• 전자공학회논문지SC
• /
• 제37권6호
• /
• pp.50-63
• /
• 2000

본 논문에서는 소프트 컴퓨팅 기법을 이용한 새로운 근전도 신호 패턴 분류 방법을 제안한다. 재활 로봇시스템에서 기존에 사용되었던 여러 가지 입력 장치(음성, 레이저 포인터, 키패드, 3차원 입력기 등)에 비해 근전도 신호를 이용한 방식이 가지는 장점을 서술한다. 기존의 근전도 신호 분류 방법의 문제점인 사용자 의존성을 줄이기 위해 제안한 사용자 독립적인 특징 선택 방법에 대해 상술한다. 선택된 특징 집합을 이용하여 퍼지 패턴 분류기 및 퍼지 최대-최소 신경망을 구성하여 학습 전(퍼지 패턴 분류기)과 학습 후(퍼지 최대-최소 신경망)에 각각 83%와 90%의 분류 성공률을 얻어 제안된 방법의 유용성을 확인할 수 있었다.

• 한국지능시스템학회논문지
• /
• 제15권1호
• /
• pp.69-74
• /
• 2005

본 논문에서는 신경망 학습을 위한 데이터 획득 시 생길 수 있는 오차론 줄이기 위해 획득 데이터에 대한 전처리 과정을 퍼지로써 구현하는 알고리즘을 제안하였다. 신경망은 주어진 정보론 이용하여 학습을 가능하게 함으로써 시스템의 특징을 추출하는 데 매우 우수한 능력을 발휘하고 있다. 그러나 이는 학습에 사용하는 데이터에 오차가 포함되지 않는다는 점을 전제로 하고 있다. 그런데 데이터 획득 과정이 인간의 주관적 판단에 의해 수작업으로 이루어지는 경우 학습 데이터는 오차가 존재할 수 있다. 학습 데이터의 오차론 줄이기 위해 초기에 획득된 데이터를 분석하고 추가적인 후보 데이터를 선정하여 그 중에서 가장 적절한 데이터를 고르게 하였다. 데이터 후보 추천 시스템은 데이터 획득 과정에서 큰 영향을 미치는 물체의 거리와 크기를 모두 고려할 수 있도록 퍼지 모델로써 구현하였다. 후보 추천 결과, 상당수의 오차 데이터를 수정할 수 있었으며 이는 심리적 요인과 신체적 요인이 크게 작용한 데이터일수록 큰 효과를 나타냈다.

• 한국지능시스템학회논문지
• /
• 제14권6호
• /
• pp.720-725
• /
• 2004

로봇은 조립, 도장, 용접 등 단순 반복 작업이나 위험한 지역의 탐사 및 산업현장에서 벗어나 좀더 다양한 분야로 발전되어지고 있다. 최근 로봇의 형태는 인간의 명령을 이행하고 스스로 학습하며, 감정을 지닐 수 있는 인공지능을 내장한 로봇이다. 활용의 예 중에는 '가족 도우미'의 역할을 수행하는 로봇으로 가사, 방범, 오락, 교육 그리고 인공비서 등의 기능을 담당하는 형태로 향후 가정의 필수품으로 자리 잡을 전망이다. 이러한 가정용 로봇의 구현을 위해서 인공지능의 요소를 활용하는 것은 당연하다. 일차적으로 로봇이 여러 가지의 기능을 수행하기 위해서는 환경 정보를 받아들이는 센서의 역할이 크며 이런 센서를 사용조건에 맞게 활용하는 것도 중요하다. 본 논문에서는 로봇에 부착된 여러 개의 센서를 응합하고 융합된 여러 종류의 센서값을 이용하여 로봇이 주변환경에 맞게 행동을 할 수 있도록 소프트 컴퓨팅 기법을 이용하였다. 또한 로봇의 행동모듈을 구성하여 인간에게 시각$.$청각적인 효과를 줄 수 있도록 인간 친화형 지능 로봇을 구현하고자 한다.

