This project will consider the design of Stochastic MPC strategies based Computational Intelligence techniques such as fuzzy models and neural networks, but focusing on achieving theoretical properties such as stability and feasibility, increasing computation speed, and can use control policies. This is novel in that these properties are hardly ever obtained when using nonlinear models in SMPC. Additionally, we will aim to systematize the stability and convergence analyses. These developments will be validated on applications such as irrigation systems, climatization systems, microgrids. This is expected to improve the performance of existing control methods for systems with stochastic uncertainty, and enable the design of advanced control systems where the lack of guarantees or slow computation does not permit the implementation of advanced control systems.
Additionally, the development of ad-hoc controllers based on SMPC with RL for Electric Vehicle Routing will be tackled in this research. This controller will be tackled following principles particularly selected for this application since stability is not a major concern here because of the finite horizon for every day of operation.