controller structures - Proportional-Derivative (PD)-type, Proportional-Integral (PI)-type, and hybrid PI-PD-type - were evaluated under typical disturbances and noise to assess robustness in a simulation environment. Implementation considerations for deployment on a real plant were also discussed. Results demonstrate that the proposed neuro-fuzzy controllers meet design specifications and improve disturbance rejection, enhancing the overall efficiency of the cooking-pressing system.

Instituciones: Universidad de Ingeniería y Tecnología