• 대한전기학회:학술대회논문집
• /
• 대한전기학회 2011년도 제42회 하계학술대회
• /
• pp.1966-1967
• /
• 2011

본 연구에서는 방사형 기저 함수를 이용한 다항식 신경회로망(Polynomial Neural Network) 분류기를 제안한다. 제안된 모델은 PNN을 기본 구조로 하여 1층의 다항식 노드 대신에 다중 출력 형태의 방사형 기저 함수를 사용하여 각 노드가 방사형 기저 함수 신경회로망(RBFNN)을 형성한다. RBFNN의 은닉층에는 fuzzy 클러스터링을 사용하여 입력 데이터의 특성을 고려한 적합도를 사용하였다. 제안된 분류기는 입력변수의 수와 다항식 차수가 모델의 성능을 결정함으로 최적화가 필요하며 본 논문에서는 Multiobjective Particle Swarm Optimization(MoPSO)을 사용하여 모델의 성능뿐만 아니라 모델의 복잡성 및 해석력을 고려하였다. 패턴 분류기로써의 제안된 모델을 평가하기 위해 Iris 데이터를 이용하였다.

• 대한기계학회논문집A
• /
• 제21권5호
• /
• pp.819-827
• /
• 1997

In this paper, a new modeling of a fine actuator for an optical pick-up has been proposed and multiobjective optimization of the actuator has been performed. The fine actuator is constituted of the bobbin which is supported by wire suspension, the coils which wind around the bobbin, and the magnets which cause the magnetic flux. If current flows in the coils, magnetic force is so produced as to be balanced with spring force of wire, so the bobbin is pisitioned. In this model the transfer function from input voltage to output displacementof bobbin has been obtained so that we can describe this integrated system with electromagnetic and mechanical parts. Wire suspension is regarded as a continuous Euler beam, damper as distributed viscous damping, and bobbin as a rigid body which can move up- and down- ward motion only. According to the model, the high frequency dynamic characteristics of the fine actuator can be known and the effect of damping can be investigated while the conventional second order model cannot. In multiobjective optimization, two objective functions have been chosen to maximize the fundamental frequency and the sensitivity with respect to the input voltage of the actuator so that Pareto's optimal solutions have been obtained using .epsilon.-constraint method. These objective functions will satisfy the trends which will enhance the access speed and reduce the tracking error in the optical pick-up technology of next generation. In the result of optimization, we obtain the designs of the optical pick-up fine actuator which has high speed, high sensitivity and low resonant peak. Furthermore, we offer the relation between two object functions so that the designer can make easy choice.

• International Journal of Aeronautical and Space Sciences
• /
• 제11권4호
• /
• pp.247-265
• /
• 2010

Multi-objective design exploration (MODE) and its applications are reviewed as an attempt to utilize numerical simulation in aerospace engineering design. MODE reveals the structure of the design space based on trade-off information. A self-organizing map (SOM) is incorporated into MODE as a visual data mining tool for the design space. SOM divides the design space into clusters with specific design features. This article reviews existing visual data mining techniques applied to engineering problems. Then, we discuss three applications of MODE: multidisciplinary design optimization for a regional-jet wing, silent supersonic technology demonstrator and centrifugal diffusers.

• 한국전산구조공학회논문집
• /
• 제33권6호
• /
• pp.419-426
• /
• 2020

본 연구에서는 철골편심가새골조 시스템을 대상으로 다목적최적화기법을 통해 설계를 수행하고 그 결과를 분석하였다. 최적화 설계를 위해 유전 알고리즘의 일종인 NSGA-II를 활용하였다. 여기서, 목적함수는 이율배반적 관계를 갖는 구조물량과 층간변위로 하여 최소화되고, 제약조건에는 구조기준에서 요구하는 내력비, 링크의 회전각 등을 포함하였다. 제약조건은 최적화 알고리즘 내에서 각 항목을 위반할수록 목적함수 값을 크게 증가시키는 벌금함수의 형태를 가지고 있다. 설계기준에서 EBF 시스템의 설계규정은 링크 부재만 항복이 허용되며 나머지 부재는 링크 항복 시 발생되는 부재력을 탄성상태에서 견디도록 의도한 역량설계법에 기초한다. 그러나 최적화를 통해 도출된 결과 중 일부는 구조기준의 설계조항은 만족하지만 특정층 링크에 소성변형이 집중되어 연약층을 형성함으로써 기준에서 의도하는 역량설계의 원칙을 위배하는 결과가 나타났다. 이를 해결하기 위해 모든 링크의 전단 초과강도계수 중 최대값이 최소값의 1.25배를 넘지 않도록 하는 제약식을 추가하였다. 새로운 제약식을 추가한 경우 모든 최적해는 설계기준과 역량설계의 원칙을 준수하는 것으로 나타났다. 모든 설계안에서 보 경간에 대한 링크의 길이비는 전단링크의 범주에 해당하는 10% ~ 14%였다. 전체적으로 설계안들은 링크의 초과강도 계수비가 가장 지배적인 제약으로 작용하였으며, 구조기준의 요구사항 중 층간변위와 내력비 등의 항목에서 허용치에 비해 매우 보수적으로 설계되었다.

