PhD Saeideh Khatiry Goharoodi - Data-Driven Symbolic Models for Mechatronic System Identification and Control

When
18-01-2022 from 15:00 to 16:00
Language
English
Website
https://tinyurl.com/PhDSaeidehKhatiry

PhD Saeideh Khatiry Goharoodi - Data-Driven Symbolic Models for Mechatronic System Identification and Control

The primary goal of this doctoral thesis is to automate the process of finding ‘white-box' models that are interpretable and consist of symbolic equations with focus on mechatronic systems.

To achieve this goal, we develop new algorithms for discovering mathematical models using symbolic regression. We focus on the application of symbolic regression towards mechatronic system identification, i.e. finding a symbolic expression for the governing input-output relationship of a complex mechatronic system; and control, i.e. discovering the control law from data.

Our primary research goal is realized through three subgoals: (i) automated model building for system identification and control, (ii) automated model selection and (iii) realization of symbolic models and their application on mechatronic systems, namely mechanical Duffing oscillator, electric induction machine and weaving mill as examples of mechatronic systems.

The developed algorithms and methods bring us many steps closer to the goal of realizing automated mathematical model extraction on physically dynamic systems in which the modeling cost is simplified and the user interaction can be limited or even eliminated.