• 한국콘텐츠학회논문지
• /
• 제11권4호
• /
• pp.74-82
• /
• 2011

비선형 공정을 퍼지 모델링 하는 것은 전체 입력의 공간 분할 및 퍼지 추론 방법에 따른 퍼지 추론 시스템의 입출력 특성을 분석하는 것이 필요하다. 이를 위해, 퍼지 모델은 입력 변수와 퍼지 입력 공간 분할 및 후반부 다항식 함수에 의한 구조 및 파라미터를 동정함으로서 표현된다. 퍼지 규칙의 전반부에서 입력 데이터의 최소 값과 최대 값을 이용하는 최소-최대 방법 및 입력 데이터를 군집으로 형성하는 C-Means 클러스터링 알고리즘이 퍼지 모델의 동정을 위해 사용되고, 소속 함수는 삼각형, 범종형, 사다리꼴형 소속함수를 사용한다. 퍼지 규칙의 후반부 동정에서 퍼지 추론은 간략 및 선형 추론과 같은 두 가지 형태를 수행한다. 각 규칙의 후반부 파라미터들, 즉 다항식의 계수들의 동정은 표준 최소자승법에 의해 수행된다. 마지막으로, 비선형 공정으로는 널리 이용되는 가스로 데이터를 이용하여 시스템 특성 및 성능을 평가한다.

• 대한원격탐사학회지
• /
• 제33권4호
• /
• pp.423-436
• /
• 2017

This paper presents a deep learning-based road segmentation framework from very high-resolution orthophotos. The proposed method uses Deep Convolutional Autoencoders for end-to-end mapping of orthophotos to road segmentations. In addition, a set of post-processing steps were applied to make the model outputs GIS-ready data that could be useful for various applications. The optimization of the model's parameters is explained which was conducted via grid search method. The model was trained and implemented in Keras, a high-level deep learning framework run on top of Tensorflow. The results show that the proposed model with the best-obtained hyperparameters could segment road objects from orthophotos at an average accuracy of 88.5%. The results of optimization revealed that the best optimization algorithm and activation function for the studied task are Stochastic Gradient Descent (SGD) and Exponential Linear Unit (ELU), respectively. In addition, the best numbers of convolutional filters were found to be 8 for the first and second layers and 128 for the third and fourth layers of the proposed network architecture. Moreover, the analysis on the time complexity of the model showed that the model could be trained in 4 hours and 50 minutes on 1024 high-resolution images of size $106{\times}106pixels$, and segment road objects from similar size and resolution images in around 14 minutes. The results show that the deep learning models such as Convolutional Autoencoders could be a best alternative to traditional machine learning models for road segmentation from aerial photographs.

• 대한산업공학회지
• /
• 제6권2호
• /
• pp.21-29
• /
• 1980

본(本) 연구(硏究)는 투자안(投資案)들간의 상호관계(相互關係) 및 위험(危險)을 고려한 자본예산문제(資本豫算問題)를 다루고 있다. 기존(旣存)의 개발(開發)된 모형(模型)은 확률제약조건계획모형(確率制約條件計劃模型) 및 기대효용극대화모형(期待效用極大化模型)의 두 범주(範疇)로 구분(區分)될 수 있다. 전자의 경우 목적함수(目的凾數)가 다소 제약적(制約的)이며 위험(危險)을 직접적인 형태로 고려하지 않고 있는 반면에 후자는 기대효용(期待效用)에 대한 근사치(近似値)를 사용하기 때문에 투자결정(投資決定)이 최적화(最摘化)되지 못할 가능성이 있다. 본(本) 연구(硏究)는 목적함수(目的凾數)를 보다 일반적(一般的)인 형태로 수정(修正) 보완(補完)함으로써 현실적용성(現實適用性)을 높이고자 하였다. 해법절차(解法節次)로는, 자본예산문제(資本豫算問題)를 우선 비선형(非線型) 0-1 정수계획(整數計劃) 문제로 정식화(定式化)하고, 이를 선형(線型) 0-1 정수계획(整數計劃)문제로 변형(變形)하여 원문제(原問題)의 하한(下限)을 찾은 후 B&B 연산법(演算法)으로 원문제(原問題)의 최적해(最適解)를 구하고 있다.

