A Recursive Prediction Error Method with Effective Use of Gradient-Functions to Adapt PMSM-Parameters Online
Chapter
Accepted version
Åpne
Permanent lenke
https://hdl.handle.net/11250/2740062Utgivelsesdato
2020Metadata
Vis full innførselSamlinger
- Institutt for elkraftteknikk [2503]
- Publikasjoner fra CRIStin - NTNU [38525]
Originalversjon
https://doi.org/10.1109/IAS44978.2020.9334744Sammendrag
This paper proposes a method for online estimation of electrical parameters of interior permanent magnet synchronous machines (IPMSM) based on the recursive prediction error method (RPEM). The parameter-sensitivity functions (herein known as the gradient functions, Ψ T ) both in dynamic and steady -states are exploited for this purpose. The RPEM has been computed using the stochastic gradient algorithm (SGA). The scalar Hessian matrix, r[k] appearing in the algorithm has been analyzed for both its steady and dynamic states. Different combinations of Ψ T and r[k] -states have been simulated and compared with respect to performance when used for parameter adaptation.