Data-Driven Modelling

Data-driven modelling
Photo credit: David Brohede

Mathematical models of a system can be estimated from measured input and output data using methods for system identification and used later for various purposes, e.g., control design, signal prediction, diagnosis, and simulation. Some of the challenges with data-driven modelling are to handle dynamics, disturbances, mismatches between system and model, nonlinearities, prior knowledge, missing data, and closed-loop or network configurations. For a long time, LiU has been a world-leading research group in the area.