Abstracts

 
1 On-and off state losses in foil windings
Lavers, J.D.

High frequency DC-DC converters require transformers that are designed for optimal efficiency, particularly with respect to the copper losses when gapped cores are used. This paper examines the copper losses that arise in foil windings, particularly under open circuit conditions that approximate the conditions that arise in transformers for flyback converter applications. The study uses 2-D and 3-D finite element models to estimate losses in open circuit spiral windings, both interleaved and not interleaved. The validity of 2-D approximate models is evaluated and the effect of air gap placement is examined.
2 An accuracy assessment of 2-D vs. 3-D finite element models for ferrite core, sheet wound transformers
Lavers, J.D.; Lavers, E.D.

This study examines the parameter estimates for representative high frequency transformer designs with a view to assessing the errors that might be expected when 2-D approximate finite element models are used for the purpose of parameter extraction. The assessment has been based on a comparison of parameter estimates obtained from the 2-D models, relative to values predicted by fully converged 3-D models. The study has shown that the widely used 2-D models provide parameter estimates that, at worst, are in the order of 25% greater than the 3-D results. In many instances, the agreement is found to be much closer than that.
3 A novel finite analytic element method for solving eddy current problems with moving conductors
Dezhi Chen; Shao, K.R.; Lavers, J.D.
A novel finite analytic element method (FAEM) is presented. The basic idea of the method is the incorporation of local analytic solution of the governing equation in the finite element method, A local analytical solution satisfying its nodal conditions is found in each element and is used for determining the shape functions. Then, a weighted residuals scheme is followed to yield the linear algebraic equations. The presented FAEM is applied to solve 1-D and 2-D eddy current problems with moving conductors. Because the problem's analytical features have been considered, the solution in each element is approximated closely and the spurious oscillations which occur in the ordinary Galerkin solutions are avoided. High accuracy is obtained with no need of very fine meshes.
4 Vector potential due to a circular loop in a toroidal cavity in a high-permeability core
Namjoshi, K.V.; Lavers, J.D.; Jain, P.K.

We present a Green's function for a circular loop in a toroidal cavity in a high-permeability material. We show that under certain conditions, Green's function may be expressed in terms of Jacobi's theta functions. We demonstrate an application of such a Green's function in analyzing the performance of pot core transformers. Results compare well with the values obtained by other analytical methods and by the measurement.
5 Application of a wavelet transform in eigenvalue problems for electromagnetic field computations
Shao, K.R.; Yang, J.C.; Lavers, J.D.

In this paper, we present the wavelet transform (WT) method to reduce the dimension of a matrix without sacrificing too much precision on eigenvalues. By using the WT, the original matrix is split into two matrices, with each being a quarter the size of the original one. One of the derived matrices contains the smaller eigenvalues, while the other has larger ones. Moreover, this procedure can be taken successively, one can have a series of reduced order matrices.
6 Preparation and characterization of polymer-coated magnetic nanoparticles
Burke, N.A.D.; Stover, H.D.H.; Dawson, F.P.; Lavers, J.D.; Jain, P.K.; Oka, H.

Composite materials consisting of polystyrene-coated iron nanoparticles were prepared by the thermal decomposition of iron pentacarbonyl in the presence of polystyrene-tetraethylenepentamine dispersants. The nanocomposites contain both simple core-shell particles of 10-20 nm diameter, and more complex particles (20-100 nm) made from the agglomeration of several core-shell particles. Electron diffraction revealed that the core was composed of iron, in contrast to the iron nitride (Fe/sub 3/N) reported for similar conditions with a polyisobutylene-based dispersant. The materials exhibit hysteresis and /spl sigma//sub s/ ranges from 7.6 to 29.3 emu/g. The coercivity goes through a maximum for particles of /spl ap/25 nm diameter. Removal of unbound dispersant from the materials greatly increases the magnetization.
7 Constriction resistance at high signal frequencies
Lavers, J.D.; Timsit, R.S.

