• 센서학회지
• /
• 제12권2호
• /
• pp.79-87
• /
• 2003

본 논문에서는 비선형 특성을 갖는 센서 배열을 사용하는 시스템에서, 센서 배열에 soft computing을 이용하여 선형적인 특성을 얻고, 배열의 특성을 이용하여 외란의 영향을 감소시켜 시스템의 성능을 향상시키는 방법을 제안한다. 대표적인 예로 배열에 사용된 광전도 센서인 CdS는 단일 CdS 셀 조차도 비선형적인 특성을 갖고 있기 때문에 비선형 특성을 소속 함수로 표현하는 퍼지 변수를 사용한다. 배열로 사용되는 센서에 각 단일 센서의 소속 함수와 특성을 이용하는 퍼지 논리를 적용하여 전체적으로 선형화된 센서 특성을 얻는다. 각 센서의 비선형적 요소를 표현하는 소속함수의 파라미터들에 유전 알고리즘을 사용하여 최적화 된 선형 특성을 얻어 시스템의 성능을 향상시킬 수 있다. 또한 보정을 위한 센서를 추가하지 않고 센서 배열의 특성을 이용하여 광전도 센서가 민감한 영향을 받는 외란을 보정하여 시스템의 외란 제거 성능을 향상시킬 수 있다. CdS 센서 배열의 자체 비선형성 뿐만 아니라 입력되는 물체의 그림자는 경계가 뚜렷하지 않고 흐릿하므로 퍼지 논리를 이용하는 방법이 거리측정과 외부 광원에 의한 외란에 대해 향상된 결과를 보인다. 제안된 방법을 적용한 센서 배열을 자기부상(Magnetic Levitation System)에서 볼의 거리 측정에 적용하여 성능을 검증한다.

• Earthquakes and Structures
• /
• 제5권6호
• /
• pp.657-678
• /
• 2013

For economical earthquake resistant design of cable-stayed bridge tower, the use of energy dissipation systems for the earthquake protection of steel structures represents an alternative seismic design method where the tower structure could be constructed to dissipate a large amount of earthquake input energy through inelastic deformations in certain positions, which could be easily retrofitted after damage. The design of energy dissipation systems for bridges could be achieved as the result of two conflicting requirements: no damage under serviceability limit state load condition and maximum dissipation under ultimate limit state load condition. A new concept for cable-stayed bridge tower seismic design that incorporates sacrificial link scheme of low yield point steel horizontal beam is introduced to enable the tower frame structure to remain elastic under large seismic excitation. A nonlinear dynamic analysis for the tower model with the proposed energy dissipation systems is carried out and compared to the response obtained for the tower with its original configuration. The improvement in seismic performance of the tower with supplemental passive energy dissipation system has been measured in terms of the reduction achieved in different response quantities. Obtained results show that the proposed energy dissipation system of low yield point steel seismic link could strongly enhance the seismic performance of the tower structure where the tower and the overall bridge demands are significantly reduced. Low yield point steel seismic link effectively reduces the damage of main structural members under earthquake loading as seismic link yield level decreases due their exceptional behavior as well as its ability to undergo early plastic deformations achieving the concentration of inelastic deformation at tower horizontal beam.

• 산업경영시스템학회지
• /
• 제39권3호
• /
• pp.1-9
• /
• 2016

NSS (Navigation satellite system) provides the information for determining the position, velocity and time of users in real time using satellite-networking, and is classified into GNSS (Global NSS) and RNSS (Regional NSS). Although GNSS services for global users, the exactitude of provided information is dissatisfied with the degree required in modern systems such as unmanned system, autonomous navigation system for aircraft, ship and others, air-traffic control system. Especially, due to concern about the monopoly status of the countries operating it, some other countries have already considered establishing RNSS. The RNSS services for users within a specific area, however, it not only gives more precise information than those from GNSS, but also can be operated independently from the NSS of other countries. Thus, for Korean RNSS, this paper suggests the methodology to design the satellite constellation considering the regional features of Korean Peninsula. It intends to determine the orbits and the arrangement of navigation satellites for minimizing PDOP (Position dilution of precision). PGA (Parallel Genetic Algorithm) geared to solve this nonlinear optimization problem is proposed and STK (System tool kit) software is used for simulating their space flight. The PGA is composed of several GAs and iterates the process that they search the solution for a problem during the pre-specified generations, and then mutually exchange the superior solutions investigated by each GA. Numerical experiments were performed with increasing from four to seven satellites for Korean RNSS. When the RNSS was established by seven satellites, the time ratio that PDOP was measured to less than 5 (i.e. better than 'Good' level on the meaning of the PDOP value) was found to 94.3% and PDOP was always kept at 10 or less (i.e. better than 'Moderate' level).

