Structure-Preserving Model Reduction for Large-scale Linear and Nonlinear Network Systems

Model reduction methods for linear and nonlinear network control systems will be further explored, which includes three aspects:
Firstly, model reduction methods for the individual linear and nonlinear subsystems will be generalized. The empirical differential balanced truncation method is generalized to the nonlinear system with non-constant input matrix and a structure-preserving approach for nonlinear systems will be proposed according to extended balancing.
Secondly, network structure preserving model reduction methods for network systems with switching topologies and nonlinear couplings will be studied. The clustering-based projection methods as well as balanced truncation based methods will be generalized to reduce the network structures with switching topologies and nonlinear couplings.
Finally, the model reduction method of the whole network system will be further researched, considering the switching topologies and nonlinearity.

Supervisors: Jacquelien Scherpen and Ming Cao
PhD: Yangming Dou