• Journal of Korea Society of Industrial Information Systems
• /
• v.12 no.4
• /
• pp.126-130
• /
• 2007

This paper presents a method to design a repetitive controller that is specified in the specified trajectory for the repetitive works. With the single-model design approach, the controller is derived by minimizing a frequency-domain based cost function that produces monotonic convergence of the tracking error as a function of repetition number. Numerical illustrations show how the proposed single-model design method produces a repetitive controller in a single nominal model of the system.

• Journal of Korea Society of Industrial Information Systems
• /
• v.13 no.2
• /
• pp.1-7
• /
• 2008

This paper presents a method to design a repetitive controller that is robust to variations in the system parameters. The uncertain parameters are specified probabilistically by their probability distribution functions. Instead of working with the distribution functions directly, the repetitive controller is designed from a set of models that are generated from the specified probability functions. With this multiple-model design approach, any number of uncertain parameters that follow any type of distribution functions can be treated. furthermore, the controller is derived by minimizing a frequency-domain based cost function that produces monotonic convergence of the tracking error as a function of repetition number. Numerical illustrations show how the proposed multiple-model design method produces a repetitive controller that is significantly more robust than an optimal repetitive controller designed from a single nominal model of the system.

• Journal of Korea Society of Industrial Information Systems
• /
• v.16 no.4
• /
• pp.99-105
• /
• 2011

This paper presents a method to design a centralized repetitive controller that is robust to variations in the multiple system parameters. The uncertain parameters are specified probabilistically by their probability distribution functions. Instead of working with the distribution functions directly, the centralized repetitive controller is designed from a set of models that are generated from the specified probability functions. With this multiple-model design approach, any number of uncertain parameters that follow any type of distribution functions can be treated. Furthermore, the controller is derived by minimizing a frequency-domain based cost function that produces monotonic convergence of the tracking error as a function of repetition number. Numerical illustrations show how the proposed multiple-model design method produces a repetitive controller that is significantly more robust than an optimal repetitive controller designed from a single nominal model of the multiple system.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.13 no.5
• /
• pp.422-428
• /
• 1988

An iterative learning control system is a control system which makes system outputs follow desired outputs by iterating its trials over a finite time interval. In a discrete time system, we proposed one method in which present control inputs can be obtained by a linear combination of the input sequence and time-shifted error sequence at previous trial. In contrast with a continous time learning control system which needs differential opreration of an error signal, the time shift operation of the error sequence is simpler in a computer control system and its effectiveness is shown by a simulation.

• Proceedings of the KAIS Fall Conference
• /
• 2000.10a
• /
• pp.218-222
• /
• 2000

본 논문에서는 피드백 사용형 2차 반복 학습제어 방법이 수렴 성능의 향상과 외란에 대한 강인성 향상에 덧붙여 학습제어의 피드백 항을 이용함으로써 초기 조건 오차가 있음에도 불구하고 이를 극복할 뿐만 아니라 기존의 알고리즘보다 더 빠른 수렴 능력이 있음을 확인한다. 또한 불안정한 결과를 낳는 높은 학습 제어 게인의 경우에도 피드백 항을 추가한 본 학습제어 방법에 의해 안정화됨으로써, 빠른 응답 특성과 강인성 향상을 가져올 수 있음을 보인다. 그리고 본 알고리즘을 선형 시변 연속 시스템이 적용된 모델 시스템을 설계하여 이를 통한 시뮬레이션 결과로서 초기 조건 오차의 극복 능력이 뛰어남을 확인하고 시스템의 안정화와 강인성 향상에 기여함을 확인한다.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.13 no.2
• /
• pp.175-179
• /
• 2003

Convergence condition determines performance of iterative learning control (ILC), for example, convergence speed, remaining error, etc. Hence, the performance can be elevated and a feasible set of learning controllers grows if a less conservative condition is obtained. In the frequency domain, the $H_{\infty}$ norm of the transfer function between consecutive errors has been currently used to test convergence of a learning system. However, even if the convergence condition based on the $H_{\infty}$ norm has a clear property about monotonic convergence, it has a few drawbacks, especially in MIMO plants. In this paper, the relation between the condition and the monotonicity of convergence is clarified and a modified convergence condition is found out using a frequency domain Lyapunov equation, which supersedes the conventional one in the frequency domain.

