System Identification – Mathematical Modeling Research

Identification of a parsimonious model of systems can be difficult depending on the nature of the data and/or priors available. It is difficult since, very often, the available data is noisy and has missing samples. Therefore, we are investigating a new randomized parsimonious system identification algorithm for Linear/Nonlinear systems. A new atomic norm Anormapproach has been used for proposed randomized algorithms.

Theoretical and Implementation Studies in the Field of Linear System Identification:

• Sparse System Identification Algorithm via Atomic Norm
• Modeling of health problems (Physical Activity, Smoking Behavior, Cancer, etc.)
• Developing a method for ghost elimination in bearing-only tracking for passive radars
• Investigating a new recursive model learning algorithm for thermal dynamical models
  
• Research on time-series forecasting methods
• System identification theory for testing/classification of the materials