Search results

Global optimization in electrical engineering using stochastic methods requires usually a large amount of CPU time to locate the optimum, if the objective function is calculated either with the finite element method (FEM) or the boundary element method (BEM). One approach to reduce the number of FEM or BEM calls using neural networks and another one using multiquadric functions have been introduced recently. This paper compares the efficiency of both methods, which are applied to a couple of test problems and the results are discussed.

Nowadays, there are strong movements towards development and usage of multimedia courseware as a means of knowledge transfer. Many authors of textbooks or lecture notes are now striving to redesign the supporting material for their major courses in a structured, highly efficient way, including interactive content and media. Thus, in order to avoid unnecessary work load resulting from updating and publishing various courseware versions, tools for improving document creation and conversion have been developed and are now being applied for the first time on a new “Electrodynamics”‐‐ courseware.

The aim of the work is to reconstruct the anisotropic complex conductivity distribution with the common Gauss‐Newton algorithm.

A cubic region with anisotropic material properties is enclosed by a larger cube with isotropic material properties. Numerical simulations are done with tetrahedral nodal finite elements of second‐order.

It can be shown that it is possible to reconstruct anisotropic complex conductivity distribution if the starting values are chosen sufficiently close to the true values of the complex conductivity.

In this paper, the anisotropic electric conductivity and the anisotropic permittivity are reconstructed in 3D.

The purpose of this paper is to extend a (μ/ρ, λ) evolution strategy to perform remarkably more globally and to detect as many solutions as possible close to the Pareto optimal front.

A C‐link cluster algorithm is used to group the parameter configurations of the current population into more or less independent clusters. Following this procedure, recombination (a classical operator of evolutionary strategies) is modified. Recombination within a cluster is performed with a higher probability than recombination of individuals coming from detached clusters.

It is shown that this new method ends up virtually always in the global solution of a multi‐modal test function. When applied to a real‐world application, several solutions very close to the front of Pareto optimal solutions are detected.

Stochastic optimization strategies need a very large number of function calls to exhibit their ability to reach very good local if not the global solution. Therefore, the application of such methods is still limited to problems where the forward solutions can be obtained with a reasonable computational effort.

The main improvement is the usage of approximate number of isolated clusters to dynamically update the size of the population in order to save computation time, to find the global solution with a higher probability and to use more than one objective function to cover a larger part of the Pareto optimal front.

This paper aims to show on a widely used benchmark problem that chaotic sequences can improve the search ability of evolution strategies (ES).

The Lozi map is used to generate new individuals in the framework of ES algorithms. A quasi‐Newton (QN) method is also used within the iterative loop to improve the solution's quality locally.

It is shown that the combined use of chaotic sequences and QN methods can provide high‐quality solutions with small standard deviation on the selected benchmark problem.

Although the benchmark is considered to be representative of typical electromagnetic problems, different test cases may give less satisfactory results.

The proposed approach appears to be an efficient general purpose optimizer for electromagnetic design problems.

This paper introduces the use of chaotic sequences in the area of electromagnetic design optimization.

Uncertainties in the design variables of non‐linear engineering optimization problems are often neglected. That could result in considerable deterioration of the target function value of an implemented design compared with the computed optimal solution. This effect can be reduced with robust optimization, where it is tried to achieve robust designs by actively embedding the uncertainties and robustness measures in the optimization process. A methodology for robust optimization of non‐linear problems is presented, including practical methods for the solution of such programs. The benefits of the approach are discussed in a numerical field calculation example.

Considers, without loss of generality, a simple linear problem, where in a certain domain the magnetic field, generated by infinitely long conductors, whose locations as well as the currents are unknown, has to meet a certain figure. The problem is solved by applying hierarchical simulated annealing, which iteratively reduces the dimension of the search space to save computational cost. A Gauss‐Newton scheme, making use of analytical Jacobians, preceding a sequential quadratic program (SQP), will be applied as a second approach to tackle this severely ill‐posed problem. The results of these two techniques will be analyzed and discussed and some comments on future work will be given.

The process of identifying unknown hidden objects by taking advantage of electromagnetic effects becomes more and more important. Clearing of mines or finding electrical conductors in concrete should be mentioned here. Magnetisation and eddy currents are the phenomena which are used in general. In this case, the layout and arrangement of the excitation coils and receiving coils influences the effectiveness and accuracy crucially. This design optimization process can be done by simulating the electromagnetic field with a 3D finite element method. Once a satisfying configuration has been found, the question arises, which quantities of the measured (and hence simulated) signals contain the most reliable information? Since the 3D finite element calculations are very time consuming, the inverse problem (detecting the ferrous object from some measured signals) is performed by approximating the corresponding electromagnetic signal by a neural network. Investigations on a ferrous conductive rod will be described in the paper.

The purpose of this paper is to solve generic magnetostatic problems by BEM, by studying how to use a boundary integral equation (BIE) with the double layer charge as unknown derived from the scalar potential.

Since the double layer charge produces only the potential gap without disturbing the normal magnetic flux density, the field is accurately formulated even by one BIE with one unknown. Once the double layer charge is determined, Biot‐Savart's law gives easily the magnetic flux density.

The BIE using double layer charge is capable of treating robustly geometrical singularities at edges and corners. It is also capable of solving the problems with extremely high magnetic permeability.

The proposed BIE contains only the double layer charge while the conventional equations derived from the scalar potential contain the single and double layer charges as unknowns. In the multiply connected problems, the excitation potential in the material is derived from the magnetomotive force to represent the circulating fields due to multiply connected exciting currents.

An effective approach to the optimal design of electromagnetic devices should take into account the effect of mechanical tolerances on the actual devices performance. A possible approach could be to match a Pareto optimality study with a Monte Carlo analysis by randomly varying the constructive parameters. In this paper it is shown how such an analysis can be used to allow an expert designer to select among different Pareto optimal designs.