• Journal of the Institute of Electronics and Information Engineers
• /
• v.51 no.11
• /
• pp.94-100
• /
• 2014

In this paper, a design of low power Current-Reuse Voltage Controlled Oscillator (VCO) which has wide tuning range about 1.95 GHz ~ 3.15 GHz is presented. Class-C type is applied to improve phase noise and 2-Step Auto Amplitude Calibration (AAC) is used for minimizing the imbalance of differential VCO output voltage which is main issue of Current-Reuse VCO. The mismatch of differential VCO output voltage is presented about 1.5mV ~ 4.5mV. This mismatch is within 0.6 % compared with VCO output voltage. Proposed Current-Reuse VCO is designed using CMOS $0.13{\mu}m$ process. Supply voltage is 1.2 V and current consumption is 2.6 mA at center frequency. The phase noise is -116.267 dBc/Hz at 2.3GHz VCO frequency at 1MHz offset. The layout size is $720{\times}580{\mu}m^2$.

• The Transactions of the Korean Institute of Power Electronics
• /
• v.16 no.5
• /
• pp.494-502
• /
• 2011

Bimodal tram is designed to run on a dedicated path in automatic mode using a magnetic track system in order to realize a combination of the accessibility of a bus and the constant regularity of a railroad. This paper presents design and test results of the series hybrid propulsion system of the bimodal tram on both test track and public road, which uses CNG (Compressed Natural Gas) engine and Lithium polymer battery pack. This paper describes the real-time data measuring equipment for the series hybrid propulsion system of the bimodal tram. Using this measurement equipment, the performance of the prototype vehicle's driving on test track and public road was verified and the fuel consumption and the efficiency of CNG engine have been investigated.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2023.07a
• /
• pp.541-544
• /
• 2023

LLM(Largescale Language Model)의 성능 향상을 위한 비용 효율적인 방법으로 ChatGPT, GPT-4와 같은 초거대 모델의 output에 대해 SLM(Small Language Model)을 finetune하는 방법이 주목받고 있다. 그러나, 이러한 접근법은 주로 범용적인 지시사항 모델을 위한 학습 방법으로 사용되며, 제한된 특정 도메인에서는 추가적인 성능 개선의 여지가 있다. 본 연구는 특정 도메인(Writing Assistant)에서의 성능 향상을 위한 새로운 방법인 Self-Guided Approach를 제안한다. Self-Guided Approach는 (1) LLM을 활용해 시드 데이터에 대해 도메인 특화된 metric(유용성, 관련성, 정확성, 세부사항의 수준별) 점수를 매기고, (2) 점수가 매겨진 데이터와 점수가 매겨지지 않은 데이터를 모두 활용하여 supervised 방식으로 SLM을 미세 조정한다. Vicuna에서 제안된 평가 방법인, GPT-4를 활용한 자동평가 프레임워크를 사용하여 Self-Guided Approach로 학습된 SLM의 성능을 평가하였다. 평가 결과 Self-Guided Approach가 Self-instruct, alpaca와 같이, 생성된 instruction 데이터에 튜닝하는 기존의 훈련 방법에 비해 성능이 향상됨을 확인했다. 다양한 스케일의 한국어 오픈 소스 LLM(Polyglot1.3B, PolyGlot3.8B, PolyGlot5.8B)에 대해서 Self-Guided Approach를 활용한 성능 개선을 확인했다. 평가는 GPT-4를 활용한 자동 평가를 진행했으며, Korean Novel Generation 도메인의 경우, 테스트 셋에서 4.547점에서 6.286점의 성능 향상이 발생했으며, Korean scenario Genration 도메인의 경우, 테스트 셋에서 4.038점에서 5.795 점의 성능 향상이 발생했으며, 다른 유사 도메인들에서도 비슷한 점수 향상을 확인했다. Self-Guided Approach의 활용을 통해 특정 도메인(Writing Assistant)에서의 SLM의 성능 개선 가능성을 확인했으며 이는 LLM에 비용부담을 크게 줄이면서도 제한된 도메인에서 성능을 유지하며, LLM을 활용한 응용 서비스에 있어 실질적인 도움을 제공할 수 있을 것으로 기대된다.

• Journal of Advanced Marine Engineering and Technology
• /
• v.38 no.1
• /
• pp.39-47
• /
• 2014

In this study, 2 degree of freedom PID controller is added to the conventional feed-forward controller for the purpose of improving its limitations such as set-point of tracking performance and disturbance suppression performance in the conventional PID controller. And the controller parameters optimization as a Real Coded Genetic Algorithm (RCGA) is used. Simulation and experiments verify the performance of the controller.

