• Journal of Korean Ophthalmic Optics Society
• /
• v.1 no.1
• /
• pp.29-39
• /
• 1996

We studied the method which would determine the appropriate additive damping factor for the damped least sequres(DLS) optimization. We calculated eigenvalues of the product of the Jacobian matrix of error function by using the singular value decomposition(SVD) method. While suitable damping factor was appiled to the additive DLS by using SVD and Gaussian elimination method, the convergence and stability of the optimization process were examined in a triplet-type camera lens-system where the condition number is well conditioned. We compared the convergence and stability of merit function when median, maximum and minimum of eigenvalues were used as a damping factor in the optimization process. When damping factor is median of eigenvalue, the convergence and stability of merit function are more excellent than in the case of two other eigenvalues. Thus, we adopt the median of eigenvalues as an appropriate damping factor. Next, by using SVD and Gaussian elimination method, we compound the convergence and stability of optimization process for triplet-type camera lens-system design. In these two method; triplet-type camera lens-system in which condition number is well conditioned, has little improvement with the combination of DLS and SVD.

• ETRI Journal
• /
• v.28 no.1
• /
• pp.59-66
• /
• 2006

Learning behaviors of a radial basis function network (RBFN) using a singular value decomposition (SVD) and stochastic gradient (SG) algorithm, together named RBF-SVD-SG, for odor sensing systems are analyzed, and a fast training method is proposed. RBF input data is from a conducting polymer sensor array. It is revealed in this paper that the SG algorithm for the fine-tuning of centers and widths still shows ill-behaving learning results when a sufficiently small convergence coefficient is not used. Since the tuning of centers in RBFN plays a dominant role in the performance of RBFN odor sensing systems, our analysis is focused on the center-gradient variance of the RBFN-SVD-SG algorithm. We found analytically that the steadystate weight fluctuation and large values of a convergence coefficient can lead to an increase in variance of the center-gradient estimate. Based on this analysis, we propose to use the least mean square algorithm instead of SVD in adjusting the weight for stable steady-state weight behavior. Experimental results of the proposed algorithm have shown faster learning speed and better classification performance.

• Journal of Welding and Joining
• /
• v.29 no.2
• /
• pp.64-71
• /
• 2011

In this paper, crack propagation analyses in the inner diameter (ID) repair weld of the dissimilar metal weldment of a nozzle were performed using a finite element alternating method (FEAM). To calculate the theoretical solution for the crack tip stress intensity factor, a weak type singular integral equation consisted of crack surface traction and dislocation density function was constructed and solved in conjunction with the FEAM. A two-dimensional axisymmetric finite element nozzle model was prepared and ID repair welding was simulated. An initial crack, 10% depth of weld thickness, was assumed and crack propagation trajectory from the initial crack to the 75% depth of thickness was calculated using the FEAM. Crack growth versus time curve was also calculated and compared with the curves obtained from ASME code method. With the method constructed in this paper, crack propagation trajectory and crack growth time were calculated automatically and effectively.

• The Journal of Korea Robotics Society
• /
• v.6 no.3
• /
• pp.237-246
• /
• 2011

Robots have been used in many fields due to its performance improvement and variety of its functionality, to the extent which robots can replace human tasks. Individual feature and better performance of robots are expected and required to be created. As their performances and functions have increased, systems have gotten more complicated. Multi mobile robots can perform complex tasks with simple robot system and algorithm. But multi mobile robots face much more complex driving problem than singular driving. To solve the problem, in this study, driving algorithm based on the energy method is applied to the individual robot in a group. This makes a cluster be in a formation automatically and suggests a cluster the automatic driving method so that they stably arrive at the target. The energy method mentioned above is applying attractive force and repulsive force to a special target, other robots or obstacles. This creates the potential energy, and the robot is controlled to drive in the direction of decreasing energy, which basically satisfies lyapunov function. Through this method, a cluster robot is able to create a formation and stably arrives at its target.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.30 no.8 s.251
• /
• pp.949-956
• /
• 2006