• Journal of Advanced Marine Engineering and Technology
• /
• 제26권3호
• /
• pp.299-306
• /
• 2002

The load frequency control(LFC) of power system is one of important subjects in view of system operation and control. That is, even though the rapid load disturbances are applied to the given power system, the stable and reliable power should be supplied to the users, converging unconditionally and rapidly the frequency deviations and the tie-line power flow ones of each area into allowable boundary limits. Nonetheless of such needs, if the internal parameter perturbation and the sudden load variation are given, the unstable phenomena of power systems can be often brought out because of the large frequency deviation and the unsuppressible power line one. So, an optimal modulation controller for UC of multi-area power system is designed by a recursive algorithm that determines the state weighting matrix Q of a linear quadratic performance criterion. The optimal modulation controller is based on optimal control and can obtain the exact dynamic response of the UC of multi-area power system in the time domain. The performances of the resultant optimal modulation control, that is, the steady-state deviations of frequency and tie-line power flow and the related dynamics, were investigated and analyzed in detail by being applied to the UC of multi-area power system in the perturbations of predetermined internal parameters. Through the simulation results tried variously in this paper for disturbance of stepwise load changes, the superiorities of the proposed optimal modulation controller in robustness and stability were proved.

• 대한기계학회논문집A
• /
• 제20권7호
• /
• pp.2205-2212
• /
• 1996

In order to move the body of a walking robot translationally, and step over the obstacles, the walking robot must have at least 3 degrees of freedom for each leg. Therefore each leg of the general walking robots can be composed of 3-link system with 3 revolute joints. In this paper, the colsed form of inverse kinimatic solutions is shown for this general 3R linkage. Moreover, in order to have efficient walking volume in rough terrain, the workspace of each log is obtained considering the twist angles and the offsets in D-H parameters. When we design a walking robot, the information of the walking volume is needed for planning desired trajectories of the feet effectively. Appropriate knowledge of the walking volume can also be used to maximize linear or angular velocity of minimize power of stress. However, since it is impossible to obrain the information of walking volume in 3-D space directly from the kinematic equations, the walking volume can be searched through the edge detection algorithm using the triangle tracer with closed from inverse kinematic solutions. In this study, we present the closed form inverse kinematic solutions for 3R linkage model, and the walking volume of 6 legged walking robot which is modeled after the darking bettle, Eleodes obscura sulcipennis, through the method of edge detection for an arbitrary 2 dimensional shape using triangle tracer.

• 대한기계학회논문집A
• /
• 제35권3호
• /
• pp.299-308
• /
• 2011

본 논문은 항만공사용 로봇의 실린더 길이 측정을 위한 압력 옵서버 개발이다. 로봇의 유압 실린더 제어를 위해 변위 센서가 필요하며 일반적으로 LVDT, 리니어 스케일 등이 사용된다. 이러한 센서는 실린더 외부몸통에 장착되므로 건설현장과 같은 열악한 환경에서 사용할 경우 내구성이 좋지 못하다. 본 논문에서는 압력센서를 이용하여 간접적으로 실린더 길이를 측정한다. 압력센서는 유압 밸브박스 내부에 장착되어 외부충격으로부터 보호되며 방수가 용이하다. 오일을 압축성 매개체로 간주하여 실린더의 위치와 속도 함수인 동적압력 방정식을 유도하고 RLS를 이용하여 실린더가 전진, 후진 리미트에 도달할 때마다 파라미터를 갱신한다.

• 한국강구조학회 논문집
• /
• 제16권1호통권68호
• /
• pp.123-134
• /
• 2004

본 연구에서는 효율적인 형상최적화를 위해 다단계 분할기법으로 트러스 구조물의 형상 최적화를 시도하였다. 1단계에서는 단면적을 설계변수로 하여 중량, 또는 체적을 목적함수로 하고 다하중 재하조건 하의 거동제약조건과 부가적인 제약조건을 고려하여 비선형 최적화 문제를 형성한다. 이 비선형 계획문제를 축차 선형계획 문제로 변환하여 개선된 허용방향법으로 최적화하였다. 이때 필요한 도함수는 다른 연구와 달리 효율적이라고 알려진 거동공간법으로 구하였고, 최적화 과정 중 이를 이용하여 부재력를 근사화 함으로써 계산의 효율성을 높였다. 2단계에서는 형상 설계변수만을 고려한 무제약 최적화 문제로 형성한 후 일방향 탐사기법을 적용하여 형상을 최적화하였다. 이와 같이 구성된 본 연구의 알고리즘을 몇 가지 트러스 구조물에 적용하여 본 알고리즘의 적용성과 효율성 및 타당성을 증명하였다.

