Model-order reduction techniques for linear electromagnetic problems - an overview

Simeoni, M.; Vandenbosch, Guy A. E.; Lager, Ioan E.

doi:10.1109/eumc.2005.1610131

Cited by 4 publications

(5 citation statements)

References 42 publications

(38 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Their ability to generate compact simulation models of electromagnetic devices starting from their spatially discretized linear models has been demonstrated in several works [14], [67], [68], [60], [22]. Less focus has been put on applying model order reduction techniques to the nonlinear models of electromagnetic devices.…”

Section: Thesis Contributionmentioning

confidence: 99%

Model-Order Reduction of Moving Nonlinear Electromagnetic Devices

et al. 2008

View full text Add to dashboard Cite

This dissertation delivers a contribution to the field of order reduction of large-scale nonlinear models of electromagnetic devices. In particular, it enables applying model order reduction techniques to an important class of electromagnetic devices that contain moving components and materials with nonlinear magnetic properties. Such devices include among others rotating electrical machines, electromagnetic valves, electromagnetic solenoids, and electromechanical relays.The presented methods exploits the trajectory piecewise linear (TPWL) approach in approximating the nonlinear dependency of materials properties on the applied magnetic field. Additionally, the model nonlinearity that is caused by the movement of the device components is handled using a novel approach that updates the electromagnetic (EM) field model permanently according to the new components positions.The order of the large-scale electromagnetic field model is reduced by approximating the original electromagnetic field distribution by a linear combination of few virtual field distributions that are found using the proper orthogonal decomposition (POD) approach.The challenge of selecting the number and the position of the linearization points in the TPWL model is tackled using a new approach that considers the change in the magnetic properties of the device materials among all the simulated state-vectors.The new presented methods are extended to enable generating parametric reduced order models of moving nonlinear EM devices. Such models enable a fast and accurate prediction of the behavior of the EM device and its variations that result from changing the values of its design parameters. Additionally, several algorithms for generating an optimal reduction subspace of the parametric model are presented and compared.Finally, an approach for overcoming the challenge of generating reduced order models of EM devices while considering the strong influence of their power electronics driving circuits is introduced and applied to the example of a rotating electrical machine coupled to a power electronics driving circuit.The new methods presented in this work are validated by applying them on the examples of three industrial devices. An electrical transformer, an electromagnetic valve, and a rotating electrical machine. DEDICATIONTo my parents, my wife, and my daughter. ACKNOWLEDGMENTS

show abstract

Section: Thesis Contributionmentioning

confidence: 99%

Model-Order Reduction of Moving Nonlinear Electromagnetic Devices

et al. 2008

View full text Add to dashboard Cite

show abstract

“…The parameterized model is expressed only in terms of one parameter at a single frequency. Parameterizing both with respect to frequency [8] and geometry or parameterizing with respect to more than one geometrical parameter at a time would result in approximate cross terms that aggrandize the error values accumulatively.…”

Section: Calculation Of the Higher Order Derivativesmentioning

confidence: 99%

“…A prerequisite of calculating derivatives through (8) is to have the derivatives of the mesh elements with respect to geometry deformation. However, it is not possible to use finite differences for evaluating the element derivatives (A ðiÞ t ).…”

Section: Mesh Parameterizationmentioning

confidence: 99%

“…Another issue of interest is the region in where the parameterized model is trustable. There has been a great deal of previous work is reported on this subject Length of screw 4-4 2.60 p 10 Length of screw 5-5 1.50 p 11 Length of screw 2-3 0.75 p 12 Length of screw 4-5 1.25 12.3230 f 3 12.3240 f 4 12.3280 f 5 12.3300 f 6 12.3321 f 7 12.3340 f 8 12.3381 f 9 12.3430 f 10 12.3500 f 11 12.3560 f 12 12.3610 f 13 12.3660 f 14 12.3680 f 15 12.3680 f 16 12.3710 f 17 12.3750 f 18 12.3760 f 19 12.3820 (e.g., [24].) The parameterization results with respect to all design parameters for the nominal structure (Dimensions shown in Tables I and II) are shown in Figure 4.…”

Section: B Parameterized Model Verificationmentioning

confidence: 99%

“…The major approach for accelerating the EM solvers has included reduced order models [8] and frequency parameterization [8][9][10][11]. These methods have made it possible to calculate the objective functions in terms of field quantities over a frequency band with just few calculations.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Optimization of passive microwave devices through geometrical parameterization-Application to the design of a cylindrical dual-mode cavity filter

Mahanfar¹,

Bila²,

Aubourg³

et al. 2009

Int J RF and Microwave Comp Aid Eng

View full text Add to dashboard Cite

A new algorithm, based on geometrical parameterization and finite element method is presented for the optimization of microwave devices. Using geometrical parameterization, the field quantities are expressed as a polynomial in design parameters. Automatic differentiation is used for calculation of higher order derivatives. To ensure continuous gradients, an integrated mesh deformation algorithm is deployed to morph initial finite elements mesh into the perturbed geometry. The resulting parametric model is deployed through quadratic surface reconstruction to find local minimum at each stage. The convergence of the optimization through the reconstructed surfaces is discussed. As an example, a 5-pole dual-mode cavity filter is designed and optimized using the presented algorithm.

show abstract