• 제어로봇시스템학회:학술대회논문집
• /
• 1994.10a
• /
• pp.243-246
• /
• 1994

In this paper, a self-learning fuzzy controller is designed with a fuzzy approximation of an inverse model. The aim of an identification is to find an input command which is control of a system output. It is intuitional and easy to use a classical adaptive inverse modeling method for the identification, but it is difficult and complex to implement it. This problem can be solved with a fuzzy approximation of an inverse modeling. The fuzzy logic effectively represents the complex phenomena of the real world. Also fuzzy system could be represented by the neural network that is useful for a learning structure. The rule of a fuzzy inverse model is modified by the gradient descent method. The goal is to be obtained that makes the design of fuzzy controller less complex, and then this self-learning fuzz controller can be used for nonlinear dynamic system. We have applied this scheme to a nonlinear Ball and Beam system.

• 제어로봇시스템학회:학술대회논문집
• /
• 1996.10b
• /
• pp.1300-1303
• /
• 1996

In this paper a typical problem is examined that a light, general aviation airplane, such as Cessna or Navion, in gliding turn flight shows helical-dive phenomenon when pilots try to stop the descent by using elevator only. It is known from pilot's experience that in a certain flight trim it is impossible to recover from helical-dive by using elevator only. From this study it is shown that helical-dive phenomenon is involved with longitudinal/lateral dynamics coupling to airplane's aerodynamics. Also this phenomenon consists of three parts of flight dynamics; first of all, fast longitudinal motion occurs, then is followed by a little slow lateral motion, and finally logitudinal/lateral coupled motion is fully developed.

• Journal of Institute of Control, Robotics and Systems
• /
• v.5 no.8
• /
• pp.914-922
• /
• 1999

A study on the improvement of tracking performance of a 3 DOF planar parallel manipulator is performed. A class of adaptive tracking control sheme is designed using self tuning adaptive fuzzy logic control theory. This control sheme is composed of three classical PD controller and a multi learning type self tuning adaptive fuzzy logic controller set. PD controller is tuned roughly by manual setting a priori and fuzzy logic controller is tuned precisely by the gradient descent method for a global solution during run-time, so the proposed control scheme is tuned more rapidly and precisely than the single learning type self tuning adaptive fuzzy logic control sheme for a local solution. The control performance of the proposed algorithm is verified through experiments.

• Proceedings of the IEEK Conference
• /
• 1998.06a
• /
• pp.933-935
• /
• 1998

The HMM-Net is an architecture for a neural network that implements a hidden markov model (HMM). The architecture is developed for the purpose of combining the discriminant power of neural networks with the time-domain modeling capability of HMMs. Criteria used for learning HMM-Net classifiers are maximum likelihood(ML) and minimization of mean squared error(MMSE). In this paper we propose an efficient learning method of HMM_Net classifiers using a ML-MMSE hybrid criterion and report the results of an experimental study comparing the performance of HMM_Net classifiers trained by the gradient descent algorithm with the above criteria. Experimental results for the isolated numeric digits from /0/ to /9/ show that the performance of the proposed method is better than the others in the repects of learning and recognition rates.

• The Journal of the Acoustical Society of Korea
• /
• v.15 no.4E
• /
• pp.21-27
• /
• 1996

In this paper, we propose a discriminative training algorithm for the stochastic segment model (SSM) in continuous speech recognition. As the SSM is usually trained by maximum likelihood estimation (MLE), a discriminative training algorithm is required to improve the recognition performance. Since the SSM does not assume the conditional independence of observation sequence as is done in hidden Markov models (HMMs), the search space for decoding an unknown input utterance is increased considerably. To reduce the computational complexity and starch space amount in an iterative training algorithm for discriminative SSMs, a hybrid architecture of SSMs and HMMs is programming using HMMs. Given the segment boundaries, the parameters of the SSM are discriminatively trained by the minimum error classification criterion based on a generalized probabilistic descent (GPD) method. With the discriminative training of the SSM, the word error rate is reduced by 17% compared with the MLE-trained SSM in speaker-independent continuous speech recognition.

