• International Journal of Fuzzy Logic and Intelligent Systems
• /
• v.9 no.4
• /
• pp.294-300
• /
• 2009

The neural network with radial basis function is introduced for position tracking control of AC servo drive with the existence of system uncertainties. An adaptive robust term is applied to overcome the external disturbances. The proposed controller is implemented on a high performance digital signal processing DSP TMS320C6713-300. The stability and the convergence of the system are proved by Lyapunov theory. The validity and robustness of the controller are verified through simulation and experimental results

• Journal of information and communication convergence engineering
• /
• v.2 no.3
• /
• pp.172-176
• /
• 2004

In special effects industry there is a high demand to convincingly mimic the appearance and behavior of natural phenomena such as smoke, waterfall, rain, and fire. Particle systems are methods adequate for modeling fuzzy objects of natural phenomena. This paper presents particle system graphics library for generating special effects in video games and virtual reality applications. The library is a set of functions that allow c++ programs to simulate the dynamics of particles for special effects in interactive and non-interactive graphics applications, not for scientific simulation.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• v.16 no.4
• /
• pp.270-275
• /
• 2016

We present an approach for acceleration actor-critic algorithm for reinforcement learning with continuous action space. Actor-critic algorithm has already proved its robustness to the infinitely large action spaces in various high dimensional environments. Despite that success, the main problem of the actor-critic algorithm remains the same-speed of convergence to the optimal policy. In high dimensional state and action space, a searching for the correct action in each state takes enormously long time. Therefore, in this paper we suggest a search accelerating function that allows to leverage speed of algorithm convergence and reach optimal policy faster. In our method, we assume that actions may have their own distribution of preference, that independent on the state. Since in the beginning of learning agent act randomly in the environment, it would be more efficient if actions were taken according to the some heuristic function. We demonstrate that heuristically-accelerated actor-critic algorithm learns optimal policy faster, using Educational Process Mining dataset with records of students' course learning process and their grades.

• Journal of the Korea Convergence Society
• /
• v.4 no.2
• /
• pp.43-50
• /
• 2013

Sliding mode control(SMC) are carried out in this literature. And to make the controllers perform better, fuzzy logic was chosen,it makes PID controller auto-tuning parameters and reduced the chattering problem of sliding mode control. Since SMC take error and derivative of error as inputs, after comparison some results are obtained.PID controller response faster yet sliding mode control is much steadier. However certain problems cannot be ignored that the chattering phenomenal cannot be reduced entirely and this motion may hurt the machine; this project only considered a simple system, there is no guarantee PID can work as well as in this case for a much more complex system. MATLAB simulink was the main approach to obtain the performance of the two controllers: to observe the control output of the two controllers, electric circuit and special controllers are designed and tested in MATLAB.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• v.1 no.1
• /
• pp.125-130
• /
• 2001

In this paper, we propose a new type of evolution strategies combined with reinforcement learning. We use the variances of fitness occurred by mutation to make the reinforcement signals which estimate and control the step length of mutation. With this proposed method, the convergence rate is improved. Also, we use cauchy distributed mutation to increase global convergence faculty. Cauchy distributed mutation is more likely to escape from a local minimum or move away from a plateau. After an outline of the history of evolution strategies, it is explained how evolution strategies can be combined with the reinforcement learning, named reinforcement evolution strategies. The performance of proposed method will be estimated by comparison with conventional evolution strategies on several test problems.

• Journal of KIISE:Software and Applications
• /
• v.26 no.1
• /
• pp.130-130
• /
• 1999

In this paper, we propose a novel design method for improving performance of existing FCM-type clustering algorithms. First, we define the performance measure which focuses on bothcompactness and separation of clusters. Next, we optimize this measure using evolution program.Especially the proposed method has following merits: ① using evolution program, it solves suchproblems as initialization, number of clusters, and convergence to local optimum ② it reduces searchspace and improves convergence speed of algorithm since it represents chromosome with possiblepotential centers which are selected possible candidates of centers by density measure ③ it improvesperformance of clustering algorithm with the performance index which embedded both compactnessand separation Properties ④ it is robust to noise data since it minimizes its effect on center search.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• v.5 no.4
• /
• pp.297-301
• /
• 2005

