Search

This paper addresses the problem of extracting sensitivity information from a modified adaptive-network-based fuzzy inference system (ANFIS) topology, i.e., from a Parallel Layer Perception network. This topology was recently proposed and presents computational advantages if compared with the traditional ANFIS. The indirect extraction of gradient information is useful in optimization problems when a high computational effort is involved in the evaluation of the functions, for instance when finite element analysis is used to solve the electromagnetic field problem. An analytical and an electromagnetic optimization problem are discussed. The results show the effectiveness, i.e., simplicity, accuracy and saving in Cpu time, of this novel topology for the extraction of sensitivity information. [ABSTRACT FROM AUTHOR]

This paper discusses the application of response surface methodology (RSM) to design optimization for brushless dc permanent magnet motors for a hard disk driver to reduce cogging torque. The feasibility of using RSM with the finite-element method in practical engineering problems is investigated with computational examples and comparison between the fitted response and the results obtained from an analytical solution. [ABSTRACT FROM AUTHOR]

In this paper, a novel two-dimensional ray tracing procedure is presented for the localization of electromagnetic (EM) sources in urban environment in the presence of multipaths, reflections, and diffractions. The EM source localization is achieved by solving an inverse problem in which input data are few field measurements taken at accessible points. In the procedure, the number and positions of EM field sources are obtained by minimizing a suitable cost function, evaluated by exploiting the ray tracing method. A numerical experiment is also presented in order to show the applicability of this procedure to realistic situations. [ABSTRACT FROM AUTHOR]

In reconstruction (i.e., determining the states of a model from measurements of model outputs), one is often forced to search for a regularized solution due to poor sensitivity of model outputs with respect to the model states. The amount of regularization is controlled by the regularization parameter, a scalar value multiplied with the so-called regularization term. The choice of the regularization parameter is crucial for the reconstruction process. In this paper, a new method to estimate the regularization parameter in an adaptive way is proposed. A condition-number based estimate of the regularization parameter for the first iteration step is required to choose the weighting factor for adapting the regularization parameter iteratively. By virtue of controlling the regularization term, a kind of edge preservation can be achieved. The validity of this method will be demonstrated for a capacitance tomography problem, which is solved applying a Gauss-Newton scheme. [ABSTRACT FROM AUTHOR]

Models of electromagnetic devices using reluctance networks are very interesting, especially for sizing and optimization. Unfortunately, too often, designers do not use them because they are time consuming to implement, mainly for complicated aspects like the need to write and solve the network equations. We have developed a tool that automates all those symbolic and numeric tasks. We also generate software components that allow to use those models very efficiently in an optimization process by providing the formal right sensitivity. We present the methodology and software, as well as their application on examples. This paper proposes also an analysis of the global design and modeling process with a reluctance network that defines the relative place of the designer versus methodologies and tools. [ABSTRACT FROM AUTHOR]

The optimal design of a magnetic recording head is one important issue for high magnetic recording of 600-Gb/in² and beyond. In this paper, a three-dimensional topology optimization method (ON/OFF method) combined with a finite-element method is applied for the optimization of a cusp field single-pole-type head. In the ON/OFF method, the unknown variable is the reluctivity of magnetic material in the design domain and there is no special constraint. Therefore, the possibility of finding more optimal topology is increased compared with the conventional optimization method in which the design variables are specified. The optimal shape and dimension of a head is determined so that the flux density in a recording bit becomes the maximum and the leakage flux in the adjacent bit and adjacent track becomes the minimum. The recording field of 1.75 T and the field gradient of 0.038 T/nm (380 Oe/nm) are obtained for the head-soft underlayer distance of 25 nm. [ABSTRACT FROM AUTHOR]

This paper describes a fitting algorithm suitable for simultaneously approximating the real and imaginary parts of transformer admittance curves. The algorithm follows a unique strategy to determine the best initial guess. It optimizes the parameters one group at a time. This technique allows the fitting routine to find the best solution even when the number of optimization parameters is large. Since the optimization of each group of parameters is well controlled at each stage, the algorithm is suitable for constrained optimization. Another advantage is that the starting point for each stage can be very simple. Examples demonstrate that the algorithm produces good results for a variety of transformer admittance curves. [ABSTRACT FROM AUTHOR]

This paper describes work done on optimization of design of permanent magnet motors using response surface modeling and genetic algorithms (GAs). Finite element computations have been used for numerical experiments on geometrical design variables in order to evaluate the coefficients of a second-order model for the response surfaces representing machine parameters (λm, Ld, and Lq). GAS were used to optimize the torque and speed of the machine in terms of these variables. [ABSTRACT FROM AUTHOR]