• 한국지능시스템학회논문지
• /
• 제14권4호
• /
• pp.402-410
• /
• 2004

본 논문에서는 복잡하고 비선형적인 시스템을 위하여 최적 면역 알고리즘의 개선된 클론선택에 기반을 둔 최적FNN 설계방법을 제안한다. FNN은 퍼지추론의 간략 추론과 학습방법으로는 오류역전파 알고리즘을 하였고 멤버쉽함수의 파라미터, 학습률 및 모멘텀 계수들을 선정하기 위하여 개선된 클론 선택을 사용하는 방법을 도입하였다. 제안한 알고리즘은 생체의 면역반응에 기초를 둔 면역알고리즘의 클론선택을 기본으로 분화율을 조절하여 성능을 개선하였다. 그 과정을 통하여 다양한 항체들을 생성하고 목적함수나 제한조건과 같은 항원들에 대하여 가장 높은 친화도를 가지는 항체를 최적 항체로 선택하였다. 제안된 알고리즘의 성능을 평가하기 위하여 가스로공정과 교통경로선택 공정을 사용한다.

• Structural Engineering and Mechanics
• /
• 제40권1호
• /
• pp.85-103
• /
• 2011

Integration of finite element analysis (FEA) software into various software platforms is commonly used in coupling systems such as systems involving structural control, fluid-structure, wind-structure, soil-structure interactions and substructure method in which FEA is used for simulating the structural responses. Integrating an FEA program into various other software platforms in an efficient and simple way is crucial for the development and performance of the entire coupling system. The lack of simplicity of the existing integration methods makes this integration difficult and therefore entails the motivation of this study. In this paper, a novel practical technique, namely CS technique, is presented for integrating a general FEA software framework OpenSees into other software platforms, e.g., Matlab-$Simulink^{(R)}$ and a soil-structure interaction (SSI) system. The advantage of this integration technique is that it is efficient and relatively easy to implement. Instead of OpenSees, a cheap client handling TCL is integrated into the other software. The integration is achieved by extending the concept of internet based client-server concept, taking advantage of the parameterization framework of OpenSees, and using a command-driven scripting language called tool command language (TCL) on which the OpenSees' interface is based. There is no need for any programming inside OpenSees. The presented CS technique proves as an excellent solution for the coupling problems mentioned above (for both linear and nonlinear problems). Application examples are provided to validate the integration method and illustrate the various uses of the method in the civil engineering.

• 전자공학회논문지SC
• /
• 제38권1호
• /
• pp.1-13
• /
• 2001

인버터를 적용한 유압식 엘리베이터 시스템은 펌프의 마찰과 실린더 패킹 및 탑승카와 레일의 마찰특성으로 인하여 PID 제어기로는 제어가 되지 않는 데드존이 생기게 된다. 본 논문에서는, 이러한 문제를 해결하기 위하여 먼저, 퍼지제어기와 PID제어기를 혼용하는 하이브리드제어기를 이용하는 방법을 시도한다. 그러나, 인버터를 적용한 유압식 엘리베이터는 비선형성이 강한 시스템이므로 두 제어기의 출력이 절환 되어야 하는 경계층이 다수인 경우가 대부분인데 반하여, 기존의 하이브리드제어기는 두 제어기의 출력이 절환되는 특정구간의 경계층을 제외하고는 전체 운전구간에서는 어느 한 제어기에만 영향을 받게 되므로 제어성능에 문제를 가지게 된다. 이에 따라 본 논문에서는, 출력혼합기의 출력비를 퍼지로직에 의하여 변경시키는 새로운 퍼지하이브리드제어기를 제안하여 기존의 하이브리드제어기의 문제점을 해결한다. 제안된 퍼지하이브리드제어기는 시스템의 상태에 따라서 두 제어기의 출력비를 달리하여 사용하는 방법으로 향상된 제어성능을 달성한다. 시뮬레이션 결과를 통하여, 제안된 퍼지하이브리드제어기가 극저속 속도영역에서 뿐 아니라 정상상태를 포함한 전 운전 영역에서의 제어 성능이 우수함을 보였다.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 2004년도 ICCAS
• /
• pp.1033-1036
• /
• 2004

