• 음성과학
• /
• 제5권1호
• /
• pp.7-21
• /
• 1999

This paper proposes an improved training procedure in speech recognition based on the continuous density of the Hidden Markov Model (CDHMM). Of the three parameters (initial state distribution probability, state transition probability, output probability density function (p.d.f.) of state) governing the CDHMM model, we focus on the third parameter and propose an efficient algorithm that determines the p.d.f. of each state. It is known that the resulting CDHMM model converges to a local maximum point of parameter estimation via the iterative Expectation Maximization procedure. Specifically, we propose two independent algorithms that can be embedded in the segmental K -means training procedure by replacing relevant key steps; the adaptation of the number of mixture Gaussian p.d.f. and the initialization using the CDHMM parameters previously estimated. The proposed adaptation algorithm searches for the optimal number of mixture Gaussian humps to ensure that the p.d.f. is consistently re-estimated, enabling the model to converge toward the global maximum point. By applying an appropriate threshold value, which measures the amount of collective changes of weighted variances, the optimized number of mixture Gaussian branch is determined. The initialization algorithm essentially exploits the CDHMM parameters previously estimated and uses them as the basis for the current initial segmentation subroutine. It captures the trend of previous training history whereas the uniform segmentation decimates it. The recognition performance of the proposed adaptation procedures along with the suggested initialization is verified to be always better than that of existing training procedure using fixed number of mixture Gaussian p.d.f.

• 한국추진공학회:학술대회논문집
• /
• 한국추진공학회 2011년도 제37회 추계학술대회논문집
• /
• pp.127-134
• /
• 2011

유전알고리즘을 사용하여 액체로켓엔진의 연소실 압력과 노즐 확장비, O/F 비 등 주요 설계변수를 최적화하였다. 대상엔진은 LO2/RP-1을 추진제로 사용하는 개방형 가스발생기 사이클을 대상으로 하였다. 연소실의 물성치는 CEA2를 이용하였으며, 무게 산출은 참고문헌을 바탕으로 모델링 하였다. 최적설계의 목적함수는 비추력과 추력중량비를 다중목표로 설정하여 가중치 방법을 사용하였다. 유전알고리즘을 최적화 과정을 거친 결과 비추력은 최대 4%, 추력중량비는 최대 23% 정도 증가하였다. 또한 다양한 추력에 대해서 Pareto frontier line을 얻었다.

• 한국정밀공학회지
• /
• 제19권1호
• /
• pp.119-126
• /
• 2002

In this paper, we present a new approach based on an elitist genetic algorithm for the task of aligning the position of a robot gripper using CCD cameras. The vision-based control scheme for the task of aligning the gripper with the desired position is implemented by image information. The relationship between the camera space location and the robot joint coordinates is estimated using a camera-space parameter modal that generalizes known manipulator kinematics to accommodate unknown relative camera position and orientation. To find the joint angles of a robot manipulator for reaching the target position in the image space, we apply an elitist genetic algorithm instead of a nonlinear least square error method. Since GA employs parallel search, it has good performance in solving optimization problems. In order to improve convergence speed, the real coding method and geometry constraint conditions are used. Experiments are carried out to exhibit the effectiveness of vision-based control using an elitist genetic algorithm with a real coding method.

• 한국공작기계학회논문집
• /
• 제17권5호
• /
• pp.30-35
• /
• 2008

As the quality of a weld feint is strongly influenced by process parameters during the welding process, an intelligent algorithms that can predict the bead geometry and shape to accomplish the desired mechanical properties of the weldment should be developed. This paper focuses on the development of mathematical models fur the selection of process parameters and the prediction of bead geometry(bead width, bead height and penetration) in robotic GMA(Gas Metal Arc) welding. Factorial design can be employed as a guide for optimization of process parameters. Three factors were incorporated into the factorial model: arc current, welding voltage and welding speed. A sensitivity analysis has been conducted and compared the relative impact of three process parameters on bead geometry in order to verify the measurement errors on the values of the uncertainty in estimated parameters. The results obtained show that developed mathematical models can be applied to estimate the effectiveness of process parameters for a given bead geometry, and a change of process parameters affects the bead width and bead height more strongly than penetration relatively.

• Structural Engineering and Mechanics
• /
• 제72권1호
• /
• pp.83-97
• /
• 2019

In this research, the vulnerability of some reinforced concrete frames with different stories are studied based on the Park-Ang Damage Index. The damages of the frames are investigated under various earthquakes with nonlinear dynamic analysis in IDARC software. By examining the most important characteristics of earthquake parameters, the damage index and vulnerability of these frames are investigated in this software. The intensity of Erias, velocity spectral intensity (VSI) and peak ground velocity (PGV) had the highest correlation, and root mean square of displacement ($D_{rms}$) had the lowest correlation coefficient among the parameters. Then, the particle swarm optimization (PSO) algorithm was used, and the sinusoidal waves were equivalent to the used earthquakes according to the most influential parameters above. The damage index equivalent to these waves is estimated using nonlinear dynamics analysis. The comparison between the damages caused by earthquakes and equivalent sinusoidal waves is done too. The generations of sinusoidal waves equivalent to different earthquakes are generalized in some reinforced concrete frames. The equivalent sinusoidal wave method was exact enough because the greatest difference between the results of the main and artificial accelerator damage index was about 5 percent. Also sinusoidal waves were more consistent with the damage indices of the structures compared to the earthquake parameters.

