This project focuses on modeling, identification, control synthesis considering classical and automatic learning and artificial intelligence tools for complex dynamical systems. The modeling and identification problems will be concentrated on infinite-dimensional systems through the theory of partial differential equations (PDEs). Engineering applications can be found in many fields such as chemical and thermal processes, distribution and energy production systems, and systems related to fluid transport and medical science. Interestingly, several of these applications have parameters and mathematical functions that are little known in practice. In this context, we intend to address problems such as identifying the unknown structure of the equations that describe the underlying phenomenon given only measurements of the system’s output and integrating automatic learning and artificial intelligence techniques into the classical modeling and identification methodologies.
In the topics related to control theory, we will address those that arise in complex systems modeled by several coupled nonlinear PDEs. In general, the available results can be applied only in particular cases with fairly conservative assumptions. In this sense, we aim at developing control design methods combining innovative methods of systems theory, with modern optimization and artificial intelligence tools for the control and stability analysis. Our goal is to address these issues for the parabolic and hyperbolic cases. It turns out that artificial intelligence and machine learning tools will be used in all research areas of the project. This will make it possible to address some novel challenges in control and systems theory by combining well-known control engineering tools with artificial intelligence methods.
- Establish a strong international and interdisciplinary network between Brazil, Chile, and France to deal with the topics of modeling, identification, automation, and systems theory.
- Study of the properties of infinite-dimensional systems with engineering applications, such as chemical and thermal processes, distribution and energy production systems, and systems related to fluid transport and medical science.
- Study and development of model-based identification methods based on artificial intelligence techniques in order to obtain approximated models or to learn key parameters of the aforementioned systems.
- Development of control methodologies combining innovative and classical methods of systems theory to guarantee the asymptotic stability of the closed-loop systems under study.
- Application of the methods in prototypes available in the laboratories of the participating institutions, such as a Rijke tube, a 3D printer based on selective laser sintering, and a drilling system.