• Tribology and Lubricants
• /
• 제35권6호
• /
• pp.369-375
• /
• 2019

This paper presents a new method on how to calculate the thrust forces acting on an auto-leveling device in headlamps for passenger vehicles. The leveling device is used to lower the angle of lights when a load in the trunk of the vehicle lifts it. In the process of the headlamp design, it is imperative to predict the external forces so that the designers can decide whether to proceed or not. The device is composed of three pivot joints with no reaction moment, a plate that holds the lamp, and a leveling motor that changes rotation to linear motion. In this study, force balance, moment balance, and geometric compatibility are applied to the leveling device system so that a nonlinear system of equations can be derived; the multi-dimensional Newton-Raphson algorithm is then used to solve these. A sensitivity analysis is carried out to verify which design variables affect the system the most: the mass of the lamp and the height between the pivot and leveling device affect the thrust forces the most. Then, considering the friction forces between the moving parts, the hysteresis of the forces are derived. An experimental apparatus, designed and developed in this study, is used to verify the exactness of the derived equations. The results from experiments coincide well with the calculated results. The friction hysteresis, in particular, proves this upon analysis.

• International Journal of Automotive Technology
• /
• 제8권6호
• /
• pp.731-744
• /
• 2007

A method for dynamic analysis and design calculation of a Powertrain Mounting System(PMS) including Hydraulic Engine Mounts(HEM) is developed with the aim of controlling powertrain motion and reducing low-frequency vibration in pitch and bounce modes. Here the pitch mode of the powertrain is defined as the mode rotating around the crankshaft of an engine for a transversely mounted powertrain. The powertrain is modeled as a rigid body connected to rigid ground by rubber mounts and/or HEMs. A mount is simplified as a three-dimensional spring with damping elements in its Local Coordinate System(LCS). The relation between force and displacement of each mount in its LCS is usually nonlinear and is simplified as piecewise linear in five ranges in this paper. An equation for estimating displacements of the powertrain center of gravity(C.G.) under static or quasi-static load is developed using Newton's second law, and an iterative algorithm is presented to calculate the displacements. Also an equation for analyzing the dynamic response of the powertrain under ground and engine shake excitations is derived using Newton's second law. Formulae for calculating reaction forces and displacements at each mount are presented. A generic PMS with four rubber mounts or two rubber mounts and two HEMs are used to validate the dynamic analysis and design calculation methods. Calculated displacements of the powertrain C.G. under static or quasi-static loads show that a powertrain motion can meet the displacement limits by properly selecting the stiffness and coordinates of the tuning points of each mount in its LCS using the calculation methods developed in this paper. Simulation results of the dynamic responses of a powertrain C.G. and the reaction forces at mounts demonstrate that resonance peaks can be reduced effectively with HEMs designed on the basis of the proposed methods.

• Journal of Biosystems Engineering
• /
• 제35권4호
• /
• pp.257-261
• /
• 2010

Surface energy is the principal factor of anti-condensation. High surface energy appears hydrophilic itself and low surface energy represents hydrophobic itself. The contact angle is widely being used for measurement of surface energy of materials, evaluation of coating performances, measurement of wettability, and so on. However, the existing contact angle measuring system is so expensive for purchasing and complicated, so it takes a lot of time and money to use. This study was conducted to develop the algorithm for evaluating hydrophilic performance through measuring the contact angle of water droplet automatically, and fabricate relatively simple measuring system using a low-cost monochrome camera and image processing. A constant amount of water was firstly allocated on a slide by a micropipette, and then the image of water droplet was captured by monochrome digital camera and sent to a computer. The image was binarized and then reduced noises by labeling. Finally, the contact angle of water droplet was computed by using three points (left, right, and top coordinates), simple linear mathematics, and trigonometric function. The experimental results demonstrated the accuracy and reproducibility of the developed system showing less deviations and deviation ratio.