N. Salerno

A self-adaptive niching genetic algorithm for multimodal optimization of electromagnetic devices

A new niching genetic algorithm (NGA) called self-adaptive NGA (SANGA) is proposed. The main innovation is that the niche radius is estimated as an additional variable of the optimization problem and is not assigned a priori, as is usually done in other standard NGAs. SANGA, coupled with the deterministic pattern search method, forms a hybrid optimization method which works well in the optimization of multimodal functions and in the design of electromagnetic devices

Iteratively‐improved Robin boundary conditions for the finite element solution of scattering problems in unbounded domains

An iterative procedure is described for the finite-element solution of scalar scattering problems in unbounded domains. The scattering objects may have multiple connectivity, may be of different materials or with different boundary conditions. A fictitious boundary enclosing all the objects involved is introduced. An appropriate Robin (mixed) condition is initially guessed on this boundary and is iteratively improved making use of Greens formula. It will be seen that the best choice for the Robin boundary condition is an absorbing-like one. A theorem about the theoretical convergence of the procedure is demonstrated. An analytical study of the special case of a circular cylindrical scatterer is made. Comparisons are made with other methods. Some numerical examples are provided in order to illustrate and validate the procedure and to show its applicability whatever the frequency of the incident wave. Although particular emphasis is laid in the paper on electromagnetic problems, the procedure is fully applicable to other kinds of physical phenomena such as acoustic ones.

Finite element iterative solution of skin effect problems in open boundaries

The paper proposes an iterative procedure, called current iteration, for the finite element solution of two-dimensional steady-state skin effect problems in open boundaries. In the procedure a fictitious boundary is introduced enclosing all the conductors. On it, the magnetic vector potential is first guessed and then iteratively updated according to the current density computed in the conductors. Conditions are obtained implying convergence to the exact solution of the unbounded problem whatever the initial guess. The choice of the fictitious boundary and the selection of the relaxation parameter in such a way that computational efficiency is obtained are discussed. The greatest advantage of the procedure is its ease of implementation in a pre-existing finite element code for bounded problems. An axisymmetric version of the procedure is also described since implementation only involves minor changes as compared with the 2-D one. Examples are provided in order to clarify and validate the procedure and compare it with other techniques.

An improved solution scheme for open-boundary skin effect problems

FEM/DBCI (Dirichlet Boundary Condition Iteration) is a hybrid method which has been successfully applied to the finite element solution of static and quasistatic unbounded field problems. It is based on the iterative improvement of a Dirichlet condition on a fictitious boundary enclosing all the conductors, making use of the free-space Greens function. In this paper it is shown that for two-dimensional skin effect problems a more robust solving strategy can be pursued by means of GMRES, obtaining a reduction in both CPU and storage requirements.

A theoretical study of charge iteration

Charge iteration is an iterative procedure for the finite element computation of unbounded electrical fields, created by voltaged conductors. It makes use of a fictitious boundary, enclosing all the conductors, on which the electrical potential is first guessed and then iteratively improved according to the charge lying on the conductor surfaces. Highlights the theoretical foundations of the procedure outside any numerical context. From this useful insight, obtains a model which can aid the user in utilization of the numerical version of the procedure.

An Iterative Solution to FEM-BEM Algebraic Systems for Open-Boundary Electrostatic Problems

This paper shows that the global system of equations in the hybrid FEM-BEM solution of open-boundary electrostatic problems can be efficiently solved in an iterative way by making use of the conjugate gradient solver for the FEM equations and the direct LU solver for the BEM equations. To do so, it is convenient to implement the BEM equations in a non-conventional way, by making the nodes for the potential non coinciding with the nodes for its normal derivatives

A proposal for a universal parameter configuration for genetic algorithm optimization of electromagnetic devices

Genetic algorithms (GAs) are widely used in the optimization of electromagnetic devices. However, if finite element analyses are needed for the evaluation of the objective function, GAs require a long computation time. They also need a difficult tuning session before starting optimization, to select the best configuration of the parameters that control the algorithm. This paper proposes a universal choice of the GA parameters in order to spare the designer the tuning session and to reduce the computing time for the whole procedure.

An iterative solution to scattering from cavity-backed apertures in a perfectly conducting wedge

In this paper it is shown how the Robin iteration procedure, already proposed by the authors for the solution of electromagnetic scattering problems, can be easily adapted to scattering from cavities embedded in a perfectly-conducting wedge. The procedure couples a differential equation for the interior problem with an integral equation for the exterior one. A suitable choice of the Robin (mixed) boundary condition on the fictitious boundary dividing the interior and exterior domains avoids resonances.

Axisymmetric unbounded electrical field computation by charge iteration

An iterative procedure is presented for the finite-element computation of axisymmetric electrostatic fields in unbounded domains. In the procedure the original unbounded problem is solved by using successive evaluations of the potential on a fictitious boundary enclosing all the conductors of the problem, according to the charge lying on their surface. For simplex Lagrangian elements universal matrices are used for a fast and efficient computation of this charge. The main advantage of this procedure lies in its ease of implementation in a standard bounded problem finite-element code and in the accuracy of the computed results. >

A Parallel Version of the Self-Adaptive Low-High Evaluation Evolutionary-Algorithm for Electromagnetic Device Optimization

The self-adaptive low-high evaluation evolutionary-algorithm (SALHE-EA) is used to solve multimodal optimization problems. SALHE-EA is able to find the multiple optima of a single objective function (OF) and to give an idea of the fitness landscape in the neighborhood of these optima. This aspect is of crucial importance when the single OF is obtained using the weighted sum of the OFs, each related to a different goal of the optimization problem. This paper presents an improved version of SALHE-EA. This new version has several new features and, mainly, the suitability for parallelization.