• Journal of Biosystems Engineering
• /
• v.21 no.3
• /
• pp.283-292
• /
• 1996

Automatic guidance of farm tractors would improve productivity by reducing operator fatigue and increasing machine performance. To control tractors within $\pm$5cm of the desired path, fuzzy and adaptive steering controllers were developed to evaluate their characteristics and performance. Two input variables were position and yaw errors, and a steering command was fed to tractor model as controller output. Trapezoidal membership functions were used in the fuzzy controller, and a minimum-variance adaptive controller was implemented into the 2-DOF discrete-time input-output model. For unit-step and composite paths, a dynamic tractor simulator was used to test the controllers developed. The results showed that both controllers could control the tractor within $\pm$5cm error from the defined path and the position error of tractor by fuzzy controller was the bigger of the two. Through simulations, the output of self-tuning adaptive controller was relatively smooth, but the fuzzy controller was very sensitive by the change of gain and the shape of membership functions. Contrarily, modeling procedure of the fuzzy controller was simple, but the adaptive controller had very complex procedure of design and showed that control performance was affected greatly by the order of its model.

• Proceedings of the KIEE Conference
• /
• 2008.10b
• /
• pp.403-404
• /
• 2008

자원의 고갈과 각종 환경규제 및 정부의 전력거래 방침 등으로 인해 점점 발전소에서는 전력 계통에 대한 신뢰도 및 전기품질을 유지하기가 어려워지고있다. 발전소의 부하대별 주요 운전 설계값은 효율과 바로 직결되는 사항으로 각 부하별로 온도와 압력 등 운전설계값을 최적의 상태로 유지하는 것은 발전소 수명과 발전효율, 전력거래 등에 있어 중요한 요소들이다. 전력시장에 진입하는 발전소는 전력계통의 갑작스런 불안성 상황이나 전력거래소 요청시 경사변동폭, 출력변통율, 무효전력 출력, 자동발전제어, 주파수조절량 확보 등을 수행할 수 있어야 한다. 본 논문에서는 고급공정제어기를 이용하여 운전설계계값을 효과적으로 제어 하면서 기존의 제어로직보다 전력계통상에서 요구되는 발전기 출력을 최대한 신속히 제어하는 과정을 기술하였다. 우선 보일러 마스터와 터빈 마스터, 급수 마스터로 구성된 제어로직을 설계한 뒤 이들 마스터에 대한 발전기 출력, 주증기 압력, 기수분리기 출구온도 각각의 영향을 모델링 기법을 이용하여 적합한 모델을 구했다. 각각의 모델을 고급공정제어기에 적용하고 발전기 출력제어에 대한 기존의 발전소 응답보다 좀 더 효율적이고 실제적용이 가능한 결과를 얻을 수 있도록 튜닝을 시행했으며 그 과정과 결과를 기술했다.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.13 no.6
• /
• pp.2421-2427
• /
• 2012

This paper is discussed the design on non-linear adaptive attitude controller for quadrotor UAV. Quadrotor UAV featured to have four rotor, required the special controller to compensate for the model parameter uncertainties as the unstable nonlinear system. In this research, we designed the adaptive controller to compensate for the payload changes even though it is changed with industrial applications. Especially, based on the mathematical model of UAV, non-linear adaptive controller is suggested and the stability is verified using the Lyapunov function and finally proved its performance and effectiveness of update laws with various payload by simulation.

• Proceedings of the KOSOMBE Conference
• /
• v.1998 no.11
• /
• pp.141-142
• /
• 1998

The urine analysis system implemented to measuring the primary color reaction of urinalysis strip. Fuzzy classifier based on fuzzy theory implemented so as to classify of 9 items in the urinalysis strip and proposed the automatic turning algorithm of mambership function in the fuzzy classifier to progress the reproduction of classify. To evaluation of clinical capability, the fuzzy classifier and automatic turning algorithm apples to standard strip and standard reagent.

• Journal of Institute of Control, Robotics and Systems
• /
• v.7 no.12
• /
• pp.973-979
• /
• 2001

Sliding mode control guarantees robustness in the presence of modeling uncertainties and external disturbances. However, this can be obtained at the cost of high control activity that may lead to chattering As one way to alleviate this problem a boundary layer around sliding surface is typically used. In this case the selection of controller gain, control ban width and boundary layer thickness is a crucial problem for the trade-off between tracking error and chattering. The parameter tuning is usually done by trail-and-error in practice causing significant effort and time. An auto tuning method based on fuzzy rules is proposed in the paper in this method tracking error and chattering are monitored by performance indices and the controller tunes the design parameters intelligently in order to compromise both indices. To demonstrate the efficiency of the propose method a mass-spring translation system and a roboic control system are simulated and tested It is shown that the proposed algorithm is effective to facilitae the parameter tuning for sliding mode controllers.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2021.10a
• /
• pp.394-396
• /
• 2021

Additional pathological tests using imaging equipment are essential before diagnosing cancer cells. Recently, in order to reduce the need for time and human resources in these fields, research related to the establishment of a system capable of automatic classification of cancer cells using artificial intelligence is being actively conducted. However, in both previous studies, there were relatively limited deep learning algorithms and cell types, and limitations existed with low accuracy at the same time. In this study, a method of performing 4class Classification on four types of cancer cells through the Convolution Neral Network, a type of in-depth learning. EfficientNet, ResNet, and Inception were used, and finally Resnet was used to obtain an accuracy of 96.11 on average for k-fold.