• Proceedings of the Korean Society for Noise and Vibration Engineering Conference
• /
• 1995.10a
• /
• pp.280-283
• /
• 1995

계층형 신경회로망은 학습능력이나 비선형사상능력을 가지고 있고, 그 특징을 이용하여 패턴인식이나 동정 및 제어 등에의 적용이 시도되어 성과를 올리고 있다. 현재, 그 학습법으로 널리 이용되고 있는 것이 역전파학습법으로 최급 강하법이나 공액경사법 등의 최적화 방법이 적용되고 있지만, 학습에 많은 시간이 걸리는 점, 국소적 최적해(local minima)에 해의 수렴이 이루어져 오차가 충분히 작게 되지 않는 점 등이 문제점으로 지적되고 있다. 본 논문에서는 Hu에 의해 고안된 random 탐색법과 조합된 random tabu 탐색법으로 최적결합계수를 구하는 학습알고리즘으로, 국소적 최적해에 수렴하는 것을 방지하고, 수렴정도를 개선하는 새로운 방법을 이용하여 회전기계의 이상진동진단에 적용가능성을 검토하고 오차역전파법에 의한 진단결과와 비교검토한다.

• 한국해양학회지
• /
• v.21 no.3
• /
• pp.184-192
• /
• 1986

A study on the fine structure of the marine diatom has been carried out for taxonomical purpose during the period from Fib. 1982 to Feb. 1986 in Korean coastal waters. Fine structure of small Thalassiosira (Bacillariophyceae) was studied by means of light and scanning electron microscopy. Firstly, four species of Thalassiosira were identified and described. Of these Thalassiosira weissflogii, T.minima and T.oestrupii var venrickae were new records for Korean coastal waters.

• The Bulletin of The Korean Astronomical Society
• /
• v.35 no.1
• /
• pp.50.1-50.1
• /
• 2010

We have analyzed times of minima for of 6 binary systems. Three binary systems show period decrease at rate $3.19{\times}10-5$ yr -1 for SV Cen, $1.35{\times}10-7$ yr -1 for RT Scl and $1.14{\times}10-7$ yr -1 for AD Phe. Two systems show period increase $5.696{\times}10-8$ yr -1 for SX Aur and $6.93{\times}10-8$ yr -1 for GO Cyg. One system shows cyclic period variation. We estimated the mass transfer rate for 5 binary systems. Four systems show asymmetric light curves. Two asymmetric light curves (SV Cen and RT Scl) are due to hot spot caused by mass transfer. And two asymmetric light curves (AD Phe and TY Boo) are due to cool spot caused by magnetic activities on the cooler component. We also obtain absolute dimensions from photometric solution and spectroscopic solution by analyzing their light curves and radial velocity curves, which are collected from literatures, using 2007 version Wilson and Deviney computer code.

• Journal of Electrical Engineering and Technology
• /
• v.3 no.4
• /
• pp.451-459
• /
• 2008

Accurate load modeling is essential for power system static and dynamic analysis. By the nature of the problem of parameter estimation for power system load modeling using actual measurements, multiple local optimal solutions may exist and local methods can be trapped in a local optimal solution giving possibly poor performance. In this paper, Trust-Tech, a novel methodology for global optimization, is applied to tackle the multiple local optimal solutions issue in measurement-based power system load modeling. Multiple sets of parameter values of a composite load model are obtained using Trust-Tech in a deterministic manner. Numerical studies indicate that Trust-Tech along with conventional local methods can be successfully applied to power system load model parameter estimation in measurement-based approaches.

• Journal of Electrical Engineering and Technology
• /
• v.8 no.3
• /
• pp.464-471
• /
• 2013

This paper proposes a hybrid searching algorithm based on parameter identification for power system load models. Hybrid searching was performed by the combination of particle swarm optimization (PSO) and a complex method, which enhances the convergence of solutions closer to minima and takes advantage of global searching with PSO. In this paper, the load model of interest is composed of a ZIP model and a third-order model for induction motors for stability analysis, and parameter sets are obtained that best-fit the output measurement data using the hybrid search. The origin of the hybrid method is to further apply the complex method as a local search for finding better solutions using the selected particles from the performed PSO procedure.

