< Back to previous page

Project

Data-driven modelling and control of nonlinear vibrations: Making the best of the system identification and machine learning worlds

Modern computers allow for increasingly easy and efficient processing of large amounts of data. For these reasons, system identification and especially machine learning techniques are becoming increasingly popular in the field of mechanics. Therefore, in line with the current developments, this research will be strongly data-driven. Moreover, it is specificially focused on nonlinear systems, since industrial systems are becoming increasingly complex, and resultingly their degree of nonlinearity is becoming increasingly high. However, in terms of control, linear models are typically assumed since they significantly ease control design and analysis. Yet, nonlinear phenomena can often not be ignored anymore, which reveals the need for future engineers that can handle such nonlinear phenomena. In this regard, possible research objectives are dynamic linearisation of input-output relations, or direct control of highly nonlinear systems, all while exploiting data-driven techniques.

Date:17 May 2021 →  Today
Keywords:Nonlinear system identification, Machine learning, Control
Disciplines:Control engineering, Signals and systems
Project type:PhD project