• 응용통계연구
• /
• 제30권1호
• /
• pp.135-145
• /
• 2017

주성분 분석(principal component analysis; PCA)은 서로 상관되어 있는 다변량 자료의 차원을 축소하는 대표적인 기법으로 많은 다변량 분석에서 활용되고 있다. 하지만 주성분은 모든 변수들의 선형결합으로 이루어지므로, 그 결과의 해석이 어렵다는 한계가 있다. sparse PCA(SPCA) 방법은 elastic net 형태의 벌점함수를 이용하여 보다 성긴(sparse) 적재를 가진 수정된 주성분을 만들어주지만, 변수들의 그룹구조를 이용하지 못한다는 한계가 있다. 이에 본 연구에서는 기존 SPCA를 개선하여, 자료가 그룹화되어 있는 경우에 유의한 그룹을 선택함과 동시에 그룹 내 불필요한 변수를 제거할 수 있는 새로운 주성분 분석 방법을 제시하고자 한다. 그룹과 그룹 내 변수 구조를 모형 적합에 이용하기 위하여, sparse 주성분 분석에서의 elastic net 벌점함수 대신에 계층적 벌점함수 형태를 고려하였다. 또한 실제 자료의 분석을 통해 제안 방법의 성능 및 유용성을 입증하였다.

• 대한기계학회논문집A
• /
• 제21권11호
• /
• pp.1844-1851
• /
• 1997

Displacement fields and interface stresses are obtained by modifying the potential energy functional with a penalty function which enforces the continuity of stresses at the interface of two-materials. Based on the displacement field and the interface stresses, a new methodology to generate a continuous stress field over the entire domain including the interface of the dissimilar materials has been proposed by combining the L$^{2}$ projection method of stress-smoothing and the Loubignac's iterative method of improving the displacement field. Stress analysis was carried out on two examples which are made of highly dissimilar materials. As a result of the analysis, it is found that the proposed method provides improved continuity of the stress field over the entire domain as well as predicting accurate nodal stresses at the interface. In contrast, the conventional displacement-based finite element method provides significant stress discontinuties at the interfaces. In addition, it was found that the total strain energy evaluated from the improved continuous stress field converge to the exact value as increasing the number of iterations in the proposed method.

• Structural Engineering and Mechanics
• /
• 제23권4호
• /
• pp.387-407
• /
• 2006

A Genetic Algorithm (hereinafter GA) based optimum design algorithm and program for plane steel frames with partially restrained connections is presented. The algorithm was incorporated with the refined plastic hinge analysis method, in which geometric nonlinearity was considered by using the stability functions of beam-column members and material nonlinearity was considered by using the gradual stiffness degradation model that included the effects of residual stress, moment redistribution by the occurrence of plastic hinges, partially restrained connections, and the geometric imperfection of members. In the genetic algorithm, a tournament selection method and micro-GAs were employed. The fitness function for the genetic algorithm was expressed as an unconstrained function composed of objective and penalty functions. The objective and penalty functions were expressed, respectively, as the weight of steel frames and the constraint functions which account for the requirements of load-carrying capacity, serviceability, ductility, and construction workability. To verify the appropriateness of the present method, the optimum design results of two plane steel frames with fully and partially restrained connections were compared.

• 한국공작기계학회:학술대회논문집
• /
• 한국공작기계학회 2003년도 춘계학술대회 논문집
• /
• pp.504-509
• /
• 2003

In this paper, a multi-step optimization using a G.A.(Genetic Algorithm) with variable penalty function is introduced to the structural design optimization of a high speed machining center. The design problem, in this case, is to find out the best cross-section shapes and dimensions of structural members which minimize the static compliance, the dynamic compliance, and the weight of the machine structure simultaneously. The first step is the cross-section shape optimization, in which only the section members are selected to survive whose cross-section area have above a critical value. The second step is a static design optimization, in which the static compliance and the weight of the machine structure are minimized under some dimensional constraints and deflection limits. The third step is a dynamic design optimization, where the dynamic compliance and the structure weight are minimized under the same constraints as those of the second step. The proposed design optimization method was successful applied to the machining center structural design optimization. As a result, static and dynamic compliances were reduced to 16% and 53% respectively from the initial design, while the weight of the structure are also reduced slightly.

