Steffen Weigel

Steffen Weigel

Auto-generated profile, based on publicly available information

Steffen Weigel

Institute of Electrical Energy Conversion, University of Stuttgart, Stuttgart, Germany

Research Focus & Metrics

Academic Career Length

2Years (2025-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
21 February 2025 in IEEE Transactions on Magnetics
Verified
Journal Article
Closed access
YGYuancong Gong
Yuancong Gong
AGAndreas Gneiting
Andreas Gneiting
SWSteffen Weigel
Steffen Weigel
NPNejila Parspour
Nejila Parspour
ZAZhouxing An
Zhouxing An
The design of traction machines for electric vehicles (EVs) typically aims to minimize total energy losses across an entire drive cycle, a process that can be highly time-consuming. To enhance the time efficiency in drive cycle-oriented optimization, this work proposes a surrogate model-based methodology for optimizing drive cycle energy loss, integrating both vehicle dynamics requirements and machine constraints. Initially, multiple modeling techniques for drive cycle energy loss, including point-by-point (PbP) search, loss map (LossMap), and k-means clustering, are assessed to identify the most accurate method. Subsequently, three prediction approaches using Kriging model are employed within the optimization framework. The results demonstrate that the PbP search method provides the highest accuracy in modeling drive cycle energy loss, while the predicted lookup table-driven sequential energy loss prediction (PLUT-SELP) achieves the optimal Pareto front among the prediction methods. Furthermore, while the PLUT-SELP method significantly improves the Pareto front and reduces computation time by 47.1% compared to the full-FEA optimization routine, the alternative methods hybrid input energy loss prediction (HIELP) and the combination of direct energy loss prediction (DELP) with PLUT-SELP achieve further time reductions of 50.7% and 53.9%, respectively, with a slight trade-off in optimization quality.

6 February 2025 in IEEE Transactions on Magnetics
Verified
Journal Article
Closed access
ASAdrian Schäfer
Adrian Schäfer
SWSteffen Weigel
Steffen Weigel
NPNejila Parspour
Nejila Parspour
For the efficient simulation and optimization of electrical machines, the soft magnetic behavior of core materials is typically represented by the magnetization curve rather than complex hysteresis models, and hysteresis losses are calculated and accounted for in post-processing instead. Currently, either interpolation methods or phenomenological magnetization models are used to describe the magnetization curve. However, the existing methods exhibit several disadvantages concerning identification, comparability, implementation, and the resulting behavior of the nonlinear solver. Based on these identified disadvantages, a set of criteria for magnetization models are developed. An analysis of the state of the art reveals that the existing methods do not meet these criteria. Therefore, three novel models are proposed to advance the state of the art: a simple algebraic model, a special arctangent model, and a special hyperbolic tangent model. To evaluate these models, the Langevin function, which describes anhysteretic behavior based on a physical approach in models, such as the Jiles-Atherton (JA) model, is used. The new methods are fitted to the anhysteretic Langevin functions of different materials and show excellent agreement in regression as well as improved computational efficiency compared to the Langevin function. The best models reach R-squared values exceeding 0.99 for the four reference materials. In particular, the algebraic and arctangent models exhibit excellent regression performance, and beyond, the algebraic model achieves outstanding computational efficiency, rendering them viable alternatives to the existing models for magnetization curves in technical simulations.