• Journal of the Korean Chemical Society
• /
• v.3 no.1
• /
• pp.3-8
• /
• 1954

Fowler obtained thermodynamic quantities assuming the theory which could be derived by representing the system with microcanonical ensemble, in order to introduce the temperature T of the system proper, he considered the combined systems which are composed of the system proper and another arbitrary system that is in thermal contact with the former, and represented the combined system by a microcanonical ensemble, here, he used the steepest descent method in his calculation. This Fowler's treatment is not only unsatisfactory at the point of theoretical view but also he could not make the formulation of free energy of Helmholtz's so that this formular was forced to be assumed. From the point of Quantum Statistical Mechanical view, the materially closed system which is in an equilibrium state with the temperature T is best represented by canonical ensemble. At the actual derivation of the distribution law and thermodynamic quantities, however, in order to avoid the difficulty of calculation Tolman proceeded his calculation either representing the system proper by the grand-canonical ensemble or adding a certain limitation.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• v.5 no.1
• /
• pp.21-28
• /
• 2005

This paper proposes a new method for accelerating the search speed of genetic algorithms by taking derivative evaluation and conditional random selection into account in their evolution process. Derivative evaluation makes genetic algorithms focus on the individuals whose fitness is rapidly increased. This accelerates the search speed of genetic algorithms by enhancing exploitation like steepest descent methods but also increases the possibility of a premature convergence that means most individuals after a few generations approach to local optima. On the other hand, derivative evaluation under a premature convergence helps genetic algorithms escape the local optima by enhancing exploration. If GAs fall into a premature convergence, random selection is used in order to help escaping local optimum, but its effects are not large. We experimented our method with one combinatorial problem and five complex function optimization problems. Experimental results showed that our method was superior to the simple genetic algorithm especially when the search space is large.

• 한국정보통신설비학회:학술대회논문집
• /
• 2003.08a
• /
• pp.70-72
• /
• 2003

본 논문에서는 설계민감도 해석법과 위상최적화 기법을 사용하여 산란체의 물질상수 분포를 알기위한 수치해석 알고리즘을 제안하였다. 설계민감도 해석법과 보조변수법을 사용하여 복소 유전율에대한 목적함수의 미분정보를 계산하였고 이 민감도 정보를 통해 물질정보를 최적화 하였다. 최적화 기법으로 최대경사법(Steepest descent Method)을 사용하였으며 이 제안한 해석 기법을 2차원 TMz 모델에 적용함으로써 검증하였다.

• Proceedings of the KIPE Conference
• /
• 1999.07a
• /
• pp.526-530
• /
• 1999

This paper presents a newly developed speed sensorless drive using Neural Network algorithm. Neural Network algorithm can be divided into three categories. In the first one, a Back Propagation-based NN algorithm is well-known to gradient descent method. In the second scheme, a Extended Kalman Filter-based NN algorithm has just the time varying learning rate. In the last scheme, a Recursive Least Square-based NN algorithm is faster and more stable than the classical back-propagation algorithm for training multilayer perceptrons. The number of iterations required to converge and the mean-squared error between the desired and actual outputs is compared with respect to each method. The theoretical analysis and experimental results are discussed.

• KIEE International Transaction on Electrical Machinery and Energy Conversion Systems
• /
• v.3B no.1
• /
• pp.23-31
• /
• 2003

A novel 3D shape optimization algorithm is presented for electromagnetic devices carry-ing eddy current. The algorithm integrates the 3D finite element performance analysis and the steepest descent method with design sensitivity and mesh relocation method. For the design sensitivity formula, the adjoint variable vector is defined in complex form based on the 3D finite element method for eddy current problems. A new 3D mesh relocation method is also proposed using the deformation theory of the elastic body under stress to renew the mesh as the shape changes. The design sensitivity f3r the sur-face nodal points is also systematically converted into that for the design variables for the parameterized optimization application. The proposed algorithm is applied to the optimum design of the tank shield model of the transformer and the effectiveness is proved.