Dynamic Mode Decomposition for Real-Time System Estimation of Induction Motor Drives

Abstract

There are many methods for real-time estimation and identification for induction motor (IM) drives. In this study, dynamic mode decomposition with control (DMDc) and its variants are explored for voltage source inverter fed IM. For real-time identification, online variants of the DMDc are explored, including windowed online DMDc. These algorithms are tested in simulation for healthy and faulty behaviors of the IM drive. They are compared for estimation accuracy and implementation complexity. Lastly, these algorithms are tested on a dSpace experimental platform to further prove their potential and applicability for both healthy and faulty cases. With accurate DMDc estimation results across different variants, their applications have potentials in condition monitoring or fault-tolerant control.