Search results

The purpose of this paper is to present an effective optimization strategy applied in a physical structure optimization of a semiconductor power metal oxide semiconductor field‐effect transistor (MOSFET), with an expensive integration constraint computation.

In order to deal with inaccuracy due to inevitable numerical errors in the objective function calculation (the power losses of the power MOSFET) and in the constraint computation, the paper proposes to use the progressive quadratic response surface method (PQRSM).

The paper focuses on four aspects: the inevitable numerical errors in the power loss and the integration constraint computation; the response surface approximation (RSA) method; the PQRSM principle; and finally the comparisons of several optimization methods applied on this application problem.

An original optimization method, PQRSM, is proposed for reducing the oscillation problem of a semi‐analytical model. The optimization results of PQRSM have been compared with the evolution strategy (ES) algorithm, with similar results but faster computation.

This paper compares three different radial basis function neural networks, as well as the diffuse element method, according to their ability of approximation. This is very useful for the optimization of electromagnetic devices. Tests are done on several analytical functions and on the TEAM workshop problem 25.

The aim of this paper is to accelerate the convergence of iterative methods on ill‐conditioned linear systems of equations.

First a brief numerical analysis is given of left preconditioners on ill‐conditioned linear systems of equations. From this result, it is deduced that a double preconditioning approach may be better. Then, a double preconditioner based on an iterative diagonal scaling method and an incomplete factorization method is proposed. The efficiency of this approach is illustrated on two finite element models produced by computational electromagnetism.

The double preconditioning approach is efficient for 2D and 3D finite element problems. The bi‐conjugate gradient algorithm always converges when it is double preconditioned. This is not the case when a simple incomplete factorization method is applied. Furthermore, when the two preconditioning techniques lead to the convergence of the iterative solving method, the double preconditioner significantly reduces the number of iterations in comparison with the simple preconditioner. On the proposed 2D problem, the speed‐up is between 6 and 32. On the proposed 3D problem, the speed‐up is between 13 and 20. Finally, the approach seems to reduce the growth of the condition number when higher‐order finite elements are used.

The paper proposes a particular double preconditioning approach which can be applied to any invertible linear system of equations. A numerical evaluation on a singular linear system is also provided but no proof or analysis of stability is given for this case.

The paper presents a new preconditioning technique based on the combination of two very simple and elementary methods: a diagonal scaling method and an incomplete factorization process. Acceleration obtained from this approach is quite impressive.

This paper deals with the problem of magnetization identification. We consider a ferromagnetic body placed in an inductor field. The goal of this work is, from static magnetic field measurements taken around the device, to obtain an accurate model of its magnetization. This inverse problem is usually ill‐posed and its solution is non‐unique. It is then necessary to use mathematical regularization. However, we prefer to transform it to a better posed one by incorporating our physical knowledge of the problem. Our approach is tested on the magnetization's identification of a real ferromagnetic sheet.

This paper presents a method to improve inverse problem resolution. This method focuses on the measurement set and particularly on sensor position. Based on experiment, it aims at finding sensor position criteria to insure the least bad inverse problem solving.

The studied device is a magnetized steel sheet measured by four sensors. Three optimization techniques are compared: condition number, solid angle and signature optimization.

An efficient criterion to compare the inverse problem resolution quality is presented. The comparison of optimization techniques shows that only signature optimization gives accurate results.

A relative simple case is studied in this paper: only four sensors are used to measure a steel sheet. Moreover magnetostatic low‐field case is supposed. Nevertheless techniques presented could be applied to more complex studies. Condition number and solid angle optimizations techniques should be tested with more sensors to confirm or infirm their inefficiency.

This paper presents the first step of a larger study concerning ships for naval application. The aim is to predict magnetic anomaly created by ship to compensate it. This anomaly could be computed through the resolution of an inverse problem based on internal measurements. The signature optimization technique could be used to find the optimal sensor location onboard.

Traditional regularization techniques are focusing on adding mathematical or physical information to the system in order to improve it. This paper provides another approach to improve inverse problem resolution through measurement set. It shows that sensor position optimization should be efficient.

The purpose of this paper is to make easily accessible models to test and compare the optimization algorithms we develop.

For this, the paper proposes an optimization framework based on software component, web service, and plugin to exploit these models in different environments.

The paper illustrates the discussion with optimizations in Matlab™ and R (www.r-project.org) of a transformer described and exploitable from the internet.

The originality is to make easy implementation of simulation model and optimization algorithm coupling using software component, web service, and plugin.