Validating Methods for Non-Linear System Identification
Supervisors: Dr Rob Barthorpe and Prof Keith Worden

Understanding non-linear systems is central to a multitude of engineering applications, from evaluating the performance of civil structures to the development of advanced automotive shock absorbers. Developing such understanding requires the development of non-linear models and the application of appropriate system identification techniques.

Performing such identification is generally challenging, and the cost of failing to develop a model of a non-linear system that works in reality can be high. A small variation in system parameters can lead to large qualitative changes in the nature of the system - a phenomena known as bifurcation. The danger is that a failure to correctly identify the model parameters may result in instability or failure in the real-world system.

The aim of this PhD project is to address the problems that lead to non-linear models failing to represent reality, and to address the challenges that arise when seeking to experimentally validate non-linear system identification methods.

Previous experience / requirements: The project would ideally suit a candidate with previous experience in numerical modelling, statistical methods and/or dynamic testing.
Apply online