The method of Greenwood and Williamson is extended to obtain a solution to the coupled non-linear problem of steady-state electrical and thermal conduction across a crack in a conductive layer, for which the electrical resistivity and thermal conductivity are functions of temperature. The problem can be decomposed into the solution of a pair of non-linear algebraic equations involving boundary values and material properties. The new mixed-boundary value problem given from the thermal and electrical boundary conditions for the crack in the conductive layer is reduced in order to solve a singular integral equation of the first kind, the solution of which can be expressed in terms of the product of a series of the Chebyshev polynomials and their weight function. The non-existence of the solution for an infinite conductor in electrical and thermal conduction is shown. Numerical results are given showing the temperature field around the crack.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.2 no.4
• /
• pp.184-193
• /
• 2008

We examine the performance of genetic algorithms using binary encoding, with respect to a change of basis. Changing the basis can result in a change in the linkage structure inherent in the fitness function. We test three simple functions with differing linkage strengths and analyze the results. Based on an empirical analysis, we show that a better basis results in a smoother fitness landscape, hence genetic algorithms based on the new encoding method provide better performance.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.25 no.3B
• /
• pp.570-576
• /
• 2000

This paper proposed LQ-Servo PI controller by considering LQ-Servo structure as PI controller with a partial state feedback and concerns about the development of the flexible design algorithm by introducing weights to the design parameters of the previous LQ-Servo design method. the propose algorithm improves the matchings of the maximum and minimum singular values at high and low frequencies of the design loop transfer function as well as its loop shaping for performance.

• Structural Engineering and Mechanics
• /
• v.87 no.1
• /
• pp.95-106
• /
• 2023

This paper investigates the dynamic response of a functionally graded material (FGM) coated half-plane excited by distributed time harmonic loading. Three types of typical distributed surface loads, including uniform load, Hertz load, and square-root singular load, are considered. The mass density and elastic modulus of the FGM coating are supposed to be described by the exponential function. The material damping is modelled by a linearly hysteretic damping which is expressed by a complex modulus in the time harmonic motion. Using Fourier integral transform technique and numerical integral method, the effects of the excitation frequency, gradient index, damping, and load type on the dynamic stresses and displacements are discussed.

• Proceedings of the Korean Society for Noise and Vibration Engineering Conference
• /
• 2007.11a
• /
• pp.1018-1021
• /
• 2007

This paper explores the use of BEM based NAH to reconstruct the surface vibration of a plate in a rectangular finite cavity, in which the distances between sensors and the nearest points on the source surface are not equal. In such circumstances, different degree of information on propagating and non-propagating wave components will be detected by sensors at different positions, as well as the influence of measurement noise will vary significantly from the nearest points of measurement to the farthest ones. On the other hand, the condition number of the vibro-acoustic transfer function matrix relating normal surface velocities and field pressures will becomes high, numerically indicating an increase of linear dependency between rows of transfer function matrix. The combination of poor measurement and high condition number will result inaccurate reconstruction. Therefore, one approach to be investigated in this work is to select the measurement positions in such ways that reduce measurement redundancy, as it indicated by the condition number. The improvement is found to be significant in the numerical simulations utilizing two different criterions, spanning from over-determined to under-determined cases, and in the validation experiment.

• Smart Structures and Systems
• /
• v.4 no.5
• /
• pp.629-640
• /
• 2008

A new algorithm is proposed to determine optimal accelerometer locations (OAL) when a structure is identified by frequency domain system identification (SI) method. As a result, a guideline is presented for selecting OAL which can reflect modal response of a structure properly. The guideline is to provide a minimum number of necessary accelerometers with the variation in the number of measurable target modes. To determine OAL for SI applications effectively, the modal sensitivity effective independence distribution vector (MS-EIDV) is developed with the likelihood function of measurements. By maximizing the likelihood of the occurrence of the measurements relative to the predictions, Fisher Information Matrix (FIM) is derived as a function of mode shape sensitivity. This paper also proposes a statistical approach in determining the structural parameters with a presumed parameter error which reflects the epistemic paradox between the determination of OAL and the application of a SI scheme. Numerical simulations have been carried out to examine the proposed OAL algorithm. A two-span multi-girder bridge and a two-span truss bridge were used for the simulation studies. To overcome a rank deficiency frequently occurred in inverting a FIM, the singular value decomposition scheme has been applied.