• Wind and Structures
• /
• 제35권3호
• /
• pp.157-175
• /
• 2022

This paper highlights the minimization of drag and lift coefficient of different types both side setback tall buildings by the multi-objective optimization technique. The present study employed 48 number both-side setback models for simulation purposes. This study adopted three variables to find the two objective functions. Setback height and setback distances from the top of building models are considered variables. The setback distances are considered between 10-40% and setback heights are within 6-72% from the top of the models. Another variable is wind angles, which are considered from 0° to 90° at 15° intervals according to the symmetry of the building models. Drag and lift coefficients according to the different wind angles are employed as the objective functions. Therefore 336 number population data are used for each objective function. Optimum models are compared with computational simulation and found good agreements of drag and lift coefficient. The design wind angle variation of the optimum models is considered for drag and lift study on the main square model. The drag and lift data of the square model are compared with the optimum models and found the optimized models are minimizing the 45-65% drag and 25-60% lift compared to the initial square model.

• Wind and Structures
• /
• 제36권2호
• /
• pp.105-120
• /
• 2023

The present study is performed to find the effect of corner recession on a square plan-shaped tall building. A series of numerical simulations have been carried out to find the two orthogonal wind force coefficients on various model configurations using Computational Fluid Dynamics (CFD). Numerical analyses are performed by using ANSYS-CFX (k-ℇ turbulence model) considering the length scale of 1:300. The study is performed for 0° to 360° wind angle of attack. The CFD data thus generated is utilised to fit parametric equations to predict alongwind and crosswind force coefficients, C_fx and C_fy. The precision of the parametric equations is validated by employing a wind tunnel study for the 40% corner recession model, and an excellent match is observed. Upon satisfactory validation, the parametric equations are further used to carry out multiobjective optimization considering two orthogonal force coefficients. Pareto optimal design results are presented to propose suitable percentages of corner recession for the study building. The optimization is based on reducing the alongwind and crosswind forces simultaneously to enhance the aerodynamic performance of the building.

• 대한기계학회논문집A
• /
• 제27권1호
• /
• pp.144-151
• /
• 2003

In this paper, an optimum design problem for endosseous implant in dentistry is studied to find best implant design. An optimum design problem is formulated to reduce stresses arising at the cortical as well as cancellous bones, in which sufficient design parameters are chosen fur design definition that encompasses major implants in popular use. Optimization at once (OAO) with the large number of design variables, however, causes too costly solution or even failure to converge. A concept of multilevel optimization (MLO) is employed to this end, which is to group the design variables of similar nature, solve the sub-problem of smaller size fur each group in sequence, and this is iterated until convergence. Each sub-problem is solved based on the response surface method (RSM) due to its efficiency for small sized problem. Favorable solution is obtained by the MLO, which is compared to both solutions made by RSM and sequential quadratic programming (SQP) in the OAO problem.

• 한국자기학회지
• /
• 제7권2호
• /
• pp.103-108
• /
• 1997

본 논문에서는 퍼지결정법을 적용한 유도전동기의 최적설계 방법을 제시하였다. 이 방법은 설계자의 경험, 관점, 판단을 반영할 수 있을 뿐만 아니라 다목적 최적설계에 쉽게 적용가능하다. 특성 해석방법은 등가 자기회로법이며, 설계방법은 기존 설계법 중의 하나인 D$^{2}$L 법에 퍼지 결정법과 최적화 루틴을 결합하였다. 사용한 최적화 알고리즘은 확률론적 최적화기법인 (1+1) Evolution Strategy(ES)를 이용하였다. 제안된 알고리즘은 유도전동기의 무게최소화와 동시에 주요 동작점에서의 효율, 역률을 최대화 설계하는 다중목적 최적설계에 적용되었다.

• 한국해양공학회지
• /
• 제11권4호
• /
• pp.227-238
• /
• 1997

다목적함수 최적화를 효과적으로 수행하기 위하여 유전자 알고리즘과 직접탐색법을 결합하여 혼성형 최적화기법을 구현하였다. 이 방법은 유전자 알고리즘을 사용하여 최적점이 존재할 가능성이 높은 영역을 탐색한 후, 이 영역에서 직접탐색법을 사용하여 최종해를 찾는다. 따라서 탐색의 효율을 향상시키고 계산시간을 절약할 수 있는 장점이 있다. 그러나 최적화기법이 효율적이지만, 최적화기법을 사용하기 위해서는 전문가의 전문지식이 필요하다. 따라서 실제 최적화를 수행하기 위해서는 관련 분야의 전문지식과 최적화기법이 효율적으로 결합되는 것이 필요하다.

• 한국생산제조학회지
• /
• 제19권5호
• /
• pp.691-697
• /
• 2010

Reliability-based topology optimization (RBTO) is to get an optimal topology satisfying uncertainties of design variables. In this study, reliability-based topology optimization method is applied to the inner reinforcement of vehicle's hood based on BESO. A multi-objective topology optimization technique was implemented to obtain optimal topology of the inner reinforcement of the hood. considering the static stiffness of bending and torsion as well as natural frequency. Performance measure approach (PMA), which has probabilistic constraints that are formulated in terms of the reliability index, is adopted to evaluate the probabilistic constraints. To evaluate the obtained optimal topology by RBTO, it is compared with that of DTO of the inner reinforcement of the hood. It is found that the more suitable topology is obtained through RBTO than DTO even though the final volume of RBTO is a little bit larger than that of DTO. From the result, multiobjective optimization technique based on the BESO can be applied very effectively in topology optimization for vehicle's hood reinforcement considering the static stiffness of bending and torsion as well as natural frequency.