Constriction resistance arises in practical electrical interfaces because contact is made at discrete spots as defined by the surface roughness and contact pressure. This paper describes the dependence of constriction resistance on signal frequency. This dependence was calculated for circular constrictions ranging in diameter from 10 to 100 /spl mu/m, and for frequencies ranging from DC to 1 GHz. The results indicate that the magnitude of constriction resistance does not deviate appreciably from values predicted by Helm's classical analytical expression, as long as the skin depth is large compared with the constriction radius. For skin depths that are much smaller than the constriction radius, constriction resistance decreases with increasing frequency to an apparent limiting value independent of the constriction radius. At high frequencies, constriction resistance constitutes only one of two components of the total connection resistance measured in practice. The second component of connection resistance is determined by details of the geometry and dimensions of the contact interface, and increases with signal frequency.
8 On the use of recurrent neuro-fuzzy networks for predictive control
Sadeghian, A.R.; Lavers, J.D.

This paper presents the application of recurrent neuro-fuzzy networks for the predictive control of nonlinear, multivariable, complex systems such as electric arc furnaces. The main objectives are to investigate the capability of adaptive neuro-fuzzy networks to predict the V-I characteristics of electric arc furnaces and to compare the performance of the proposed predictors with that of the feedforward neuro-fuzzy predictors. The novelties of this work are to propose the notion of approximate prediction and to implement it using a recurrent neuro-fuzzy structure suitable for long-term prediction. Successful implementations of recurrent neuro-fuzzy predictors are described and their performances are illustrated
9 Global accuracy of the FEM solutions based on Ampere's law for a highly saturated magnetic device with a large air-gap
Sharifi, M.; Lavers, J.D.

The finite element method is widely used to model electromagnetic devices. Ensuring that the FE model is accurate is particularly important where measured data of the large prototypes are not available. This paper proposes to use Ampere's circuital law to obtain a global accuracy criterion that measures the accuracy of the whole FEM model and solutions. The accuracy of the FEM approximate solutions for highly saturated, nonlinear, magnetostatic field problems will provide a central focus for this paper.
13 Neuro-fuzzy predictors for the approximate prediction of v-i characteristic of electric arc furnaces
Sadeghian, A.R.; Lavers, J.D.

This paper presents application of feedforward neuro-fuzzy networks for single-step/multi-step prediction of the v-i characteristic of nonlinear, multi-variable, complex systems such as electric arc furnaces. The main objective is to investigate the capability of adaptive neuro-fuzzy networks to predict the v-i characteristics of electric arc furnaces. The novelties of this work are to propose the notion of approximate prediction and to it using a feedforward neuro-fuzzy suitable for long-term prediction. Successful implementations of feedforward neuro-fuzzy predictors are described and their performances are illustrated using the results obtained from adaptive neuro-fuzzy networks and recorded data.
14 Recurrent neuro-fuzzy predictors for multi-step prediction of v-i characteristics of electric arc furnaces
Sadeghian, A.R.; Lavers, J.D.

Presents an application of recurrent neuro-fuzzy systems to predict electric arc furnaces voltage and current. The primary objective is to investigate capability of adaptive fuzzy systems to predict the v-i characteristics of nonlinear, multivariable, complex systems such as electric furnaces. The novelties of this work are proposing a combination of recurrent neuro-fuzzy networks deemed suitable for prediction and using a wider window of observation whereby multi-step predictions can be made. In particular, the paper investigates the likelihood of long-term prediction for both furnace current and voltage. Successful implementations of recurrent neuro-fuzzy predictors are described and their performances are illustrated using the results obtained from adaptive neuro-fuzzy networks and recorded data.
15 Application of feedforward neuro-fuzzy networks for current prediction in electric arc furnaces
Sadeghian, A.R.; Lavers, J.D.

Presents the application of a class of hybrid neuro-fuzzy networks to the solution of a particular complex problem. The primary objectives are both to investigate the capability of adaptive neuro-fuzzy networks and to justify their application to predict the v-i characteristics of nonlinear, multi-variable, complex systems such as electric arc furnaces. The novelty of the work is proposing a feedforward neuro-fuzzy structure suitable for long-term prediction. Successful implementations of feedforward neuro-fuzzy predictors are described and their performances are illustrated using the results obtained from adaptive neuro-fuzzy networks and recorded data.

 

BACK