In this paper, we propose a method to estimate camera motion parameter based on invariant point features. Typically, feature information of image has drawbacks, it is variable to camera viewpoint, and therefore information quantity increases after time. The LM(Levenberg-Marquardt) method using nonlinear minimum square evaluation for camera extrinsic parameter estimation also has a weak point, which has different iteration number for approaching the minimal point according to the initial values and convergence time increases if the process run into a local minimum. In order to complement these shortfalls, we, first propose constructing feature models using invariant vector of geometry. Secondly, we propose a two-stage calculation method to improve accuracy and convergence by using homography and LM method. In the experiment, we compare and analyze the proposed method with existing method to demonstrate the superiority of the proposed algorithms.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• v.5 no.4
• /
• pp.310-315
• /
• 2005

This paper proposes a fast optimization method by combining queen-bee evolution and derivative evaluation in genetic algorithms. These two operations make it possible for genetic algorithms to focus on highly fitted individuals and rapidly evolved individuals, respectively. Even though the two operations can also increase the probability that genetic algorithms fall into premature convergence phenomenon, that can be controlled by strong mutation rates. That is, the two operations and the strong mutation strengthen exploitation and exploration of the genetic algorithms, respectively. As a result, the genetic algorithm employing queen-bee evolution and derivative evaluation finds optimum solutions more quickly than those employing one of them. This was proved by experiments with one pattern matching problem and two function optimization problems.

• Proceedings of the Korea Institute of Convergence Signal Processing
• /
• 2003.06a
• /
• pp.149-152
• /
• 2003

플랜트 모델이나 경험에 근거하여 설계된 퍼지제어기를 실제 플랜트에 적용할 경우, 모델링 오차와 플랜트에 대한 관련지식의 부족으로 만족할 만한 제어 결과를 나타내지 못할 경우가 있다. 이 경우 제어성능을 향상시키기 위해 제어기의 제어인자를 다시 조정하여야 하고, 이 조정과정은 시행착오 방법으로 수행되기 때문에 많은 시간과 비용을 필요로 한다. 본 논문에서는 정규화 된 오차와 오차 변화량를 사용하여 플랜트 응답에 따라 입력과 출력의 적절한 스케일 계수를 조정하는 퍼지제어기를 제안한다. 정규화 된 오차를 출력 소속함수의 중심과 폭에 곱해 출력 범위를 재조정하고, 플랜트 응답에 의해 입력의 스케일 계수를 결정한다. 이를 확인하기 위해 2차 플랜트에 적용하여 모의 실험을 수행하였다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2018.10a
• /
• pp.457-459
• /
• 2018

지난 해 롤스로이스의 무인 선박 개발 프로젝트(AAWA) 본격화 등 선박과 관련된 무인 자동화 기술 개발이 활발해지면서 국내의 R&D 사업이 많이 증가하였다. 특히 국내에서는 퍼지 모델을 이용한 RVC 지능 시스템 등 퍼지 이론을 사용한 기술들이 최근까지도 발표되고 있다. 퍼지 모델을 결정하기 위해선 해당 시스템에 대한 전문 지실뿐만이 아니라 다양한 환경에서의 반복적인 실험과 수정을 필요로 하기 때문에 시뮬레이터를 만들어 실험하게 되는데 다양한 환경에서의 반복적인 실험과 수정을 필요로 하기 때문에 시뮬레이터를 만들어 실험하게 되는데 대부분의 연구에서 시뮬레이터가 제작되는 비용과 시간에도 불구하고 해당 퍼지 모델을 위해서만 쓰이게 된다. 따라서 본 논문에서는 다양한 환경에서 퍼지 모델을 반복 실험할 수 있도록 시뮬레이터를 개발하였으며 기존의 퍼지 모델 일부를 본 시뮬레이터에 적용하여 같은 실험을 할 수 있음을 보이고 이를 통해 퍼지 모델을 만드는데 드는 시간과 비용을 줄일 수 있음을 보였다.