FWOAL867 - Identification of nonlinear continuous-time systems from noisy input – noisy output observations
In depth understanding of the complex dynamics of physical processes is the first step towards prediction (e.g. weather forecasting), control (e.g. driverless cars), and (computer aided) design (e.g. electronic circuits, drugs). These complex dynamics are often described by nonlinear differential equations with a priori unknown dynamic order, unknown parameters, and unknown nonlinear structure. The major objective of this project is to extract (estimate) the unknown/missing information from (well-designed) experiments. Therefore, a frequency domain system identification framework for estimating nonlinear differential equations from noisy input – noisy output observations will be developed that requires as little user interaction as possible. Such a nonlinear modelling framework is currently not available, while it has potential impact on very diverse scientific disciplines ranging from bio-mechanics, nonlinear modal analysis, to the characterization of material properties and growth modelling in economics.