• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.33B no.12
• /
• pp.95-105
• /
• 1996

This paper presents structure optimization of a feedforward neural netowrk controller using the genetic algorithm. It is important to design the neural network with minimum structure for fast response and learning. To minimize the structure of the feedforward neural network, a genralization of multilayer neural netowrks, the genetic algorithm uses binary coding for the structure and floating-point coding for weights. Local search with an on-line learnign algorithm enhances the search performance and reduce the time for global search of the genetic algorithm. The relative fitness defined as the multiplication of the error and node functions prevents from premature convergence. The feedforward neural controller of smaller size outperformed conventional multilayer perceptron network controller.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.27 no.3
• /
• pp.343-348
• /
• 2003

In this study, optimum designs of the air-bearing surface (ABS) are achieved using the reduced basis concept which can effectively reduce the number of design variables without cutting down on the design space. Even though the optimization method is easier and more applicable to handle than the trial-and-error method, its efficiency is largely dependent on the number of the design variables. Hence, the reduced basis concept is applied, by which the desired design can be defined as a linear combination of basis designs. The simulation results show the effectiveness of the proposed approach by obtaining the optimum solutions of the sliders whose target flying heights are 25, 20, and 15nm.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.26 no.12
• /
• pp.2527-2534
• /
• 2002

An efficient method is proposed for the design of six-bar function generators with complex design tasks. Especially, the desired functions are defined for the entire motion ranges of the input variables. The design problem is defined as a nonlinear optimization problem. A concept of a weighted structural error is introduced for the definition of the objective function. Also simple branch identifiers are incorporated to eliminate the branch problems commonly encountered in a typical linkage synthesis problem. Two example problems of designing a Watt-II type double dwell mechanism and a Stephenson-III type double beat-up mechanism are demonstrated with numerical results. Constraints such as on the Grashof conditions and on the transmission angles are included for practical solutions.

• Proceedings of the KSME Conference
• /
• 2001.11a
• /
• pp.272-277
• /
• 2001

In order to evaluate variation of fatigue data of the LP steam turbine blade steel, it is important to estimate P - S - N curves to accurately define the probability distributions. In this study, new procedure is introduced to determine the expression of P - S - N curves. For this purpose, 3-parameter Weibull distribution was found to be most appropriate among assumed distributions when the probability distributions of the fatigue life were examined by the proposed analysis. Furthermore, parameter estimation for P - S - N curves was performed using various optimization to maximize the correlation coefficient. As a result of this, sequential linear programing method is used for estimation of P - S - N curves.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.31 no.9
• /
• pp.924-933
• /
• 2007

An impeller is a important part of turbo-machinery. It has a set of twisted surfaces because it consists of many blades. Five-axis machining is required to produce a impeller because of interference between tool and workpiece. It can obtain good surface integrity and high productivity. This paper proposes finish cutting method for machining impeller with 5-axis machining center and optimization of cutting condition by response surface method. Firstly, cutting methods are selected by consideration of operation characteristics. Secondly, response factors are determined as cutting time and cutting error for prediction of productivity. Experiments are projected by central composite design with axis point. Thirdly, regression linear models are estimated as single surface in the leading edge and as dual surface in the hub surface cutting. Finally, cutting conditions are optimized.

• Proceedings of the Korean Society of Precision Engineering Conference
• /
• 2000.11a
• /
• pp.683-686
• /
• 2000

A new approach to the task of aligning a robot using camera is presented in this paper. We apply an elitist GA to find the joints angles of manipulator to reach target position instead of using nonlinear least error method. Since it employs parallel search and have good performance in solving optimization problems. In order to improve convergence speed, the floating coding method and geometry constraint conditions are used. Experiments are carried out to exhibit the effectiveness of vision-based control using elitist genetic algorithm.

• Journal of Mechanical Science and Technology
• /
• v.16 no.5
• /
• pp.619-632
• /
• 2002

In recent study on design of experiments, the complicate metamodeling has been studied because defining exact model using computer simulation is expensive and time consuming. Thus, some designers often use approximate models, which express the relation between some inputs and outputs. In this paper, we review and compare the complicate metamodels, which are expressed by the interaction of various data through trying many physical experiments and running a computer simulation. The prediction model in this paper employs interpolation schemes known as ordinary kriging developed in the fields of spatial statistics and kriging in Design and Analysis of Computer Experiments (DACE) model. We will focus on describing the definitions, the prediction functions and the algorithms of two kriging methods, and assess the error measures of those by using some validation methods.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.24 no.5 s.176
• /
• pp.1166-1174
• /
• 2000

In this paper, we introduce a positioner based on magnetic levitation to eliminate the friction which is the most severe effect to limit high resolution on the micro level. Differently from existing electromagnetic device, the proposed positioner consists of air core solenoid and permanent magnet. Although the combination produces small magnetic force, it is suitable for realizing micro motion repeatedly without the accumulation of error because there is no hysteresis caused by ferromagnetic materials, no eddy current loss, no flux saturation. First, the approximate modeling of stiffness and damping effects between the magnetic elements is made and verified experimentally. Then, we have formulated the dynamic equation of one d.o.f magnetic levitation positioner using linear perturbation method and discussed the necessity of optimization for the chief design parameters to maximize the stability performance.

• 제어로봇시스템학회:학술대회논문집
• /
• 1999.10a
• /
• pp.129-132
• /
• 1999

This paper deals with the optimal filtering problem constrained to input noise signal corrupting the measurement output for linear discrete-time systems. The transfer matrix H$_2$and/or H$_{\infty}$ norms are used as criteria in an estimation error sense. In this paper, the mixed $H_2/H_{\infty}$ filtering Problem in lineal discrete-time systems is solved using the LMI approach, yielding a compromise between the H$_2$and H$_{\infty}$ filter designs. This filter design problems we formulated in a convex optimization framework using linear matrix inequalities. A numerical example is presented.

• Journal of Institute of Control, Robotics and Systems
• /
• v.4 no.5
• /
• pp.592-599
• /
• 1998

In this paper, we design a tracking controller which satisfies transient response specifications and maintains tracking error within a tolerable limit for the uncertain track-following system of an optical disk drive. To this end, a robust $H_{\infty}$ control problem with regional stability constraints and sinusoidal disturbance rejection is considered. The internal model principle is used for rejecting the sinusoidal disturbance caused by eccentric rotation of the disk. We show that a condition satisfying the regional stability constraints can be expressed in terms of a linear matrix inequality (LMI) using the Lyapunov theory and S-procedure. Finally, a tracking controller is obtained by solving an LMI optimization problem involving two linear matrix inequalities. The proposed controller design method is evaluated through an experiment.