• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.67 no.10
• /
• pp.1292-1297
• /
• 2018

In this paper, efficiency optimization design of 600W class IPMSM was performed by using Quasi-Newton method. The output was limited to 600W to meet the same output as the basic model. The behavior of each variable as the design progressed was judged on the efficiency, which is the target value through correlation analysis. The design variables were set as the width of the stator, the position of the permanent magnet from the end of the rotor, the thickness of the permanent magnet, and the width of the permanent magnet.

• Journal of applied mathematics & informatics
• /
• v.8 no.3
• /
• pp.743-753
• /
• 2001

We provide local convergence results in affine form for inexact Newton-like as well as quasi-Newton iterative methods in a Banach space setting. We use hypotheses on the second or on the first and mth Frechet-derivative (m≥2 an integer) of the operator involved. Our results allow a wider choice of starting points since our radius of convergence can be larger than the corresponding one given in earlier results using hypotheses on the first-Frechet-derivative only. A numerical example is provided to illustrate this fact. Our results apply when the method is, for example, a difference Newton-like or update-Newton method. Furthermore, our results have direct applications to the solution of autonomous differential equations.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 1998.06a
• /
• pp.118-121
• /
• 1998

In this paper, we propose another paradigm(QNBP) to be capable of overcoming Limitations of the traditional backpropagation(SDBP). QNBPis based on the method of Quasi -Newton(variable metric) with the nomalized direction vectors and computes step size through the linear search. Simulation results showed that QNBP was definitely superior to both the stochasitc SDBP and the deterministic SDBP in terms of accuracy and rate of convergence and might sumount the problem of local minima. and there was no different between DFP+SR1 and BFGS+SR1 combined algrothms in QNBP.

• Journal of the Korean Society for Industrial and Applied Mathematics
• /
• v.3 no.1
• /
• pp.71-79
• /
• 1999

We consider multi-step quasi-Newton methods for unconstrained optimization. These methods were introduced by Ford and Moghrabi [1, 2], who showed how interpolating curves could be used to derive a generalization of the Secant Equation (the relation normally employed in the construction of quasi-Newton methods). One of the most successful of these multi-step methods makes use of the current approximation to the Hessian to determine the parameterization of the interpolating curve in the variable-space and, hence, the generalized updating formula. In this paper, we investigate new parameterization techniques to the approximate Hessian, in an attempt to determine a better Hessian approximation at each iteration and, thus, improve the numerical performance of such algorithms.

• Journal of Electrical Engineering and Technology
• /
• v.5 no.4
• /
• pp.666-671
• /
• 2010

Multi-layer perceptron neural networks have been constructed to classify four types of defects in steam generator tubes. Three features are extracted from the signals of the eddy current testing method. These include maximum impedance, phase angle at the point of maximum impedance, and an angle between the point of maximum impedance and the point of half the maximum impedance. Two hundred sets of these features are used for training and assessing the networks. Two approaches are involved to train the networks and to classify the defect type. One is the conjugate gradient method and the other is the Broydon-Fletcher-Goldfarb-Shanno method which is recognized as the most popular algorithm of quasi-Newton methods. It is found from the computation results that the training time of the Broydon-Fletcher-Goldfarb-Shanno method is much faster than that of the conjugate gradient method in most cases. On the other hand, no significant difference of the classification performance between the two methods is observed.

• Journal of the Microelectronics and Packaging Society
• /
• v.10 no.4
• /
• pp.15-20
• /
• 2003

A method that predicts the flying state of the head slider in an optical disk drive (ODD) or a hard disk drive(HDD) was investigated. The dual solver based on the Newton method and the quasi-Newton method have been developed to simulate the steady-state flying conditions. The numerical results show the effectiveness and reliability of this new solver.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.37 no.4A
• /
• pp.250-259
• /
• 2012

This paper compared and investigated the choice of optimal algorithm affects on the directivity synthesis of linear array in the satisfaction to the design specification of the desired directivity, convergence characteristic, and adaptability. Optimal algorithms use a quasi-Newton method(DFP and BFGS method) for realizing the desired directivity, used a quasi-ideal beam, steering beam, and a multi-beam, chosen as desired directivity. In the numerical result, this paper verified the effectiveness of the quasi-Newton method to the directivity synthesis, and offered a solving approach of occurred problems in the numerical simulation process.

• Journal of the Korean Society for Industrial and Applied Mathematics
• /
• v.15 no.4
• /
• pp.267-275
• /
• 2011

This paper proposes a new quickly convergent pattern search quasi-Newton algorithm that employs the multi-step version of the Symmetric Rank One (SRI). The new algorithm works on the factorizations of the inverse Hessian approximations to make available a sequence of convergent positive bases required by the pattern search process. The algorithm, in principle, resembles that developed in [1] with multi-step methods dominating the dervation and with numerical improvements incurred, as shown by the numerical results presented herein.

• Journal of the Korean Society for Industrial and Applied Mathematics
• /
• v.6 no.1
• /
• pp.91-107
• /
• 2002

This paper presents two new self-scaling variable-metric algorithms. The first is based on a known two-parameter family of rank-two updating formulae, the second employs an initial scaling of the estimated inverse Hessian which modifies the first self-scaling algorithm. The algorithms are compared with similar published algorithms, notably those due to Oren, Shanno and Phua, Biggs and with BFGS (the best known quasi-Newton method). The best of these new and published algorithms are also modified to employ inexact line searches with marginal effect. The new algorithms are superior, especially as the problem dimension increases.

• Journal of applied mathematics & informatics
• /
• v.30 no.5_6
• /
• pp.971-980
• /
• 2012

In this paper, a method associating with one new form of nonmonotone linesearch technique is proposed, which can be regarded as a generalization of the Perry-Shanno memoryless quasi-Newton type method. Under some reasonable conditions, the global convergence of the proposed method is proven. Numerical tests show its efficiency.