Supervisor:
Frede Blaabjerg

Co-Supervisor:
Subham Sahoo

Ariya Sangwongwanich

Assessment Committee:
Baoze Wei (Chair)
Dr. Stefan Vassilev Mollov, Austrian Silicon Labs, Austria
Professor Rosa Anna Mastromauro, University of Florence, Italy

Moderator:
Saeed Peyghami

Abstract:

The reliability of power electronic converters is increasingly critical as their deployment grows with renewable energy and electricity demand. This thesis focuses on data-driven condition monitoring, which extracts health indicators directly from measured data. Four key challenges are addressed: estimation accuracy, noninvasive online implementation, systematic tool development, and sampling schemes. Contributions include: (i) improved capacitor health estimation using time-domain measurements and particle swarm optimization; (ii) a noninvasive monitoring method for power semiconductors using control variables; (iii) loss-landscape analysis to guide multi-parameter estimation; and (iv) switching signal reconstruction to enhance sampling efficiency. Overall, the work advances low-cost, accurate, and practical monitoring methods to improve converter reliability.