Consideration of Magnetic Coupling in Segmented Dual Three-Phase Electrical Machines with Residual Model Approach
Year: 2026
Type: conference-paper
Abstract: Segmented asymmetrical machines (SAM) form a subclass of dual three-phase electrical machines that provide additional degrees of freedom for electromagnetic design and operation. A key challenge for these machines is accurately modeling the magnetic coupling between the two three-phase systems, which typically requires computationally expensive coupled finite element analysis (FEA). This paper introduces a residual-based modeling approach as a universal strategy for the simulation of asymmetrical machine behavior. Probabilistic machine learning models are employed to represent the current deviations induced by magnetic coupling, and uncertainty quantification is used to identify relevant asymmetric operating points within a scattered FEA framework. The proposed method reduces the required coupled FEA computation time by a factor of 8.7 while preserving accuracy in predicting dynamic torque characteristics and loss behavior. The results show that the scattered residual-based approach agrees well with fully coupled gridded models and is particularly effective for capturing losses and magnetic coupling effects, thereby enabling efficient design optimization of SAM with explicit consideration of mutual coupling.
Institution: Karlsruhe Institute of Technology
Language: English
Subfield: Electrical and Electronic Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W7225200051