Development of tools for multi-objective optimization problems with uncertainty: Application to the robust control design process for nonlinear systems
- Autores: Uriel Veyna Robles
- Directores de la Tesis: Juan Manuel Herrero Durá (dir. tes.), Francesc Xavier Blasco Ferragud (dir. tes.)
- Lectura: En la Universitat Politècnica de València ( España ) en 2025
- Idioma: español
- Tribunal Calificador de la Tesis: Eloy Irigoyen Gordo (presid.), Javier Sanchis Sáez (secret.), Jean Christophe Farges (voc.)
- Programa de doctorado: Programa de Doctorado en Automática, Robótica e Informática Industrial por la Universitat Politècnica de València
- Materias:
- Enlaces
- Tesis en acceso abierto en: RiuNet
- Resumen
Advanced control systems are typically tuned using dynamic models represented by differential equations, with parameters that may exhibit a certain degree of uncertainty. Additionally, it is common to aim at satisfying multiple design objectives, which are often in conflict with each other. Traditionally, control design problems addressed under this approach aim to obtain the Pareto set of solutions that optimizes a nominal parameter model, which is adjusted using experimental data in an identification stage. However, in practical applications, even slight parameter variations can significantly impact the performance of theoretically optimized solutions, leading to reduced efficiency.
This thesis contributes significantly to the multi-objective optimization design process for robust control tuning applications. It introduces innovative strategies, methods, and guidelines to improve and streamline the stages of problem formulation, optimization process, and decision-making. To enhance problem formulation, a design methodology is introduced for modelling parameter uncertainty in multivariable nonlinear systems. This method aims to define a set of uncertainty scenarios with highly representative properties that enable a feasible computational cost optimization process that leads to tuning robust solutions without attributing excessive conservatism. A new definition of robustness for multi-objective problems is established to support the optimization process. This definition underpins an innovative strategy for obtaining robust solutions in the presence of uncertainty. This approach succeeds in defining solutions that stand out for both optimality and robustness properties, aligning with conventional and unconventional multi-objective methods.
Two robust control design problems are addressed to demonstrate the benefits and advantages of the strategies and methods proposed in this thesis. The first involves tuning the temperature control of a proton exchange membrane fuel cell stack, while the second focuses on system control design for the transitional flight of an unmanned aerial vehicle tail-sitter. The analysis of results across both case studies underscores the versatility and effectiveness of the employed design approach. A comprehensive evaluation of performance under uncertainty highlights the tuned controllers for their diverse optimality and robustness attributes, enabling designers to tailor the final solution more specific preferences during the decision-making stage.