• 대한기계학회논문집
• /
• 제9권4호
• /
• pp.528-538
• /
• 1985

본 논문에서는 crank-rocker 운동에 의해 함수 y=sinx를 만족시키는 3차원 R-S-S-R 함수발생기구를 FDP(Fletcher-Davidon-Powell) 최적화 기법을 응용, 운동학적 으로 무오차점을 고려, 설계하려고 한다. 한편 상기와 같이 최적 설계된 기구의 실 제 운동과 요구된 운동과의 오차분석 및 운동 가능성 여부(mobility check)를 검토하 였으며 FDP 최적화 기법에 필요한 벌칙함수(penalty function)로써 Fiacco-McCormic- krhk Powell의 벌칙함수를 사용하여 그 결과를 비교하였다.

• Structural Engineering and Mechanics
• /
• 제23권1호
• /
• pp.29-42
• /
• 2006

This paper describes the design scheme of the three-dimensional structures based on the concept of the cellular automata simulation. The cellular automata simulation is performed according to the local rule. In this paper, the local rule is derived in the mathematical formulation from the optimization problem. The cell density is taken as the design variable. Two objective functions are defined for reducing the total weight of the structure and obtaining the fully stressed structure. The constraint condition is defined for defining the local rule. The penalty function is defined from the objective functions and the constraint condition. Minimization of the penalty function with respect to the design parameter leads to the local rule. The derived rule is applied to the design of the three-dimensional structure first. The final structure can be obtained successfully. However, the computational cost is expensive. So, in order to reduce the computational cost, the material parameters $c_1$ and $c_2$ and the value of the cell rejection criterion (CRC) are changed. The results show that the computational cost depends on the parameters and the CRC value.

• 제어로봇시스템학회논문지
• /
• 제13권11호
• /
• pp.1048-1052
• /
• 2007

This paper presents a rank-constrained linear matrix inequality(LMI) approach to simultaneous linear-quadratic(LQ) optimal control by static output feedback. Simultaneous LQ optimal control is formulated as an LMI optimization problem with a nonconvex rank condition. An iterative penalty method recently developed is applied to solve this rank-constrained LMI optimization problem. Numerical experiments are performed to illustrate the proposed method, and the results are compared with those of previous work.

• 소성∙가공
• /
• 제12권1호
• /
• pp.26-33
• /
• 2003

This paper is concerned with the numerical analysis of frictional contact problems in axisymmetric bodies using the rigid-plastic finite element method. A contact finite element method, based on a penalty function, are derived from variational formulations. The contact boundary condition between two deformable bodies is prescribed by the proposed algorithm. The program which can handle frictional contact problem is developed by using pre-existing rigid-plastic finite element code. Some examples used in this paper illustrate the effectiveness of the proposed formulations and algorithms. Efforts focus on the deformation patterns, contact force, and velocity gradient through the various simulations.

• 제어로봇시스템학회논문지
• /
• 제15권1호
• /
• pp.67-71
• /
• 2009

This paper presents a rank-constrained linear matrix inequality (LMI) approach to the design of a multi-objective controller such as $H_2/H_{\infty}$ control. Multi-objective control is formulated as an LMI optimization problem with a nonconvex rank condition, which is imposed on the controller gain matirx not Lyapunov matrices. With this rank-constrained formulation, we can expect to reduce conservatism because we can use separate Lyapunov matrices for different control objectives. An iterative penalty method is applied to solve this rank-constrained LMI optimization problem. Numerical experiments are performed to illustrate the proposed method.

• 경영과학
• /
• 제13권3호
• /
• pp.187-198
• /
• 1996

The traveling salesman problem has been studied for many years since the model can be used for various applications such as vehicle routing, job sequencing, clustering a data array, and so on. In this paper one of the typical exact algorithms for TSP, Minty's, will be modified to improve the performance of the algorithm on the applications without losing simplicity. The Little's algorithm gives good results, however, the simple and plain Minty's algorithm for solving shortest-route problems has the most intuitive appeal. The suggested Minty's modification is based on the creation of penalty-values on the matrix of a TSP. Computer experiments are made to verify the effectiveness of the modification.