|D'AMICO WILLIAM||Cycle: XXXVI |
Section: Systems and ControlMajor Research topic
:Direct Nonlinear Control Design with Recurrent Neural NetworksAbstract:
Nowadays, data science is increasingly spreading due to the recent introduction of innovative tools and algorithms for extracting information from data. The goal of the project is to apply Recurrent Neural Networks, and in particular Echo State Networks, Long Short Term Memory networks and Gated Recurrent Unit networks, to direct data-driven methods for control of nonlinear systems. The stability, robustness and disturbance rejection properties together with the choice of the most suitable reference model will be investigated. Notable case studies will be considered to compare direct methods with indirect ones and traditional control strategies.