The study on speech recognition and understanding has been done for many years. In this paper, we propose a new type of recurrent neural network architecture for speech recognition, in which each output unit is connected to itself and is also fully connected to other output units and all hidden units [1]. Besides that, we also proposed the new architecture and the learning algorithm of recurrent neural network such as Backpropagation Through Time (BPTT, which well-suited. The aim of the study was to observe the difference of Arabic's alphabet like "alif" until "ya". The purpose of this research is to upgrade the people's knowledge and understanding on Arabic's alphabet or word by using Recurrent Neural Network (RNN) and Backpropagation Through Time (BPTT) learning algorithm. 4 speakers (a mixture of male and female) are trained in quiet environment. Neural network is well-known as a technique that has the ability to classified nonlinear problem. Today, lots of researches have been done in applying Neural Network towards the solution of speech recognition [2] such as Arabic. The Arabic language offers a number of challenges for speech recognition [3]. Even through positive results have been obtained from the continuous study, research on minimizing the error rate is still gaining lots attention. This research utilizes Recurrent Neural Network, one of Neural Network technique to observe the difference of alphabet "alif" until "ya".

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 2003년도 ICCAS
• /
• pp.2641-2647
• /
• 2003

Conventional DLLs estimate the delay times of satellite signals individually and feed back these measurements to the VCO independently. But VDLL estimates delay times and user position directly and then estimate the feedback term for VCO using the estimated position changes. In this process, input measurements are treated as vectors and these vectors are used for navigation. First advantage of VDLL is that noise is reduced in all of the tracking channels making them less likely to enter the nonlinear region and fall below threshold. Second is that VDLL can operate successfully when the conventional independent parallel DLL approach fails completely. It means that VDLL receiver can get enough total signal power to track successfully to obtain accurate position estimates under the same conditions where the signal strength from each individual satellite is so low or week that none of the individual scalar DLL can remain in lock when operating independently. To operate VDLL successfully, it needs to know the initial user dynamics and position and prevents total system from the divergence. The suggested integration method is to use the inertial navigation system to provide initial dynamics for VDLL and to maintain total system stable. We designed the GPS/INS integrated navigation system. This new type of integrated system contained the vector pseudorange format generation block, VDLL signal processing block, position estimation block and the conversion block from position change to delay time feedback term aided by INS.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 2003년도 ICCAS
• /
• pp.2182-2186
• /
• 2003

This paper presents the differential type resonant accelerometer (DRXL) and its performance test results. The DRXL is the INS grade, surface micro-machined sensor. The proposed DRXL device produces a differential digital output upon an applied acceleration, and the principle is a gap-dependent electrical stiffness variation of the electrostatic resonator with torsion beam structures. Using this new operating concept, we designed, fabricated and tested the proposed device. The final device was fabricated by using the wafer level vacuum packaging process. To test the performance of the DRXL, a nonlinear self-oscillation loop is designed using describing function technique. The oscillation loop is implemented using discrete electronic elements. The performance test of the DRXL shows that the sensitivity of the accelerometer is 12 Hz/g and its long term bias stability is about $2mg(1{\sigma})$. The turn on repeatability, bandwidth, and dynamic range are 4.38 mg, 100 Hz, and ${\pm}\;70g$, respectively.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 2003년도 ICCAS
• /
• pp.2331-2334
• /
• 2003

The basic difference between the EKF(Extended Kalman Filter) and UKF(Unscented Kalman Filter) stems from the manner in which Gaussian random variables(GRV) are represented for propagating through system dynamics. In the EKF, the state distribution is approximated by a GRV, which is then propagated analytically through the first-order linearization of the nonlinear system. This can possibly introduce large errors in the true posterior mean and covariance of the transformed GRV, which may lead to sub-optimal performance and sometimes divergence of the filter. However, the UKF addresses this problem by using a deterministic sampling approach. The state distribution is also approximated by a GRV, but is now represented using a minimal set of carefully chosen sample points. These sample points completely capture the true mean and covariance of the GRV, and UKF captures the posterior mean and covariance accurately up to the 2nd order(Taylor series expansion) for any nonlinearity. This paper utilizes the UKF to determine spacecraft orbit when only magnetometer is available. Several catastrophic failures of spacecraft in orbit have been attributed to failures of the spacecraft mission. Recently studies on contingency-major sensor failure cases- have been performed. For mission success, contingency design or plan should be implemented in case of a major sensor failure. Therefore the algorithm presented in this paper can be used for a spacecraft without GPS or contingency design in case of GPS failure.