Thesis Modelamiento y control MPC de un estanque presurizado
Loading...
Date
2025
Journal Title
Journal ISSN
Volume Title
Program
Ingeniería Civil Electrónica
Departament
Campus
Campus Casa Central Valparaíso
Abstract
El presente trabajo aborda el modelamiento y control de un estanque presurizado multivariable, utilizando Control Predictivo Basado en Modelos (MPC) en sus variantes lineal y no lineal. Se desarrolla primero un modelo matemático no lineal del proceso, el cual captura las interacciones y restricciones inherentes entre las variables de control del sistema. Posteriormente, este modelo es linealizado alrededor de un punto de operación de interés, lo que permite implementar un controlador predictivo basado en un modelo lineal para una comparación directa entre ambas aproximaciones. El trabajo se enfoca en implementar y evaluar esquemas de control predictivo en el entorno MATLAB/Simulink, permitiendo analizar el desempeño del sistema en condiciones controladas y realistas. A lo largo de las simulaciones, se evalúan aspectos clave como el seguimiento de referencias, la capacidad de rechazo de perturbaciones externas y los efectos de la sintonización de los horizontes de predicción y control. Además, se realiza un análisis detallado de los efectos de los pesos en la función de costo sobre el desempeño general de los controladores. Los resultados obtenidos muestran que el MPC lineal, debido a su menor costo computacional, permite ampliar los horizontes de predicción y control, lo que resulta en un mejor manejo de la dinámica lenta del nivel de líquido en comparación con otros métodos(...).
This work addresses the modeling and control of a multivariable pressurized tank system, using Model Predictive Control (MPC) in both linear and nonlinear variants. Initially, a nonlinear mathematical model of the process is developed, which captures the inherent interactions and constraints between the system’s control variables. Subsequently, this model is linearized around a point of interest, allowing for the implementation of a linear model-based predictive controller for a direct comparison between the two approaches. The work focuses on implementing and evaluating predictive control schemes within the MATLAB/Simulink environment, enabling the analysis of system performance under both controlled and realistic conditions. Throughout the simulations, key aspects such as reference tracking, disturbance rejection, and the effects of tuning prediction and control horizons are evaluated. Additionally, a detailed analysis of the effects of the cost function weights on the overall controller performance is performed. The results obtained show that linear MPC, due to its lower computational cost, allows for the expansion of prediction and control horizons, resulting in better management of the slow dynamics of the liquid level compared to other methods(...).
This work addresses the modeling and control of a multivariable pressurized tank system, using Model Predictive Control (MPC) in both linear and nonlinear variants. Initially, a nonlinear mathematical model of the process is developed, which captures the inherent interactions and constraints between the system’s control variables. Subsequently, this model is linearized around a point of interest, allowing for the implementation of a linear model-based predictive controller for a direct comparison between the two approaches. The work focuses on implementing and evaluating predictive control schemes within the MATLAB/Simulink environment, enabling the analysis of system performance under both controlled and realistic conditions. Throughout the simulations, key aspects such as reference tracking, disturbance rejection, and the effects of tuning prediction and control horizons are evaluated. Additionally, a detailed analysis of the effects of the cost function weights on the overall controller performance is performed. The results obtained show that linear MPC, due to its lower computational cost, allows for the expansion of prediction and control horizons, resulting in better management of the slow dynamics of the liquid level compared to other methods(...).
Description
Keywords
Control Predictivo, Modelamiento matemático, Linealización de sistemas, Función de costo, Redes neuronales, Perturbaciones