• Annual Conference on Human and Language Technology
• /
• 1995.10a
• /
• pp.30-34
• /
• 1995

신경망을 이용하는 연속 음성 인식에서 학습이라 함은 인위적으로 분할된 음성 데이타를 토대로 진행되는 것이 지배적이었다. 그러나 분할된 음성데이타를 마련하기 위해서는 많은 시간과 노력, 숙련 등을 요구할 뿐만아니라 그 자체가 인식도메인의 변화나 확장을 어렵게 하는 하나의 요인 되기도 한다. 그래서 분할된 음성데이타의 사용을 가급적 피하고 그러면서도 성능을 떨어뜨리지 않는 신경망 학습법들이 나타나고 있다. 본 논문에서는 학습된 인식기를 이용하여 자동으로 한국어 음성데이타를 분할한 후 그 분할된 데이타를 이용하여 다시 인식기를 재학습시켜나가는 반복 과정을 소개하고자 한다. 여기에는 TDNN이 인식기로 사용되며 인식단위는 음소이다. 학습은 cross-validation 기법을 이용하여 제어된다.

• Journal of KIISE:Software and Applications
• /
• v.30 no.11
• /
• pp.1054-1061
• /
• 2003

Q-Learning, a representative algorithm of reinforcement learning, experiences repeatedly until estimation values about all state-action pairs of state space converge and achieve optimal policies. When the state space is high dimensional or continuous, complex reinforcement learning tasks involve very large state space and suffer from storing all individual state values in a single table. We introduce Q-Map that is new function approximation method to get classified policies. As an agent learns on-line, Q-Map groups states of similar situations and adapts to new experiences repeatedly. State-action pairs necessary for fine control are treated in the form of rule. As a result of experiment in maze environment and mountain car problem, we can achieve classified knowledge and extract easily rules from Q-Map

• Korean Chemical Engineering Research
• /
• v.44 no.5
• /
• pp.425-434
• /
• 2006

The operability and productivity of continuous processes, especially in petrochemical industries have made remarkable improvement during the past twenty years through advanced process control (APC) typified by model-based predictive control. On the other hand, APC have not been actively practiced in industrial batch processes typified by batch polymerization reactors. Perhaps the main cause for this has been the lack of reliable batch process APC techniques that can overcome the unique problems in industrial batch processes. Recently, some noteworthy progress is being made in this area. New high-performance batch process control techniques that can accommodate and also overcome the unique problems of industrial batch processes have been proposed on the basis of iterative learning control (ILC). In this review paper, recent advancement in the batch process APC techniques are presented, with a particular focus on the variations of the so called Q-ILC method, with the hope that they are widely practiced in different industrial batch processes and enhance their operations.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• v.9 no.2
• /
• pp.366-369
• /
• 2005

Background subtraction is mainly used for the real-time extraction and tracking of moving objects from image sequences. In the outdoor environment, there are many changeable factor such as gradually changing illumination, swaying trees and suddenly moving objects, which are to be considered for the adaptive processing. Normally, GMM(Gaussian Mixture Model) is used to subtract the background adaptively considering the various changes in the scenes, and the adaptive GMMs improving the real-time performance were worked. This paper, for on-line background subtraction, applied the improved adaptive GMM, which uses the small constant for learning rate ${\alpha}$ and is not able to speedily adapt the suddenly movement of objects, So, this paper proposed and evaluated the dynamic control method of ${\alpha}$ using the adaptive selection of the number of component distributions and the global variances of pixel values.