Gradient-Informed Machine Learning in Electromagnetics
This work addresses the computational bottleneck of repeated finite element simulations in electrical device design, offering a more efficient surrogate modeling approach for nonlinear parametric models.
The authors combined isogeometric analysis with proper orthogonal decomposition and Gaussian process regression to create a non-intrusive surrogate model for a parametric nonlinear permanent magnet synchronous machine, using gradient information from IGA to enhance surrogate accuracy.
Simulation techniques such as the finite element method are essential for designing electrical devices, but their computational cost can be prohibitive for repeated or real-time computations. Projection-based model order reduction techniques mitigate this by reducing the model size and complexity, yet face challenges when extended to nonlinear or non-affine parametric models. In this work, Isogeometric Analysis (IGA) is combined with proper orthogonal decomposition and Gaussian process regression to construct a non-intrusive surrogate model of a parametric nonlinear model of a permanent magnet synchronous machine. The differentiable nature of IGA allows for computationally efficient extraction of parametric sensitivities, which are leveraged for gradient-enhanced surrogate modeling.