• Journal of Korea Multimedia Society
• /
• v.6 no.6
• /
• pp.968-976
• /
• 2003

In this paper, we propose a novel method for lip recognition. Lip model is built based on the concatenated gray level distribution model, and the recognition problem is simplified as the minimization problem of matching object function. The Down Hill Simplex Algorithm is used for the minimization with the proposed novel method for setting initial condition, which can refrain Iteration from converging to local minima. The proposed algorithm shows extracting lip shape from the test image where Active Shape Model fails.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2001.04b
• /
• pp.1017-1020
• /
• 2001

최근 들어 관심의 대상이 되고 있는 CRM, eCRM은 비즈니스 분야에 중요한 역할을 담당하고 있다. 이를 위해 여러 방법들이 사용되고 있으나, 그 중 데이터 마이닝은 핵심 기술이라 할 수 있다. 다양한 데이터 마이닝 기법가운데 군집화 기법은, 데이터 집합을 유사한 데이터 개체들의 군집들로 분할하여 데이터 속에 존재하는 의미 있는 정보를 얻는 과정이다. 그런데 기존의 군집화 알고리즘들은 사전에 군집의 개수를 미리 결정해져야 하며, 지역적 최적해(local minima)에 수렴할 수 있다는 문제점을 가지고 있다. 본 논문에서는 진화 알고리즘을 사용하여 자동적으로 적절한 군집의 개수를 결정하여 군집화 될 수 있도록 하고, 병렬 탐색을 통해 지역적 최적해에 수렴되는 문제점을 개선한 알고리즘과 적합도 함수를 제안한다.

• Proceedings of the KIEE Conference
• /
• 2002.11b
• /
• pp.76-79
• /
• 2002

In this paper, we propose a Probabilistic Tabu Search(PTS) method for hydro-thermal scheduling. Hydro scheduling has many constraints and very difficult to solve the optimal schedule because it has many local minima. To solve the problem effectively, the proposed method uses two procedures, one is Tabu search procedure that plays a role in local search, and the other is Restarting procedure that enables to diversify its search region. To adjust parameters such as a reducing rate and initial searching region, search strategy is selected according to its probability after Restarting procedure. In order to show the usefulness of the proposed method, the PTS is applied on two cases which have dependent hydro plants and compared to those of other method. The simulation results show it is very efficient and useful algorithm to solve the hydro-thermal scheduling problem.

• Proceedings of the Korean Information Science Society Conference
• /
• 2001.04b
• /
• pp.310-312
• /
• 2001

최근 관심의 대상이 되고 있는 CRM, eCRM에는 데이터 마이닝 기법이 핵심 기술로 이용되고 있다. 이러한 데이터 마이닝 기법가운데 가장 널리 사용되고 있는 군집화는, 데이터 집합을 유사한 데이터의 군집들로 분할하여 데이터 속에 존재하는 의미 있는 정보를 얻는 것이다. 그런데 기존의 군집화 알고리즘은 사전에 군집의 개수를 미리 결정해줘야 하고 잡음에 민감하여 지역적 최적해(local minima)에 수렴할 수 있다는 문제점을 가지고 있다. 이러한 문제점의 개선을 위해, 본 논문에서는 유사도 개념을 적합도 함수로 사용하는 유전자 알고리즘을 적용한 군집화 기법을 제안하다. 특히 적합도 하수에 사용된 군집의 대표값 개념은 요약 정보만을 이용하여 계산속도가 향상되기 때문에 대용량 데이터를 다루는 마이닝에 적합할 것을 기대된다.

• Progress in Superconductivity
• /
• v.1 no.2
• /
• pp.95-98
• /
• 2000

We have studied critical currents of $YBa_2Cu_3O_7$ step-edge junctions with different step orientations with respect to the major axes of $SrTiO_3$ (100) substrates. The junctions were prepared by pulsed laser deposition and argon ion milling with photoresist mask. We investigated current-voltage characteristics and critical current of the junctions as a function of the angle. The junction critical current showed an angle dependent modulation with maxima near 0 or 90 degree and minima near 45 and 135 degrees. The experimental results were analyzed based on the microstructure of the junction along the step and the d-wave symmetry of $YBa_2Cu_3O_7$ superconductor.