• 제어로봇시스템학회논문지
• /
• 제22권5호
• /
• pp.326-331
• /
• 2016

This paper presents a novel robust adaptive AC8B exciter system against synchronous generators for ships. A PID (proportional integral derivative) control framework, which is a part of the AC8B exciter system, is simply composed of nominal and auxiliary control configurations. For selecting these proper parameter values, the former is conventionally chosen based on the experience and knowledge of experts, and the latter is optimally estimated via a neural networks optimization procedure. Additionally, we propose an online parameter learning-based auxiliary control to practically cope with deterioration of control performance owing to uncertainty in electric generator systems. Such a control mechanism ensures the robustness and adaptability of an AC8B exciter to enhance control performance in real-time implementation. We carried out simulation experiments to test the reliability of the proposed robust adaptive AC8B exciter system and prove its superiority through a comparative study in which a conventional PID control-based AC8B exciter system is similarly applied to our simulation experiments under the same simulation scenarios.

• Journal of Mechanical Science and Technology
• /
• 제18권12호
• /
• pp.2080-2094
• /
• 2004

A model predictive controller is designed to control thermal power in a nuclear reactor. The basic concept of the model predictive control is to solve an optimization problem for finite future time steps at current time, to implement only the first optimal control input among the solved control inputs, and to repeat the procedure at each subsequent instant. A controller design model used for designing the model predictive controller is estimated every time step by applying a recursive parameter estimation algorithm. A 3-dimensional nuclear reactor analysis code, MASTER that was developed by Korea Atomic Energy Research Institute (KAERI), was used to verify the proposed controller for a nuclear reactor. It was known that the nuclear power controlled by the proposed controller well tracks the desired power level and the desired axial power distribution.

• 한국강구조학회 논문집
• /
• 제21권4호
• /
• pp.361-373
• /
• 2009

유전자 알고리즘을 이용한 닐센아치교의 최적설계기법을 이 논문에서 제시하였다. 설계 매개변수로는 닐센아치교의 아치-라이즈비와 강중비에 대해서 최적화기법을 적용하여 각각의 거동을 분석하고, 적정성을 평가하여 최적의 매개변수 값을 결정하였다. 매개변수의 결정은 구조물의 안전성과 사용성 그리고 경제성에 중요한 설계인자로서 정형화가 요구된다. 이를 위해 최적화 기법으로 전역 최적해 탐색능력이 우수한 유전자 알고리즘을 사용하였으며, 설계 목적함수로는 구조물의 총 중량을 사용하였고, 제약조건으로는 변위, 응력, 시공성 제약조건을 두었다. 구조해석은 미소변위이론에 의한 탄성해석을 수행하여 유전자 알고리즘과 조합하여 병렬연산으로 수행시간을 단축시켰다. 이 연구에서 개발된 최적설계기법을 사용하여 최적의 아치-라이즈비와 강중비, 최적설계영역을 제시 하였으며 실무에 적용할 수 있도록 하였다.

• 한국시뮬레이션학회논문지
• /
• 제18권2호
• /
• pp.49-55
• /
• 2009

본 연구는 확률적 시스템에서 관심 성과함수의 기대치의 최적을 유도하는 서비스 자원의 최적 배분 문제를 조사하였다. 이러한 목적으로 통제변수를 활용하여 성과함수 기대치에 대한 서비스 자원 파라미터의 gradient를 구하는 방법을 제안하고 이를 최적화 기법의 탐색과정에 적용하여 가용 자원의 최적 배분 문제를 분석하였다. 제안된 gradient 추정 방법은 시뮬레이션 실험에서 입력 파라미터의 차원이 증가하더라도 추가로 표본점의 수를 증가시킬 필요가 없이 단일점에서 시뮬레이션 반응 결과만을 활용하고 또한 시뮬레이션의 발전과정에서 성과함수와 입력 파라미터 사이의 논리적인 관계를 기술할 필요가 없어 적용하기에 편리하다고 볼 수 있다. 본 연구의 결과를 다 차원 파라미터 공간으로의 확장하는 문제와 다양한 형태의 시뮬레이션 모형으로 적용 문제는 향후 연구해야 할 과제로 생각된다.

• 한국수자원학회논문집
• /
• 제37권12호
• /
• pp.1055-1063
• /
• 2004

공간적 위치 및 유량 값에 따라 각 계산점마다 조도계수의 값이 달리 주어질 수 있도록 하는 가변 조도변수 부정류 계산모형을 수립하였다. 유량과 조도계수의 관계식으로는 계단함수 또는 멱함수를 적용할 수 있도록 하였다. 수립된 모형을 충주댐부터 팔당댐까지의 남한강 구간에 적용하여 최적화에 의한 매개변수의 추정을 수행하였다. 가변 매개변수 모형의 보정 결과, 계단함수 도형 및 멱함수 모형 모두 유량이 커질수록 조도계수가 감소하는 경향이 일관되게 나타났다. 이러한 경향은 여주 지점 상류구간의 경우에 더욱 현저한 것으로 나타났다. 가변 조도계수 모형의 매개변수 추정에 따른 오차가 고정 조도계수 모형의 경우보다 작아짐을 알 수 있었다.