• Geomechanics and Engineering
• /
• 제37권4호
• /
• pp.383-393
• /
• 2024

Monopile foundations of offshore wind turbines embedded in soft clay are subjected to the long-term cyclic lateral loads induced by winds, currents, and waves, the vibration of monopile leads to the accumulation of pore pressure and cyclic strains in the soil in its vicinity, which poses a threat to the safety operation of monopile. The researchers mainly focused on the hysteretic stress-strain relationship of soft clay and kinds of stiffness degradation models have been adopted, which may consume considerable computing resources and is not applicable for the long-term bearing performance analysis of monopile. In this study, a modified cyclic stiffness degradation model considering the effect of plastic strain and pore pressure change has been proposed and validated by comparing with the triaxial test results. Subsequently, the effects of cyclic load ratio, pile aspect ratio, number of load cycles, and length to embedded depth ratio on the accumulated rotation angle and pore pressure are presented. The results indicate the number of load cycles can significantly affect the accumulated rotation angle of monopile, whereas the accumulated pore pressure distribution along the pile merely changes with pile diameter, embedded length, and the number of load cycles, the stiffness of monopile can be significantly weakened by decreasing the embedded depth ratio L/H of monopile. The stiffness degradation of soil is more significant in the passive earth pressure zone, in which soil liquefaction is likely to occur. Furthermore, the suitability of the "accumulated rotation angle" and "accumulated pore pressure" design criteria for determining the required cyclic load ratio are discussed.

• International Journal of Naval Architecture and Ocean Engineering
• /
• 제11권1호
• /
• pp.59-69
• /
• 2019

This paper focuses on the numerical simulation of ice abrasion induced by unbreakable ice floe. Under the assumption that unbreakable floes behave as rigid body, the Discrete Element Method (DEM) was applied to simulate the interaction between a fixed structure and ice floes. DEM is a numerical technique which is eligible for computing the motion and effect of a large number of particles. In DEM simulation, individual ice floe was treated as single rigid element which interacts with each other following the given interaction rules. Interactions between the ice floes and structure were defined by soft contact and viscous Coulomb friction laws. To derive the details of the interactions in terms of interaction parameters, the Finite Element Method (FEM) was employed. An abrasion process between a structure and an ice floe was simulated by FEM, and the parameters in DEM such as contact stiffness, contact damping coefficient, etc. were calibrated based on the FEM result. Resultantly, contact length and contact path length, which are the most important factors in ice abrasion prediction, were calculated from both DEM and FEM and compared with each other. The results showed good correspondence between the two results, providing superior numerical efficiency of DEM.

• Computers and Concrete
• /
• 제28권6호
• /
• pp.635-642
• /
• 2021

In this paper, soft learning techniques are used to predict the ultimate torsional capacity of Reinforced Concrete beams strengthened with Fiber Reinforced Polymer. Soft computing techniques, namely Artificial Neural Network, trained by various back propagation algorithms, and Particle Swarm Optimization (PSO) algorithm, have been used to model and predict the torsional strength of Reinforced Concrete beams strengthened with Fiber Reinforced Polymer. The performance of each model has been evaluated by using statistical parameters such as coefficient of determination (R²), Root Mean Square Error (RMSE), and Mean Absolute Percentage Error (MAPE). The hybrid PSO NN model resulted in an R² of 0.9292 with an RMSE of 5.35 for training and an R² of 0.9328 with an RMSE of 4.57 for testing. Another model, ANN BP, produced an R² of 0.9125 with an RMSE of 6.17 for training and an R² of 0.8951 with an RMSE of 5.79 for testing. The results of the PSO NN model were in close agreement with the experimental values. Thus, the PSO NN model can be used to predict the ultimate torsional capacity of RC beams strengthened with FRP with greater acceptable accuracy.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제8권10호
• /
• pp.3361-3377
• /
• 2014

Synchronous computing models provided by real-time synchrony protocols, such as TTA [1] and PALS [2], greatly simplify the design, implementation, and verification of real-time distributed systems. However, their application to real systems has been limited since their assumptions on underlying systems are hard to satisfy. In particular, most previous real-time synchrony protocols hypothesize the existence of underlying fault tolerant real-time networks. This, however, might not be true in most soft real-time applications. In this paper, we propose a practical approach to a synchrony protocol, called Quality-Aware PALS (qPALS), which provides the benefits of a synchronous computing model in environments where no fault-tolerant real-time network is available. qPALS supports two flexible global synchronization protocols: one tailored for the performance and the other for the correctness of synchronization. Hence, applications can make a negotiation flexibly between performance and correctness. In qPALS, the Quality-of-Service (QoS) on synchronization and consistency is specified in a probabilistic manner, and the specified QoS is supported under dynamic and unpredictable network environments via a control-theoretic approach. Our simulation results show that qPALS supports highly reliable synchronization for critical events while still supporting the efficiency and performance even when the underlying network is not stable.

• Journal of Advanced Marine Engineering and Technology
• /
• 제30권7호
• /
• pp.791-799
• /
• 2006

The aim of this paper is to design an adaptive speed control system of a marine diesel engine by fusion of hard computing based proportional integral derivative (PID) control and soft computing based fuzzy control methods. The model of a marine diesel engine is considered as a typical non oscillatory second order system. When its model and the actual marine diesel engine ate not matched, it is hard to control the speed of the marine diesel engine. Therefore, this paper proposes two methods in order to obtain the speed control characteristics of a marine diesel engine. One is an efficient method to determine the PID control parameters of the nominal model of a marine diesel engine. Second is a reference adaptive speed control method that uses a fuzzy controller and derivative operator for tracking the nominal model of the marine diesel engine. It was found that the proposed PID parameters adjustment method is better than the Ziegler & Nichols' method, and that a model reference adaptive control is superior to using only PID controller. The improved control method proposed here, could be applied to other systems when a model of a system does not match the actual system.

• 한국CDE학회논문집
• /
• 제5권2호
• /
• pp.127-135
• /
• 2000

Emergent computing paradigms such as genetic algorithms have found increased use in problems in engineering design. These computational tools have been shown to be applicable in the solution of generically difficult design optimization problems characterized by nonconvexities in the design space and the presence of discrete and integer design variables. Another aspect of these computational paradigms that have been lumped under the bread subject category of soft computing, is the domain of artificial intelligence, knowledge-based expert system, and machine learning. The paper explores a machine learning paradigm referred to as teaming classifier systems to construct the high-quality global function approximations between the design variables and a response function for subsequent use in design optimization. A classifier system is a machine teaming system which learns syntactically simple string rules, called classifiers for guiding the system's performance in an arbitrary environment. The capability of a learning classifier system facilitates the adaptive selection of the optimal number of training data according to the noise and multimodality in the design space of interest. The present study used the polynomial based response surface as global function approximation tools and showed its effectiveness in the improvement on the approximation performance.

• 한국통신학회논문지
• /
• 제34권5C호
• /
• pp.498-504
• /
• 2009

Low-density parity-check (LDPC) 부호의 복호에는 성능이 좋은 합곱 알고리즘(sum-product algorithm; SPA)과 하드웨어가 간단한 비트 반전(bit-flipping; BF) 알고리즘이 많이 쓰이고 있다. 본 논문은 이들 두 가지 방법의 장점을 가지는 저복잡도 고성능 복호 알고리즘을 제안한다. 본 제안된 유연 비트 반전(soft bit-flipping) 알고리즘은 비트와 체크 노드 사이에 전달되는 메시지를 계산하는 데 단순한 비교와 덧셈 연산만을 필요로 하며 연산량이 적다는 장점이 있다. 또한 연산이 완료된 메시지의 활용률을 높이고 비균등 양자화(non-uniform quantization)를 채용하여 1000 내외의 부호 길이에서 SPA 에 0.4dB 근접하는 신호대 잡음비(signal-to-noise ratio)를 달성하였다. 본 논문에서 제안된 알고리즘을 이용하면, 행 무게(row weight)와 열 무게(column weight)가 높아서 종래의 SPA로 구현하기 어려웠던 부호를 비교적 좋은 오율 성능을 유지하면서 실용적으로 구현할 수 있다.